Cozero, end to end.
Everything a new Head of Product needs to know about the company, the product, the platform, and the finance-grade layer Cozero is now building on top of it. Written from the code, the Notion, the Slack, the customer roster, and the numbers.
Finance-grade decarbonisation software for European enterprises.
Cozero measures, manages, and reduces corporate emissions with the same rigor a CFO applies to a P&L. Log operational data, calculate scope 1, 2, and 3 emissions against a factor database, close reporting periods, publish CSRD and CDP-compliant reports, plan reductions with climate actions and targets, and share product-level footprints as Digital Product Passports.
The product's most-repeated positioning claim, on the marketing site and in every board doc: "carbon belongs on the balance sheet." The next-repeated is "ROCI: Return on Carbon Investment." The two together are the CFO pitch. Everything else in the platform supports one of them.
The version of the pitch that turns into product is Cozero Control: a separate app in prototype, built for the CFO persona rather than the Head of Sustainability, sitting alongside Core and Carbon Intelligence. Ali Arnaout's Q2 board opening in May crystallised it: "energy = emissions = euros, decarbonization IS a pricing strategy." Fabian's ROCI 2.0 proposal (v0.15, 29.07.2026) formalises the metric behind the pitch as Carbon-Adjusted NPV and "carbon uplift" (CIRR - IRR). Phoenix Contact bought the first paying access on 23.07.
"In 2019, Helen and I sat down with a simple but stubborn conviction: the world's companies needed a serious tool to actually manage their decarbonization, not just report on it, not just tick boxes, but truly transform. Six years later, I look at what we've built together and it genuinely still takes my breath away. From a spreadsheet replacement to more and more of a finance-grade decarbonization platform, one that corporations will eventually use to run their sustainability transformation like a CFO runs a P&L."
A Berlin climate SaaS that measures its own emissions on its own product.
Incorporated 26 June 2020. Series A closed April 2024. Approximately 30 people today, materially smaller than the older 50-100 figure in external profiles. The company went through a deliberate reset in autumn 2025 that Helen and Anja refer to as "six months of rebuilding."
HQ and offsites
Berlin. Maschinenraum in Mitte. Berlin Tech Days May 11-13 at Numa Torstraße produced the Intensity Metrics prototype. Summer offsite July 1-3 at Gutshof Fredenwalde, a manor 1.5h north of Berlin.
Investors
Kvanted (Helsinki, industrial tech). ENV / EnBW New Ventures (EnBW's corporate arm). New Alpha Asset Management (Paris). Business angels. Board chaired by Henrik Larsen (ex-Maersk).
Financial snapshot
May 2026: MRR €174K, flat vs plan, ~16% below. Cash €2.97M, ~€460K above plan. EBITDA -€121K. New expansion turned positive at +2.2%. Churn managed in Q3.
Certification
ISO 27001 certified since 2024 (Proks Cert), Q3 2026 surveillance audit in prep. GHG Protocol certification in flight with GutCert: audit day 9 June 2026, non-conformance items closed by 26 June.
Awards
German Sustainability Award 2022/2023 winner. Helen Tacke: Business Insider Top 25 Zukunftsmacherinnen. Verdantix 2026 Green Quadrant: Challenger. Carbon Alliance founding participant.
Currently hiring
Principal Engineer (job ad re-posted 26 June, 19 new applications the following week). Account Executive (4 candidates active). SDR (2 screenings this week). Senior Software Engineer for the Insights + ACT squad (late stage).
Every customer's year follows the same shape.
A Cozero customer is a mid-to-large European corporate. Their sustainability, facility, or logistics team logs in through a Cognito-backed auth flow (SSO for enterprise) and lands on the home dashboard. What happens next follows a repeatable annual cycle.
The four steps below describe the current Core-app workflow every existing paying customer uses today. Cozero Control introduces a fifth motion, described in the Cozero Control section: a driver-based forecast loop the CFO or head of controlling drives.
Model the org
Set up the organisation, its business unit tree, its physical locations, and its revenue figures. Onboarding wizard walks through this once. Suppliers are added when Scope 3.1 requests start.
Ingest operational data
Manual entries, Ingestro bulk imports, custom data transformation pipelines, REST API, EcoTransIT SFTP, TravelPerk API, supplier engagement links. All roads write to log entries.
Calculate and read
The calculation engine picks factors, applies methodology, produces scope 1, 2, and 3 rows plus a location-based mirror where relevant. Customer reads via Overview (Dashboard, Digital Twin, Intensity Metrics) and Log (LEO, Data Quality, Recalculations).
Close, plan, publish
Close the reporting period to lock numbers. Plan reductions in Act (Strategy Builder, Targets, Action Simulator). Publish computed reports (CSRD E1-1 to E1-9, CDP, GHG E1-6) as XLSX, share pages, sustainability reports, and Digital Product Passports.
Four modules the customer navigates: Overview, Log, Act, Share.
The sidebar in the Core app has four visible parent nodes plus Organization Settings, Account and Billing, and personal Settings. Every product surface today lives inside one of these. 165 Prisma models sit behind the UI. Cozero Control is a separate app; the Core modules below are what every existing paying customer uses.
Read the footprint
- Digital Twin: organisation as a tree of BUs and locations with emissions overlaid
- Dashboard: filterable scope 1/2/3 totals and hotspots
- Intensity Metrics: emissions per unit of business activity (tkm, m2, revenue)
Ingest and clean
- Data Ingestion: Ingestro-powered bulk import (over 80% of platform log entries)
- Corporate Footprint: LEO table, every log entry filterable
- PCF: product-level lifecycle model
- Factors: internal factor database + supplier factor requests
- Data Transformations: CDT pipelines for messy customer exports
- Closed Periods: lock a reporting year
- Data Quality: DQ rating per log entry
- Business Activities: AI-powered activity mapping
Plan reductions
- Strategy Builder: decarbonisation canvas of climate actions
- Targets: absolute and percent reductions with milestones
- Hotspot Analysis: granular emissions view with pinned analytics
- Action Simulator: Impact Explorer: simulate an action's impact on the forecast
Publish results
- Reports hub: your reports and available templates
- Computed Reports: CSRD E1-1 to E1-9, GHG E1-6, CDP
- Balance Sheet Report: GHG Protocol balance sheet
- Certifications, Sustainability Reports, Share Pages, Stories
- Supplier Engagement: supplier-side of a factor request
- Digital Product Passport: public QR / URL / JSON-LD
Every route and every Prisma model
Deep dive into all 165 Prisma models, the exact route file per module, and how the sidebar renders. Reads the code paths in `apps/dashboard/src/organisms/NavSidebar` and `libs/prisma/schema.prisma`.
One equation, twenty-four methods, four phases, three trust layers.
The calculation engine turns customer activity data (litres of diesel, kWh of electricity, kg of steel, tonne-kilometres) into stored, versioned, compliance-grade tonnes of CO2e attributed to a scope, a category, a business unit, and a month. Every stored emission carries a versioned audit trail back to the exact factor picked and the four subscores that picked it.
The identity
emissions = ((factor × activity + addend) × period × multiplier) / 1000
All 24 dispatchable calculation methods share this base identity. Fuel consumption, spend-based purchased goods, distance-routed logistics, home office, marketing paper, consumer transport, EcoTransIT, and all the rest.
Factor sources
DEFRA (UK government dataset). Ecoinvent (LCA). Climatiq (aggregator API, pulled on demand). GLEC (logistics). IEA / EPA / national grids for country grid mix.
Custom factors uploaded by the customer always win via the "bonus" score. Supplier-specific factors enable primary-data scope 2 market-based.
The pipeline
Trigger event lands on a FIFO SQS queue. Step Functions EXPRESS chunks the work (200 log entries per chunk) and fans out to 20 workers. Four measured phases per run:
pre_processing context calculate persist
"The new engine has shown an increase in calculation speed of almost 60x. A batch of 200,000 log entries has been calculated in less than 15 minutes."
Dual scope 2 reporting
Every scope 2 log entry produces a market-based row in organization_emissions (customer's contract-specific factor) and a mirror location-based row in organization_emissions_auxiliary_location_based (country grid mix factor). Both computed on every relevant entry; the reporting layer decides which to surface where.
WTT / TTW split
Freight comes back from EcoTransIT with three numbers: total, Tank-to-Wheel (vehicle combustion), Well-to-Tank (upstream fuel). Cozero stores two rows in organization_emissions distinguished by energy_emission_source_id. Shipped Q1 2026 for Zalando under DEV-733 with a ~1.5M row backfill.
Closed periods lock the numbers
Pre-processing SQL marks any log entry whose Log's date range overlaps a closed period as BELONGS_TO_CLOSED_PERIOD, skipping calculate and persist. Once closed, no recompute changes the numbers without an explicit recalculation flow with preview diff.
The audit trail
audit_emission_factor_v2 records every candidate factor scored, the winning factor's four subscores (geography, time, factor origin, bonus), the tie-breaker step, and the emission factor IDs that went to tiebreak. Auditors reconstruct: what was entered, which candidates were considered, why one won, what was applied, which stored row represents the result.
End-to-end walkthrough of a Vorwerk diesel calculation
One thousand litres of diesel at a German warehouse in April 2026, from log entry creation through factor selection, four calculation phases, persist to two tables, and audit trail. Plus every fragile part the engineering team has explicitly flagged.
Seven paths in, one write path out.
Every log entry, no matter how it entered, lands in the same log_entries table with foreign keys pointing at the ingestion source. Over 80% of platform log entries come through bulk import. Nippon alone drove approximately 95% of the 63 June Ingestro imports.
1. Manual log entry
Single entry via the LogEntryForm. Activity type, quantity, unit, date, business unit, location, optional supplier. Used for one-off entries and small orgs.
2. Legacy CCF bulk import
Excel template download, filled in, uploaded back. Still active for EcoTransIT-driven customers and for PCF.
3. Ingestro V2 (the flagship)
April 2026 rewrite. Embedded @ingestro/importer-react (formerly Nuvo). CSV or XLSX upload, column mapping in-browser, in-app error review. Over 80% of platform log entries.
4. Custom Data Transformation Pipelines
June 2026. Cozero engineers author a Python + DuckDB pipeline for a customer's specific ERP export shape; the customer uploads their raw file, the pipeline transforms it, hands off into Ingestro. Named customers: Mosolf, Schoeffel, Berner, Sulzer.
5. Public REST API
Documented at cozero.readme.io. API keys are read-only (new default, April 2026) or full-access. Endpoints for log entries, logs, business units, locations, suppliers, computed reports.
6. Third-party connectors
EcoTransIT (logistics, SFTP), TravelPerk (business travel, API pull), Climatiq (factor library). No first-party connectors for SAP, Oracle, SharePoint, Google Drive, BigQuery.
7. Supplier engagement
Tokenised email link. Customer creates a FactorRequest, supplier receives an email, lands on Supplier Onboarding, submits data, response flows back as FactorRequestResponse. Especially important for scope 3.1 and PCF exchange.
"The CCF bulk import is one of the most-used features on the platform. Over 80% of all log entries come from bulk imports."
Every path, every table, every bottleneck
The data model behind ingestion, foreign keys back to source, validation gates, and documented bottlenecks (API 504s at C=40, Ingestro fails at 15k rows, calc processor drains at 310 entries/min).
Closed periods, computed reports, frameworks, passports.
Cozero produces four kinds of output today: (1) computed reports for regulators and disclosure bodies, exported as XLSX; (2) share pages, sustainability reports, certifications, and stories for internal and external publication; (3) supplier engagement responses back upstream; and (4) Digital Product Passports for product-level footprints. Cozero Control adds a fifth: CFO-facing carbon-adjusted business cases, covered in its own sections below.
| Framework | Support level | How Cozero maps to it |
|---|---|---|
| GHG Protocol Corporate Standard | First-class | Scope 1, 2, 3 model in the calculation engine. Dual scope 2 reporting shipped June 2026. |
| CSRD ESRS E1 | First-class | Computed reports E1-1 through E1-9. E1-6 (GHG emissions), E1-4 (transition plan), E1-5 (energy) live. |
| CDP | First-class | Climate questionnaire mapping. cdp-disclosure-2024 and 2025 keys. |
| ISO 14083 / GLEC | First-class | Logistics methodology. EcoTransIT integration with WTT/TTW split. |
| iLEAP (Pathfinder Framework) | Partial | DPP MVP shipped October 2025. Pathfinder event push for PCF exchange. |
| CBAM | Column-level | Column-level only in LEO exports. Deprecation banner exists. OrganizationEmissionsCbam table. |
| SBTi | Structural fit, no validation | Target schema mirrors SBTi patterns but no validation, no SDA pathways, no submission workflow. V2.0 published 11 June 2026, mandatory from 2028; Patrick's Layer 8 proposal shows how the V2 alignment shift folds into Control. |
| TCFD / GRI / ISSB S2 / SEC | Not first-class | Close via CSRD E1 overlap. No dedicated computed report keys. IFRS S2 disclosure regime is the natural fit for the ROCI-produced numbers (see ROCI section). |
The reporting cycle
Cozero assumes a 12-month cycle (calendar or fiscal year). No monthly or quarterly close in the code. A closed period is a date-ranged lock keyed to a business unit, with an outdatedEmissionFactors JSON column and a usedInAct flag that gates Act's forecasting.
Intensity Metrics (shipped 2026-06-11)
Admin defines custom inputs (name, unit, subcategoryIds). Metrics appear as optional fields on log entries and as dynamic Ingestro columns. Numerator and denominator both selectable as emissions or a metric id. Priority customers: Hupac, Hoyer, Mosolf, Liebherr, Redcare.
Climate actions as scenarios (in Core / ACT)
ClimateAction has annualCarbonSaving, timeToImpact, lifetime, capex/opex, cost-per-tonne. ScenarioForecast applies an action to a Forecast and writes new datapoints. Strategy Builder plots baseline forecast, target line, action-adjusted forecast. Control renames these to reduction initiatives and adds year-resolved schedules, switch semantics, and driver-priced financials.
Product Carbon Footprint
Product-level lifecycle model on Product with functionalUnit, selectedSteps. Outputs: Product Footprint Overview, DPP (QR + link + JSON-LD), iLEAP push. Main buyer: upstream suppliers responding to downstream buyer requests for Scope 3.1.
The reporting layer, framework by framework
Which report Cozero writes, what format, which enum keys, the SBTi gap, TCFD and GRI overlap, and how Cozero Control plans to change the shape of the whole reporting side.
Finance-grade decarbonisation for the CFO, in its own app.
Cozero Control is the third Cozero app, sitting alongside Core (the accounting platform every existing customer uses) and Carbon Intelligence (the self-serve insights studio). Where Core answers "what are our emissions?", Control answers "what should we invest in to reduce them, and what does that do to the P&L?" The persona is the CFO or head of controlling, not the head of sustainability.
Signed customer: Phoenix Contact (2026-07-23), first on-site delivery 2026-08-28 in Bad Pyrmont. Pipeline: DMG MORI, Liebherr, Wilo, Duisport (Compass CW31).
Digital Twin
Layer 4
The operational substrate: nested tree of business units and locations. Reused directly from Core for connected customers (read-only in Control); managed in Control with extended entity types (Group, Division, Legal entity, Department, Function, Fleet, Region, Customer segment, Product line, Asset) for stand-alone customers. Emission centers live inside the twin.
General Ledger
Layer 5
Finance-aware taxonomy: emission type / center / object, mirroring SAP-style cost type / center / object. GHG Protocol Scope 1/2/3 + 15 categories serve as the default chart of accounts for emission types. Allocation keys split emissions across objects when direct assignment is not possible.
Drivers
Layer 6
Reusable named assumptions with year-by-year sparse curves. Three types (Activity, Intensity, Financial). Bundled into scenarios (BAU / Normal Case / Best Case ship as defaults). Projects the twin forward to 2050. One scenario evaluated at a time; multiple can be shown side by side.
"Energy = emissions = euros, decarbonization IS a pricing strategy. Cozero is not another information system. It's a bypass."
Two customer modes
Every Control feature is specced against two modes decided at setup: Core-connected (twin and CCF baseline synced read-only from Core; edits made in Core) and Control-only (twin and baseline managed directly in Control; drop-zone upload for CCF, extended entity types for the twin). Drivers, scenarios, initiatives are Control-native in both modes. Phoenix Contact is Control-only.
Own app, own repo, own dev flow
Repository: github.com/Cozero/cozero-control. Development via Coder browser workspaces at coder.cozero.io/templates/control. Non-engineers create a workspace, check out a branch, build with Claude Code, push to open a PR. Explicitly the "vibe-code builder" motion Patrick and Tom fall back on when engineering capacity is short.
How it fits Core + Carbon Intelligence
Shared-cookie SSO (decided 2026-07-23): Core issues an RS256 accessToken cookie on .cozero.io with an apps entitlement claim; Control and the Supplier app verify with Core's public key. One login, one logout. For Core-connected customers, Control reads the digital twin and CCF baseline through an agreed API contract; no writes back to Core.
Why a separate app, not a bigger Core
Enriching Core's climate_actions to carry activity/intensity decomposition, year-resolved schedules, capex/opex driver linkage, versioning, and shift semantics was judged more costly than building fresh in Control (decision 2026-07-24). Control also must work stand-alone. Upside: separate release cadence, uncluttered CFO surface. Downside: two apps to keep coherent.
Full Control briefing
The thesis, the three-pillar model, both customer modes, how Control fits Core and Carbon Intelligence, who works on it (Patrick, Maxi, Nuno C, Ghaith, Ghassen, Paz, Sonja, Ali, Tom), the technical setup, prototype status, GTM approach, what's next.
Nine layers stack Control on top of Core's carbon accounting.
Patrick's "Cube" concept synthesis (2026-05-22) stacks Control on top of Core's carbon accounting. Nine layers, each with its own Notion deep-dive spec page in the house Scope → Context → FR-x → DM-x → Prototype status → Open questions format. Two layers carry two documents (Layer 6 splits into Drivers vs Reduction Initiatives, Layer 9 splits into Control Use Cases vs Reduction Analytics).
| Layer | What it is | Status |
|---|---|---|
| 1-3 GHG Data | The CCF baseline as line items with mandatory twin-element link and optional supplier/customer/product/tag context. Synced from Core LOG for connected customers; drop-zone upload (csv/xlsx/JSON) for Control-only, into user-defined 1-year closed periods. | Prototype partial: JSON upload works, csv/xlsx open. QA table view builds on the Facts tab. |
| 4 Digital Twin | Nested tree of business units and locations. Extended entity types for Control-only customers (Group, Division, Legal entity, Department, Function, Fleet, Region, Customer segment, Product line, Asset). Location is always a leaf, never a root. | Prototype visual, no CRUD yet, sync path fixture-based. |
| 5 General Ledger | Emission type (what kind) / center (which unit is accountable, a twin node) / object (what bears the emission: product, project, contract, asset). Allocation keys split when direct assignment is not possible. Yearly lock (Patrick x Tom, 2026-07-16). | Concept-scoped in Notion. Prototype patterns from SAP-style controlling. |
| 6a Drivers & Scenarios | Fourteen FRs. Driver types (Activity / Intensity / Financial with carbon-driven vs activity-driven Financial subdivision). Sparse curves interpolated linearly, horizon 2050. Assignment on twin + GHG axes with override-vs-stacking semantics. Scenarios as bundles, mutually exclusive at evaluation. Process-level intensity breakdown added 24.07 as FR-14. | Substantially implemented in prototype; presets, exposure modifiers, fine-grained targeting open. |
| 6b Reduction Initiatives | Fourteen FRs. Renamed from "climate actions" (Core's name) to fit CFO framing (decision 24.07). Core's status set and 13-lever catalog reused 1:1. Year-resolved effect schedules. Switch initiatives (before → after, incl. inline catalog creation). Financial dimension with driver-priced opex/saving lines. Guarded hard delete. | Data model substantial; year-resolved schedules, seeded lever catalog, forecast-side switch resolution open. |
| 7 Forecast | Eleven FRs. Equation: (base activity + activity-driver effects - initiative reductions) × emissions intensity. Activity and intensity stored separately per year. Line-item grain (decision 23.07). Four data points per line item per period: emissions, activity, carbon cost, activity cost. Transparency chains per dimension. Horizon 2050. | Kernel implements the canonical model at bucket grain; line-item grain, transparency UX, financial per-line-item data points open. |
| 8 Targets | Proposal draft by Patrick, 15.07.2026. Targets on any twin level, for any GHG subset, on emissions / activity / intensity. Linkable across cycles (SBTi V2.0 5-year rolling cycles). Lifecycle: validation status, sign-off, owner, lock, audit trail. V2.0 introduces alignment (measurable transition) alongside engagement. | Proposal; ACT today is the base to extend, not rebuild. |
| 9a Control Use Cases | Three registers borrowed from financial controlling: Accounting (books are right, emission on the right center/object), Controlling (plan vs actuals per center, with a signed-off budget), Planning (EBITDA+, TCO+, CapEx planning for the transition). Connectivity principle from IFRS S2 / ESRS E1. | Concept-scoped; the Controller and Planner surfaces are follow-ups. |
| 9b Reduction Analytics | Two use cases so far: 1. Find the best reduction pathway (scenario comparison + filterability + top-level KPI cards). 2. Understand key elements of a reduction pathway (waterfall charts per emission center, default two-step target windows: until 2030, until 2035). | Linear project Control: Reduction analytics screens & charts (Patrick, target 2026-08-14, Discovery). Notion deep-dive still DRAFT: FR structure to follow. |
Every layer in detail
Requirements, decisions, prototype status per layer. Where the layers connect (the twin as inheritance axis, the ledger as translation, drivers as projection, forecast as engine). What each spec page says today, what is open.
Everything the CFO sees comes from one loop.
Authored assumptions (drivers), company actions (reduction initiatives), a bundle of both (a scenario), a projection (the forecast). The mechanics below are specced in Notion Layer 6a (Drivers & Scenarios), Layer 6b (Reduction Initiatives), and Layer 7 (Forecast).
Reduction measures are everything the company does itself. Drivers are everything others do.
Drivers in three types
Activity drivers
Change how much of something the company does or consumes. "Sold units +3% YoY", "Production efficiency -1% YoY of consumption per unit of output". Machine-efficiency upgrades are Activity: they change how much energy is used, not the carbon intensity of that energy.
Intensity drivers
Change the emissions factor per unit of activity. "Electricity decarbonisation -2% YoY", "Grid emissions intensity Germany trajectory (IEA)". FR-14 (added 24.07) lets an intensity driver decompose into process-level trajectories that roll up to the aggregated curve.
Financial drivers
Price activity or emissions in euros. Split into carbon-driven (EU ETS price, EU ETS phase-in) and activity-driven (electricity price, natural gas price, WACC, inflation adjustment). Cozero ships defaults with source and provenance.
Assignment on two axes
Twin axis and GHG axis
Drivers land on data via two nested axes: the twin (org → business unit → location → asset) and the GHG axis (scope → category → activity). Refineable via context links to supplier, customer, product, tag down to a single line item. Multi-select supported: BU A + BU C × Scope 1 + Scope 3.1 in one authoring gesture.
Override, not stacking (same quantity)
Drivers modelling the same quantity at different granularity: the most specific wins and replaces the broader one. Example: supplier-specific steel data for 20% of volume, the remaining 80% falls back to the default. Override relationships must be explicitly declared and traceable in the UI, not implicit.
Stacking of distinct drivers
Drivers modelling different quantities compose multiplicatively on the same line item. Sold units × production efficiency × grid decarbonisation. Activity and intensity components stored separately per year (DM-2, shared with Forecast). Never blended.
Scenarios as bundles
A scenario has a name, a reference year, and a set of drivers (many-to-many; the same "Pumps +5% to 2030" block sits inside Best Case and Normal Case). Publish lifecycle: draft → published → superseded → archived. One scenario evaluated at a time; multiple can be shown side by side.
The forecast, in one equation
(base activity + activity-driver effects − initiative reductions) × emissions intensity
Where emissions intensity is the initiative's intensity if a switch or intensity change applies, else the base intensity × intensity-driver effects. Per line item, per period. Activity and intensity computed and stored separately per year.
Four data points per line item per period: forecasted emissions, forecasted activity, forecasted carbon cost (CO2 × carbon-driven financial driver × exposure), forecasted activity cost (activity × activity-driven financial driver). These aggregate into carbon-adjusted figures: EBITDA+, TCO+, and the CANPV foundation ROCI 2.0 builds on.
Line-item grain, no pre-aggregation (FR-2)
Every future period holds values per line item. Users filter and zoom on the full-resolution result. The prototype's bucket grain (twin × scope/category/subcategory) is an implementation state, not an accepted scope cut. Line-item grain × monthly resolution × 25-year horizon × thousands of suppliers means millions of data points per scenario version.
Explainability chains (FR-7)
Every forecasted value decomposes into one chain per dimension. Activity chain: activity last period → after drivers → after initiatives. Intensity chain: same shape. Financial chain: same shape. Plus an applied-factors summary. Success criterion: a non-technical user sees their own configuration reflected and understands how it changed the numbers.
The mechanics in full
Sparse curves, pathway quick-select, Cozero-shipped defaults, the override/stacking rules, exposure modifiers, scenarios and their publish lifecycle, the forecast integrity rules, shift initiatives and forecast-born line items, market-based decarbonisation, analytics screens, and the Phoenix-labelled Linear tickets that turn all of this into a delivery.
How carbon meets finance.
Reduction initiatives (what the company does itself) sit on one side of the Cube. The General Ledger (type / center / object) sits on the other. Together they turn a forecast of emissions into a forecast of who is accountable, what bears the emission, and how much it costs.
The Ledger's three dimensions
Emission type = what kind of emission (Scope 1 diesel combustion, Scope 2 electricity, Scope 3.1 purchased goods). GHG Protocol Scope 1/2/3 + 15 Scope 3 categories serve as the default chart of accounts.
Emission center = which twin node is accountable (Foundry Mannheim, Central Procurement, Fleet Europe).
Emission object = what bears the emission (pump line, valve line, BioPlasticConnector product, specific assets). Cost object in accounting terms.
Allocation keys split emissions across objects when direct assignment is not possible (production volume, revenue share, m2, headcount, transport tonne-kilometres). Emissions must always add up: every emission must ultimately land on an object.
Reduction initiatives, quantified in three dimensions
Every initiative optionally carries year-by-year schedules on activity change (relative fraction or absolute), intensity change (old → new), and abatement (t CO2e fallback for coarse initiatives). Sparse-entry semantics: enter only the years you have a view on. Fleet electrification's ramp is a schedule, not a linear ramp over duration.
Switch initiatives (FR-7): move activity between line items rather than reduce in place. Supplier A → supplier B. Fleet diesel → EV (Scope 1 → Scope 2 + upstream). Explicit before → after pair with a conversion rate. Receiving side may not exist in the baseline (no EV electricity yet, no biodiesel yet): the user creates a custom catalog activity inline and supplies the receiving intensity. The forecast then materialises synthetic "forecast-born line items."
The initiative financial dimension (FR-8)
Per initiative version: capex, opex / saving lines (kind saving vs cost; source activity-derived, shift-derived, or manual; either quantity × financial driver, or manual amount, with optional inflation escalation), WACC. Financial lines reference driver versions (roles price, carbon, other). Prices are never hard-copied.
Carbon cost savings are derived, never entered. Emissions reduction (FR-4, allocated per FR-13) × the carbon-price driver of the evaluated scenario × exposure. Time value of money enters only via WACC in NPV. Abatement cost, NPV, payback follow from the same inputs.
Effect breakdown to line items (FR-13)
Users author at a realistic coarse grain: "30% less Diesel B7" on the business activity Diesel B7 at the twin root. The system allocates the effect down to every covered line item. Relative effects apply uniformly per covered line item; absolute effects apportion proportionally to reference-period share. Nobody authors measures line item by line item.
Prototype has the right allocation semantics but at bucket grain, and only for emissions/activity, not for the financial data points. Line-item grain + financial data points = Phoenix ticket DEV-2666.
Terminology conventions (24.07)
Drivers
What others do. Building blocks with year-by-year curves.
Reduction initiatives
What the company does. Called "climate actions" in Core; renamed in Control to fit CFO framing.
Decarbonisation levers
Initiative taxonomy: the 13-entry Core catalog reused 1:1 (Waste management, Facilities energy optimization, Fleet electrification, ...).
Value effects
Where the value lands. Quantifiable (cost reduced, cost avoided, revenue generated, revenue protected) or directional (1-5 badge with horizon).
One metric, three answers, one story.
ROCI has been a Cozero trademark since 2020, a modelling concept since July 2021, and an informal prototype implementation in Control since spring 2026, but never a formally specified metric. Fabian's ROCI 2.0 Methodology & Product Proposal v0.15 (draft for working-group discussion, 25.07.2026, updated 29.07 after Patrick's review) formalises the direction the Control prototype has been quietly moving toward.
ROCI is the finance-language metric. It answers "what is the return on this investment once carbon is priced into the cash flows?" The sustainability-language question ("how much abatement per euro?") is answered by MACC, ROCI's designated sibling per PMF 2.0. Both views are generated from the same underlying data.
Carbon-Adjusted Net Present Value: the foundation
CANPV = Σt=0..T (ΔOpExt + ΔRevt + ΔCarbont + ΔTaxt − CapExt) / (1 + WACC)t + RVT / (1 + WACC)T
Three properties are built into the definition, all already Layer 6 design principles: scenario-conditional (a forecast evaluates one scenario at a time; the carbon price comes from that scenario's driver, so CANPV is automatically year-resolved and scenario-dependent), with-vs-without (the counterfactual is the same scenario evaluated without the initiative; the counterfactual is an explicit, inspectable model input), discount once (time value of money enters only via WACC in NPV; uncertainty is expressed as the CANPV range across scenarios).
The user-facing KPI family
| KPI | Definition | When shown | Speaks to |
|---|---|---|---|
| CANPV | Absolute € as above | Always (always defined) | Headline; value creation |
| ROCI | CANPV / PV(CapEx). Carbon-adjusted profitability index. | Only when PV(CapEx) > materiality threshold | Ranking under a capex budget |
| CIRR | Discount rate at which carbon-adjusted flows break even | Only when flows change sign once | Hurdle-rate logic: "clears WACC by X pts" |
| Carbon uplift | CIRR − IRR (IRR = same flows with ΔCarbon set to 0) | Whenever both defined | The signature number: how much does pricing carbon improve this investment? |
| Payback | Years to cumulative break-even (undiscounted and discounted) | Always | Familiar CFO shorthand |
| Abatement cost | € per tCO2e; feeds MACC | Always | The sustainability-language sibling view |
| Directional badges | 1-5 scores on flagged value effects (fines/policy, license-to-operate, brand, financing conditions, resilience) | Alongside, never blended | Honesty about the unquantifiable |
ROI is not a separate calculation: ROI ≡ ROCI with carbon flows zeroed. Displaying both side by side makes the carbon uplift permanently visible.
Quantifiable core, directional overlay (D-2)
Adopted from Deloitte's Sustainability Fusion (2026). Fines exposure, license-to-operate, brand, financing conditions, resilience are real but too uncertain for a point estimate. Score them 1-5 alongside the number instead of blending them in. Promotion to quantifiable when a specific probability-weightable exposure exists (e.g. named GEG non-compliance). Never a silent default rule.
Cash ROCI vs Steering ROCI (D-3)
Cash ROCI: real cash flows only, incl. regulated carbon prices × exposure. What the CFO takes to the board as bankable. Steering ROCI: additionally includes internal carbon price flows. What the company governs by (and, per the Phoenix conversation on 2026-05-22, potentially ties to incentive systems). Every display names the carbon-price driver, its source, and the exposure assumption.
Edge cases dissolved (D-4)
CANPV is always defined (absolute euros, no division, no sign ambiguity). Ratio and IRR views are gated behind materiality thresholds. Initiatives with zero or negative net investment and positive CANPV are labelled "no-regret / immediately accretive" instead of displaying an explosive ratio. The 2021 worked tables become regression tests.
Independence structurally retired (D-5)
The 2021 assumption that emission and cost effects are independent, retired: in Control both come out of the same driver movement. Fleet electrification's fuel-cost saving and its emissions reduction are the same litres of diesel priced two ways. Uncertainty is expressed as the metric's range across scenarios ("clears WACC in base and high-price scenarios; misses in low-price"), not via probability distributions on inputs. The 2021 Monte Carlo layer retires with it.
Eight product requirements (ROCI-R-1 through R-8)
R-1 Metric definitions
Formally define the KPI family above. Display gating per D-4. Derived metrics never persisted on the initiative.
R-2 Exposure drivers formalise liableShare
Sliders and hardcoded CBAM phase-in become exposure drivers referencing a base carbon-price driver, year-resolved. First list: EU ETS free allocation, CBAM coverage, ETS2 introduction.
R-3 Revenue line semantics
Saving lines either formally extended or complemented by a revenue kind covering feed-in remuneration, contracted green premiums.
R-4 Tax, subsidies, capex phasing, residual value
Grants + simple effective tax rate in v1 core (Patrick 29.07). CapEx year-resolvable and grant-netted. Full jurisdiction-specific depreciation out of scope. End-of-lifetime: cash and carbon flows end at lifetime unless a residual value or persistence is declared.
R-5 Directional value effects
Head-level scores 1-5 (fines/policy, license-to-operate, brand, financing conditions, resilience). Never enters CANPV. Financing conditions frequently promotable (SLL margin grids are contractual).
R-6 Portfolio aggregation
Portfolio CANPV = Σ initiative CANPVs within one scenario. Capex-budget view ranks by profitability index. Plan-level MACC from the same data.
R-7 Driver sensitivity view
Tornado over the scenario's drivers (ETS price, energy prices, activity growth). CFO-native. Fast-follow.
R-8 Ex-post ROCI (roadmap marker)
Plan-vs-actual comparison of CANPV and carbon uplift per initiative, once actuals flow in. Ex-ante business cases are consultancy territory; realised return on climate investment computed from audited activity data is the long-run moat argument for the ROCI trademark.
Your auditor will ask for these numbers anyway. Cozero produces them as a by-product of the business case you run for yourself.
The full ROCI 2.0 briefing
The identity decision D-0 through the pilot interface D-7. All eight ROCI-R-x product requirements with anchor to Layer 6. The Deloitte Fusion mapping (adopted / not adopted). The naming split. The change table 2021 → 2.0. The sources.
Four weeks to prove the finance-grade layer.
Phoenix Contact is Cozero's first paying access to Control. The deal was signed 2026-07-23. On-site result presentation is Friday 2026-08-28 in Bad Pyrmont. Between those dates: four weeks to build a driver-based forecast for two purchased material categories (copper, plastic) and one product-use-phase category, at scale (thousands of suppliers).
Phoenix bought a service, not a software subscription. Cozero doesn't plan to onboard them at the start; they get access at the end so they can play with the platform themselves. That access is an additional argument for rollout, but the focus is achieving the promised scope. Phoenix is Control-only: no Core interaction.
The five in-scope features
Import CO2 footprint
Same data structure as Core for the imported CCF data, transformed into the right structure before upload. QA table view (LEO-like, Layer 1-3 FR-9) to filter and export entries so Cozero can stand behind the numbers presented. No edit / calculation service.
Drivers & Scenarios
Runs two tracks in parallel. Builder lead starts model implementation. Climate team gathers default drivers (IEA grid trajectories, material EFs, decarbonisation journeys). Tom and Patrick spec the process-level breakdown logic (Layer 6a FR-14 added 24.07). Second iteration adds breakdown to the model.
Reduction plan
Focus categories are Scope 3, so this is not as crucial as Drivers & Scenarios, but it must show for selected initiatives what the impact on the reduction journey is (e.g. actively switching from one supplier to another). Must support upload and integration of reduction measures from an existing file, and creation of new measures.
Forecasting logic
Success criterion: a non-technical user can look at the forecast, see their configurations (reduction measures, drivers) reflected, and understand how it impacts the forecast. Transparency chains per dimension (activity chain, intensity chain, financial chain) plus an applied-factors summary.
New analytics screens
A filterable chart where different reduction pathways (scenarios) can be compared against each other and against a baseline / reference line in one overview (own target pathways + external industry trajectories). Drill down to a specific site, supplier, material. Placement: with the analytics screens in Control's Act-like area.
Coder infrastructure
Must be functional first so everyone is enabled to build on top of it. This is what lets Patrick and Tom step in as vibe-code builders where engineering capacity is short. Prototype URL Phoenix will see: control--control-ft-ux--paz.coder.cozero.io.
Timeline
| Date | Milestone |
|---|---|
| 2026-07-23 | Deal signed. Alignment call Phoenix × Cozero. |
| End of July | Data received from Phoenix (Scope 3 inventory, material/product data for 3.1 + 3.11, primary data from suppliers). |
| 2026-07-31 | Iter 1 targets: CCF import, first Drivers & Scenarios cut, forecasting logic, analytics concept. |
| 2026-08-04 | Clarification session, 08:00. |
| 2026-08-07 | Reduction plan target. |
| 2026-08-14 | Iter 2 targets: Drivers & Scenarios breakdown logic, analytics screens implementation. |
| 2026-08-28 | Result presentation, on-site Bad Pyrmont (pending final confirmation with Lennart). Impulse workshop directly after. |
Explicit "not in scope"
- Interaction with Core. Phoenix is Control-only.
- Financial Data proper. Simplified driver-priced model only.
- General Ledger. Type / center / object is Layer 5 follow-up.
- Control-related analytics screens beyond scenario comparison (Accounting, Controlling, Planning).
Where the Linear tickets live
CCF import: Import/Sync your CCF baseline data project. DEV-2607, DEV-2608, DEV-2609, DEV-2642 all Phoenix-labelled.
Drivers, Scenarios & Forecast: single project, 9 Gherkin user stories (DEV-2643 through DEV-2650, plus DEV-2673). All Phoenix-labelled.
Reduction plan: Manage your reduction initiatives project. 11 tickets, most Phoenix-labelled. Best-organised of the four: the Notion spec carries an explicit FR → ticket table.
Analytics: only DEV-2650 (scenario comparison), inside the D&S&F project. No dedicated project. Waterfall charts (Use Case 2) have no ticket.
One cross-cutting refactor riding along
Control has no currency model in code today: € is hardcoded across ~106 places and money is stored in millions behind a _m suffix. Fabian's revenue-line work on ROCI 2.0 surfaced that "which currency is this?" has no answer. The Money (currency & units) refactoring spec (30.07.2026) sets one org-wide reporting currency, plain-unit storage, derived driver units, and formatter-level abbreviation. Linear Control: Money, currency and units [refactoring], tickets DEV-2750 (currency read and applied), DEV-2751 (plain-unit storage), DEV-2752 (admin UI, backlog). Not user-facing, but a prerequisite for adding revenue lines without multiplying an existing inconsistency.
The full Phoenix briefing
Feature by feature with Notion spec anchors and Phoenix-labelled Linear tickets. Cross-cutting audit recommendations from the internal audit at ~/Desktop/claude-temp/phoenix-control-linear-audit.md.
How the next Phoenix comes in the door
The four-sentence pre-read, the v2 pitch script (nine screens, refined after Inditex), the GTM engine's four pillars (Foundation, Marketing, Sales, Delivery), the Stage 1 → Gate → Stage 2 model with three-workshop cadence, the €20K reference anchor with 50% credit into Stage 2, and the open commercial questions (ICP, standard pricing, delivery staffing, success metrics). Consolidates "How to pitch the prototype", "GTM Engine for Control", and "Project Approach" from the Notion hub.
AI showed up as onboarding acceleration and pattern-finding, not as a chat box.
Cozero's public AI story is deliberate. The four shipped or in-flight AI-driven features are all onboarding accelerators, feedback aggregators, or synthesis engines. No chat-with-your-dashboard yet. Ghassen Labidi leads the AI Transformation initiative company-wide. Ghassen is also a contributing builder on Control, and Ghaith on the Control driver-based forecast backend.
Business Activity Mapping (BAM)
Matches thousands of customer activities against Cozero's business activity database with confidence scores. Reduces onboarding from days to hours. Full audit trail in ExternalBusinessActivitiesMappingRecommendation and Correction tables. Feature flag business-activities-mappings.
Cozero Pulse
Customer feedback aggregator built by Ghassen, Mathis Mira, and Ghaith Belaazi. 609 entries, 57 customers, 84 meetings from Demodesk processed with embeddings + LLM into 28 canonical themes. Force-directed theme graph, cross-customer connections, evidence links to Linear + Slack + GitHub. Lives on an internal VPS.
Carbon Data Intelligence
Self-serve insights studio ranked highest in the June GTM scoring exercise. Repo carbon-intelligence-poc/ (Next.js 16 + FastAPI + Vercel AI SDK). Sponsor Ignacio, built by Anja + Ghassen + Santi. Target Sep 2026.
AI Demo Agent
Mathis's agent produces customised sales demos for prospect calls. Already used for Lagermax, Doka, Telekom demos (per Compass CW23). Ships context from Cozero's own product to a demo shell so the AE walks in with a pre-primed environment.
"We need to probe how good the quality of the AI-aggregated feedback is. You probably know more examples than I do where companies have reported and acted based on hallucinated AI generated dashboards in recent months."
Reporting is the real product.
The product story frames Act (planning) and Share (publishing) as differentiators, but the numbers say Log and Share dominate. Twenty-eight percent of page views are report-related, and 48% of active clients have at least one closed period. Eight enterprise accounts account for 66% of ARR.
Feature adoption reality check
PCF: positioned as a differentiator, only 1 created per month prod-wide.
Climate Actions / Targets: "strategic planning surface", 1 created per month prod-wide.
API: "enterprise integration surface", 16 calls per month per Ainoa's November 2024 report.
Intensity Metrics: seed says 22 of 28 accounts touched in first week; comparable prod signals suggest that is off by orders of magnitude.
Where the concentration is
Prod November 2024 (latest Ainoa post): Sulzer alone drove 79% of log entries that month.
Prod June 2026 (bulk import numbers): Nippon alone drove approximately 95% of the 63 Ingestro imports.
Cozero Today seed top 3 by MAU: Hapag-Lloyd, Zalando, Peek & Cloppenburg. Real prod: Sulzer + Rhenus concentration.
The seed's flat event distribution is a known limitation. Real prod is a heavy pareto.
Ainoa Ryll's monthly "PRODUCT METRICS MONTHLY" post in the product Slack channel is the canonical source for prod product metrics. The latest is November 2024, seven months stale. Worth asking her if the cadence stopped or moved.
Fifteen customers, six industries, one geography, one Control PoC.
From the Monthly Customer Health Score DB, May-June 2026, ordered by ARR. Approximately 200 accounts total, about 65 with populated ARR, total tracked ARR approximately €2.2M. Phoenix Contact joined in July 2026 as the first Cozero Control PoC (not on the ARR roster below since Control is billed as a service rather than a subscription).
| # | Customer | Industry | Region | ARR (EUR) | Tier | Health | Sentiment |
|---|---|---|---|---|---|---|---|
| 1 | Zalando | Retail | DACH | 207,087 | 1 | At Risk | Critical |
| 2 | Vodafone Group Plc | Telco | UK | 156,000 | 1 | At Risk | Critical |
| 3 | Nippon Express | Logistics | Global / EU | 147,060 | 1 | Healthy (80) | Champion |
| 4 | DMG MORI | Manufacturing | DACH | 146,200 | 1 | Healthy (78) | Champion |
| 5 | Sulzer Management AG | Industrials | CH | 120,954 | 1 | Healthy (77) | Uncertain |
| 6 | Vorwerk Group | Consumer goods | DACH | 93,000 | 1 | Neutral (72) | Uncertain |
| 7 | AIT Worldwide Logistics | Logistics | US / EU | 84,000 | 2 | Neutral (67) | Uncertain |
| 8 | Progroup | Packaging | DACH | 69,600 | 2 | Healthy (77) | Champion |
| 9 | FUNKE Mediengruppe | Media | DACH | 58,200 | 2 | At Risk | Critical |
| 10 | Redcare Pharmacy | Pharma / e-com | DACH | 57,000 | 2 | Healthy (85) | Uncertain |
| 11 | Peek & Cloppenburg | Retail / fashion | DACH | 51,840 | 2 | Healthy (81) | Champion |
| 12 | Duisburger Hafen (Duisport) | Logistics / port | DACH | 50,640 | 2 | Healthy (80) | Champion |
| 13 | Schmitt + Sohn Aufzüge | Manufacturing | DACH | 44,640 | 2 | Neutral (67) | Champion |
| 14 | Optibelt Group | Manufacturing | DACH | 39,000 | 2 | Neutral (74) | Champion |
| 15 | Mosolf | Automotive logistics | DACH | 38,071 | 2 | Neutral (67) | Champion |
Below €38K ARR: TRICOR, Wala, MUL (Medien Union Ludwigshafen), Hupac Intermodal, Rudolph Logistik, ARS Altmann, Liebherr International (€30K, group expansion planned), Deutsche Funkturm, HOYER, KAFRIT, Rheinische Post, HÜBNER Group. Historically also active or on the roster: Maersk, Hapag-Lloyd, Katjes, Mammut Sports Group, LeanIX, Gentner Verlag, Wilo.
Deutsche Telekom (€200K) is in evaluation as of Compass CW27, the largest open pipeline. Doka (€80K) also in evaluation. Recent wins: Mosolf €215K three-year renewal, Shop-Apotheke €57K, Rheinische Post €23K, Rudolf Müller €4K. Funke chose to stay after failing with Osapiens (€32K renewal). Control pipeline (Compass CW31): Phoenix Contact (signed), DMG MORI, Liebherr, Wilo, Duisport.
Head of Sustainability picks Core, CFO signs Control, Head of Supply Chain uses both.
The written ICP definition: mid-to-large European enterprises (500+ employees or €250M+ revenue) with complex supply chains, measurable carbon exposure, dedicated sustainability function, and strategic decarbonisation ambition. Geography is DACH-plus (France, Benelux, Nordics, UK). North America only through EU-led buying centres. China and most of APAC are explicitly excluded. Cozero Control sharpens the CFO persona.
Head of Sustainability
Operational owner. Evaluator. Uses Core day to day. Cares about audit-readiness, methodology defensibility, time-to-report. Under CSRD, increasingly acts as the analyst who briefs the CFO.
CFO or Finance lead
Increasingly the budget owner as CSRD forces sustainability into financial reporting. Cares about ROI on decarbonisation spend. Cozero pitches ROCI at this persona. Cozero Control is built for this persona explicitly: the CANPV + carbon uplift framing (ROCI 2.0) is the CFO's language. Ali Arnaout is the domain advisor.
Head of Supply Chain or Ops
Dominant in logistics accounts (Hapag-Lloyd, Rhenus, Nippon, Mosolf). Cares about ISO 14083, GLEC, supplier engagement. Also the persona for Phoenix Contact's Scope 3.1 supplier engagement questions.
Four triggers
CSRD
Single biggest driver for European mid-market and enterprise. Every ICP anchor leads with CSRD depth. MUL is a €33.6K customer because of CSRD onboarding for 40+ subsidiaries by 2027/2028.
CBAM
Manufacturers and industrials importing high-carbon goods. Explicit in Sulzer, Duisport, Liebherr use cases. Also a Cozero Control target: CBAM coverage phase-in is a first-list exposure driver in ROCI-R-2.
Scope 3 procurement pressure
Logistics providers pushed by shippers to report emissions on purchased transport. Mosolf, Rhenus, Hapag-Lloyd, Nippon live because customers demand ISO 14083 / GLEC.
Auditor and IFRS S2 scrutiny
PwC audited Vorwerk's Scope 3.1 in April 2026 and flagged spend-based factors as a risk. Under IFRS S2 / ESRS E1, disclosure of anticipated carbon costs, decarbonisation capex by horizon, and scenario elasticities becomes the norm. Cozero Control produces those numbers as a by-product of the business case.
Historic pricing tiers
Live pricing is not published. It lives in an internal Google Sheet.
Nine competitors matter, two are DACH-native, none carries ROCI.
Cozero self-classifies as "Challenger" in Verdantix's 2026 Green Quadrant. Watershed and Sweep are Leaders. Cozero's positioning claims to beat on price, on CSRD depth, on logistics methodology (ISO 14083, GLEC, EcoTransIT), and on the ROCI framework. ROCI 2.0 sharpens this last claim: Cozero adopts Deloitte Fusion's quantifiable-core / directional-overlay discipline, but ROCI stays Cozero's own return metric (Fusion's TACF is workflow, not a ranking number). Source: internal Competitive Battlecard 2026 (Notion, last refreshed 15 June).
Watershed
SF · Series C · ~400 people
Wins on: very large US enterprises with dedicated sustainability engineering. 60+ integrations, 500,000+ factors, AI Agents.
Cozero wins on: price (2-5x cheaper), CSRD depth, logistics methodology, ROCI framework (finance-language metric with CANPV + carbon uplift).
Sweep
Paris + Denver · Series B · ~200
Wins on: cross-team coordination UX, supplier portal, four AI agents (Sweepy, Builder, Analyst, Writer). Managing 1B+ tCO2.
Cozero wins on: logistics methodology, ROCI, DACH regulatory depth, ERP/TMS integration maturity.
Plan A
Berlin, acquired Jan 2026
Acquired by Diginex Hong Kong for €55M. Historic customers BMW, Visa, Deutsche Bank, Trivago.
Cozero wins on: acquisition uncertainty. Talk track: "integration with a Hong Kong ESG conglomerate takes 12-18 months." Softest DACH competitor to poach from.
Persefoni
Mesa AZ · Series C · ~300
Wins on: financial institutions (PCAF certified), Deloitte/Bain/Workiva partnerships, audit-ready lineage, free Pro tier.
Cozero wins on: non-financial industries, logistics methodology, EU data sovereignty.
Greenly
Paris · Series B · ~300 · 2,500+ clients
Wins on: transparent pricing ($3,800 to $12,000/year), SME scale, brand. EcoPilot AI, 100+ integrations.
Cozero wins on: enterprise depth (Greenly is SME-first), logistics, DACH regulatory depth.
Normative
Stockholm · Series B · ~150
Wins on: managed services model, GHG-certified Climate Advisor per customer, 330,000+ factors from 16 databases.
Cozero wins on: self-serve automation vs advisor dependency, scalability, logistics, ROCI.
Carbmee
Berlin · Series A €20M Dec 2024 · ~100
Closest DACH competitor. ERP-native (deep SAP), ISO 14067 PCF at transaction level. Maersk is a Carbmee customer.
Cozero wins on: transport methodology (ISO 14083 for logistics vs Carbmee manufacturing focus), ROCI, CSRD breadth, production readiness.
osapiens
Mannheim · Series C €85.8M Jan 2026 · ~550
Germany's first 2026 unicorn. Compliance suite of 25 modules (CSRD, CSDDD, EUDR, LkSG, whistleblowing, due diligence, carbon). IDC MarketScape Leader 2026 for Carbon Accounting.
Cozero wins on: carbon depth (osapiens = 1 of 25 modules; Cozero is purpose-built), logistics, ROCI, decarbonisation planning.
The full customer briefing
Complete customer roster, six ICP criteria and tiering, competitive landscape with the softer entrants, pricing tiers past and present, recent commercial signal from the last 90 days (wins, losses, at-risk accounts), and the customer success motion (CHS v3, QBRs, escalation model).
Three engineering squads for Core, one cross-functional initiative for Control.
Founders sit at C-level. Function heads own the day-to-day. Engineering is deliberately small: eight named engineers plus Arnaldo. The current three-squad structure was formalised at Compass CW23 in June 2026, replacing the older Blue / Green / Orange colour squads. Cozero Control has its own cross-functional roster on top of the squads, with Tom accountable end-to-end and Maxi as builder lead.
Helen Tacke
Public face. Currently Interim Sales Lead since the May 2026 Head of Sales exit. Business Insider Top 25 Zukunftsmacherinnen. Commercial lead for Control.
Fabian Schwarzer
Company operations, financial discipline, board relations. Author of the ROCI 2.0 Methodology & Product Proposal. Commercial lead for Control.
Anja Ruisinger
Runs the joint Product-Data-Engineering unit from Q1 2026 reorg. Co-owns the Carbon Data Intelligence prototype.
Arnaldo Tema
Joined January 2026. Owns Platform team, delivery process, DORA, hiring pipeline. Author of the monthly Engineering Newsletter and board updates.
Ignacio Ghiorzi
Owns Data Ingestion & Validations, Emissions Calculation, Reports. Sponsors Unified Data Ingestion and Data Quality & Trust.
Tom Scharnberg
End-to-end accountability for Cozero Control. Owns Insights, ACT, Intensity Metrics. Sponsor of most H2 initiatives in Linear. Contributing builder on Control; operational build coordination.
Patrick Schmitt
Ran the Climate track through the recent GutCert GHG Protocol audit. Owns industry-specific onboarding and methodology. Domain expert on Control; responsible and accountable for methodological robustness, marketing / sales / delivery GTM. Author of most Layer deep dives.
Maxi Fried
Named builder lead for Control. Also on the Engine squad in Core.
Nuno Carvalhão
Builder lead for Carbon Intelligence. Contributing builder for anything shared between CI and Control. Returned from parental leave late June 2026; capacity is watchable.
Ghaith Belaazi
Builder lead for the Control driver-based forecast backend. Contributing builder across Control.
Ghassen Labidi
Head of AI within the Data team. Built Cozero Pulse and the Cozero Control Claude prototype in a one-day challenge. Domain expert on AI engineering; contributing builder on Control.
Paz Arando
Sponsor of the UX Revamp initiative. Consulted on Control design and UX. The Control prototype's Coder branch runs on Paz's workspace.
Sonja-Elena Pennanen
Contributing domain expert on Control default drivers and scenarios (Climate team). Leads "Control: Define default scenarios and drivers" (Linear).
Prof. Dr. Ali Arnaout
Decarbonaid GmbH; former CFO/CIO at AIDA Cruises / Carnival. Not on the board or team but the single most important external voice on Control. Opened Q2 board with 45 minutes that reshaped the thesis (28 May). Returned 10 June for a five-hour hands-on Control workshop in Berlin.
Katie Boyd
One of the three GTM scorers for H2 prioritisation. Manages the enterprise book with Anouchka.
Etienne
Expanded to full GTM scope since Marcel Klock's May exit. First time marketing and sales are aligned as one motion in 2026.
Hugo Vieira
Infrastructure, CI/CD, Platform reliability agenda. Leads the next observability wave (push gateway for Lambdas, per-org labels, 32-to-10 dashboard consolidation) and the CI/CD Pipeline Overhaul.
Three engineering squads (formalised Compass CW23)
| Squad | Slack | PM | Engineers | Domain |
|---|---|---|---|---|
| Platform | #platform-squad |
Arnaldo | Hugo, Rob Berglund | Reliability, DevOps, CI/CD, auth (SSO), users, orgs, billing, API keys, logging, feature flags, cache, mail, secrets. Anything cross-cutting. |
| Engine | #engine-squad |
Ignacio | Maxi Fried, Rui Fernandes, Diogo Pires | Data ingestion (Ingestro V2, direct log entries, connectors), calculations engine V1 and V2, EcoTransIT (TTW/WTT), 22+ methodologies, LEO Snowflake refresh, ADS Mapper, BAM. |
| Insights | #insights-squad |
Tom | Santiago Rojas, Nuno Jacinto, +1 incoming | Strategy Builder, Climate Actions, Targets, Action Simulator, Hotspot Analysis, Home & Tasks, Overview Dashboard, LEO read API, CubeJS, Forecasting, Carbon Intelligence prototype, CONTROL. |
Cozero Control at the top of the H2 slate.
Cozero organises strategic bets as Linear Initiatives, cross-referenced against a Notion tracker. H2 2026 slate is three streams: Innovation, Core Product, Capabilities. Cozero Control moved from "prototype phase, AI-accelerated as time-boxed Claude build" (Compass CW25) to "first paying PoC signed, on-site delivery scheduled" (Compass CW31). Deliberate framing at CW25: "central to our differentiation, so we prove it before we scale it: build a powerful prototype, validate it in GTM, and find a paying co-creation partner before any full production build."
Cozero Control
Finance-grade decarbonisation. Ali Arnaout workshop 10 June crystallised the model: Digital Twin + General Ledger + Drivers. Phoenix Contact signed 23.07 as first paying PoC, on-site delivery 28.08. Ownership: Tom accountable, Maxi builder lead, Patrick methodology, Ali advisory. Target Dec 2026 for the productionised build.
Carbon Data Intelligence
Self-serve insights studio. Ranked highest in the June GTM scoring exercise. Repo carbon-intelligence-poc/, Next.js 16 + FastAPI + Vercel AI SDK. Sponsor Ignacio, built by Anja + Ghassen + Santi + incoming Senior SWE. Target Sep 2026.
ROCI 2.0 methodology
Fabian's v0.15 proposal (25.07, updated 29.07). CANPV + carbon uplift + Cash/Steering + quantifiable/directional split. Eight product requirements (ROCI-R-1 through R-8) written in the Layer 6 house format so they lift into the spec pages directly. Cross-project with Control and Ali's Finance-Grade Decarbonization Pilot.
Intensity Metrics
Went live mid-June. Test case for "intruder-mode" delivery: Berlin hackathon PoC to production v1 in 1-2 sprints on Katie's persistent CS signal. Priority customers Hupac, Hoyer, Mosolf, Liebherr, Redcare.
Data Quality & Trust
Retention must-have. Data Validations behind flag, CDT shipped June, ADS Mapper monitoring live. Sponsor Ignacio, target Dec 2026.
Unified Data Ingestion
Fixes the binary-batch failure mode. Aggregated validations, API-side deduplication. Sponsor Ignacio, target Oct 2026. "Ingest any file" consciously deprioritised.
Company-wide AI Transformation
Three-phase Coworkshops (basics, team plugins, agentic workflows). Claude subscriptions rollout. #ai-transformation Slack. Powered BAM, the AI Demo Agent, the Claude Skill for report generation, Cozero Pulse, and the Ali-Arnaout Control Claude prototype. Sponsor Ghassen, target end June 2026.
UX revamp
Backed by the Gaia design system starter. Sponsor Paz Arando, target Dec 2026. Consulted role on Control UX.
Stabilize Cozero's Platform
Umbrella for the four engineering priorities: Observability, Deployment Speed, Calculations Engine Reliability, Data Ingestion. Sponsor Arnaldo, target end June 2026.
"CONTROL is central to our differentiation, so we prove it before we scale it: build a powerful prototype, validate it in GTM, and find a paying co-creation partner before any full production build."
The full leadership + strategy briefing
Every named leader with their scope, every board member and observer, every squad's ceremonies, every strategic bet with Linear ticket, the full Compass narrative arc CW21 through CW27, and the current hiring pipeline.
Compass every Monday, Sprint Ice Cream now.
Team members are called Coheros. The values are Passion for Sustainability, Radical Candor, Ownership, Bias for Intentional Action, Collaborate and Include. Cadences are deliberately few: one all-hands, one all-department refinement, one shared retro, one weekly Tech Doc Review, one fortnightly Tech Forum.
Compass
Weekly all-hands, Mondays. Biweekly deep dives. Structure: AI news of the week, founders + KPI, people + customer updates, deep dives, team updates, Q&A. Every session recorded and archived.
Sprints
Biweekly per squad + PMs. Sprints have names (currently "Sprint Ice Cream"). Sprint Refinement weekly Tuesday, whole department. Sprint Retro biweekly Wednesday, single shared retro.
Tech Doc Review
Weekly Friday, Engineering + Data. Mandatory for calc engine, ingestion contracts, Prisma schema, platform infra. Fortnightly Tech Forum for cross-squad share-outs.
No-Meetings Thursday
Protected focus day.
Sheriff Rotation
Weekly, all engineering. Sheriff hosts ceremonies, triages the bug board, assigns bugs.
Offsites
Berlin Tech Days May 11-13 2026 (Numa Torstraße). Summer offsite July 1-3 2026 at Gutshof Fredenwalde, six-year anniversary.
Values
Passion for Sustainability
The umbrella. Aligning work with Cozero's mission to reverse climate change and contribute to a sustainable world.
Radical Candor
Direct + empathetic feedback. Transparency even when uncomfortable. Open debate then commitment.
Ownership
Accountability for outcomes. Root causes not symptoms. Operating with transparency and reliability such that trust replaces micromanagement.
Bias for Intentional Action
Speed of implementation, prioritisation, balance perfection with speed. "Win fast, lose cheap." Dares to take trials.
Collaborate and Include
Diverse teams, cross-team help, company needs above individual preferences.
Leadership pillars (V2)
For people managers: Psychological Safety, Lead with Trust, Coach Your People, Show Appreciation. Backed by the Cozero Leadership Academy.
The company runs its own decarbonisation on the platform.
Cozero measures its own emissions inside Cozero. Practice what we preach. Accounting periods for 2022 through 2025 all live in Notion, each with scope 1, 2, and 3 breakdown backed by the platform's own tables.
What Cozero tracks internally
- Scope 1: heating
- Scope 2: electricity, water
- Scope 3: HR purchasing, AWS server usage, tech services, business travel, employee commute, off-site emissions, use of sold product
Methodology alignment
GHG Protocol, ISO 14064, ISO 14083, GLEC, iLEAP. GHG Protocol certification in flight with GutCert: audit day 9 June 2026, non-conformance items closed by 26 June.
Net/gross calorific value separation for fuel is now live, driven by that audit.
Data sources
AWS Customer Carbon Footprint Tool for server usage. Employee survey for HQ Scope 3 (commute, off-site). Manual entries and connectors for the rest.
Not held
No B Corp certification. No ISO 14001 environmental management certification. Internal Notion page titled "Lack of certification for Cozero" is worth reading if the topic comes up. Cozero's stance: alignment with the reporting-standard methodologies is what matters for customer compliance, not operational EMS certifications.
External participation
Founding participant in Carbon Alliance: 500+ organisations representing 85,000+ carbon footprints. Consistent thought leadership in DVZ, Handelsblatt, WirtschaftsWoche, OMR Festival, BVL Supply Chain CX, Impact Festival, Greentech Live.
Recent public wins
June 2026 newsletter case study (Neil, customer): "Cozero gave us a system of record we could actually defend. We moved our CDP rating from a C to a B in a single cycle." Verdantix 2026 Green Quadrant: Challenger classification.
Where to go next.
Everything above is grounded in one of six sources: the code, Notion, Slack, the Compass minutes, the Head-of-Engineering memory files, and public press. The pointers below are the fastest paths to verify or extend any claim. Cozero Control lives primarily in Notion; every Layer FR/DM referenced in the Control sections maps to a Notion page ID below.
Code (Core)
core/apps/dashboard/src/organisms/NavSidebar/NavSidebar.useMenuItems.tsx: sidebar and module groupingcore/libs/utils/src/dashboard-routes.ts: full route mapcore/libs/prisma/schema.prisma: 165 domain modelscore/libs/calculations-v2/: v2 calc enginecore/apps/api-gateway/src/log-service/: ingestion controllerscore/apps/api-gateway/src/connector-service/climatiq/: factor API integration
Code (Control)
github.com/Cozero/cozero-control: separate repo- Coder templates:
coder.cozero.io/templates/control - Prototype URL:
control--control-ft-ux--paz.coder.cozero.io - Prototype in Claude Design (v3):
claude.ai/design/p/f1da8520-9d47-4170-9b6e-b90dd9fcec03
Notion (Core, canonical)
- Cozero Feature Map: 82040abbd0a8496eb0bb0459c5d52117
- Product Roadmap: 61a9d51afb794f51b4dfb4f7bd2e9f4f
- ICP & Tiering: 26919224b975801dae4ee7ff99259c6c
- Monthly Customer Health Score DB: f004ddc7ce484ab0be363d82f727ee79
- Competitive Battlecard 2026: 34b19224b9758161bc46ff06c58ec612
- Cozero Emissions Accounting: 629faa814a1140daa9a1216819d91ad3
- Cozero Home: f129285e89b44420bd9228b18573cc8e
Notion (Cozero Control hub)
- Cozero Control root: 36e19224b9758094b395f3f8dc288906
- Idea: The Cube: 36819224b97580bf8881efec5ced4ab3
- Layer 1-3 GHG Data: 37119224b9758066925edcd82b8c05a0
- Layer 4 Digital Twin: 39f19224b97580c2895bcf273a86c2c3
- Layer 5 General Ledger: 36819224b97580a58f58d317acd15ba7
- Layer 6a Drivers & Scenarios: 36819224b9758057b517d07a54640fc1
- Layer 6b Reduction Initiatives: 39f19224b97580808d3eefd7944ada2f
- Layer 7 Forecast: 39219224b97580828d34d573d4899f71
- Layer 8 Targets: 39e19224b97580d2b7c3f36319226788
- Layer 9a Control Use Cases: 36d19224b97580ada6b0ed0a3182b1bc
- Layer 9b Reduction Analytics [DRAFT]: 3aa19224b97580dab3f1fe1bf6d42257
Notion (Control adjacent)
- ROCI 2.0 Methodology & Product Proposal v0.15: 3a819224b97581f68069e003be7cd8a8
- Phoenix Contact hub: 3a619224b975809e841ac14f662046c0
- Phoenix Build Plan (23.07.2026): 3a619224b97580aaa12bf49cf74b03b6
- Alignment Phoenix x Cozero (23.07.2026): 3a619224b97580009e7acf5b6135cf4d
- Data requirements beyond CO2: 39f19224b9758050b841d50f13656300
- Money (currency & units): 3ad19224b975817ca2cafb9f0e66a04d
- Control <> Carbon Intelligence <> Core App: 39e19224b97580fa98cbd37b9f659933
- How to pitch the prototype: 37119224b975807e97b0e2cf04b78ef2
Linear (canonical)
- Initiative: Decarbonization Intelligence / Cozero Control
- Project: Import/Sync your CCF baseline data
- Project: Create/Sync your digital twin
- Project: Drivers, Scenarios & Forecasting
- Project: Manage your reduction initiatives
- Project: Define default scenarios and drivers (Climate)
- Project: The Ledger as reconciliation bridge
- Project: CI & Control: Auth and Permissions
- Project: Control & CI: UX integration (nav bridges)
Compass minutes (last 10 weeks)
- CW21 (May 18): customer re-allocation
- CW23 (Jun 1): Q2 board recap, Ali Arnaout, David Fischer step-down, Marcel exit
- CW24 (Jun 8): H2 bets
- CW25 (Jun 15): H2 synthesis, "Carbon = Capital", Control framing
- CW26 (Jun 22): six months of rebuilding, calc engine v2 rollout
- CW27 (Jun 29): six-year anniversary, pipeline update
- CW31 (Jul 27): Phoenix Contact signed, Control pipeline listed
Slack channels of note
#product-announcements: all release announcements#platform-squad,#engine-squad,#insights-squad#ai-transformation: AI initiative#sales_internal,#customer-success,#p1-escalation#product: Ainoa's monthly product metrics post#proj-calculation-engine-rework,#prs-and-qa-reviews
External sources cited
- Deloitte & Aspen Institute, Sustainability Fusion (2026)
- SBTi Corporate Net Zero Standard V2.0 (published 11 Jun 2026)
- Flores & Martins (2026), Sustainability-related financial information and valuation: A case study of Vale, SSRN 6977704
- SpeakFinance webinar deck (20.05.2026)
- PMF 2.0 Strategy Paper (CW-41-2025, ROCI excerpt)