Monday morning, 8:30, a coffee going cold on the corner of the desk. Anneke — head of sustainability at a 400-person precision components manufacturer with plants in Eindhoven, Wrocław, and outside Stuttgart — is staring at the FY2025 kickoff email from her CFO. Group audit wants the GHG inventory locked by end of May. Last year it took until mid-August. Last year also cost €138,000 in consulting fees, plus her own two-month opportunity cost, plus the internship her team didn't get to hire because the budget went elsewhere.
She opens last year's deliverable. Forty-two tabs. The Scope 3 workbook refuses to open in Excel without a "recovering file" dialog. Category 1 (Purchased goods and services) is populated by spend-based factors the consultants pulled from a 2019 DEFRA table. Category 11 (Use of sold products) is a single cell with the word "immaterial" in it and no methodology note. The market-based Scope 2 column is blank.
Her three plant managers haven't even sent the 2025 energy data yet.
There are three ways to deal with this. She has now tried all of them.
Option A: Hire a Consultant (The "Everyone Else Is Doing It" Approach)
This is what the board expects on year one, and often year two, and sometimes year five. You call a Big Four firm or a boutique climate advisory, get a 14-page proposal with a Gantt chart, and sign for somewhere between €100,000 and €300,000 depending on how many plants you have and how nervous you look.
Here's the actual shape of the engagement, stripped of the methodology slides:
Week 1–2: Kick-off. A senior manager and two analysts want interviews with each plant controller, the facilities team, procurement, fleet, and HR. You send them your organisational chart. They send you a 62-column data request template covering fuels, refrigerants, grid electricity per site, renewable PPAs, T&D losses, fleet mileage, business travel, employee commuting, waste, water, and the full Scope 3 suite. You forward it to your plant managers, who reply that they've never seen half of these categories before.
Week 3–5: Data collection hell. The analysts chase your controllers for utility bills. The Wrocław plant sends kWh in one file and MJ in another. The Stuttgart plant's solar PPA contract is in German legalese that the junior consultant misreads — they assume it's 100% additional when it's actually unbundled guarantees of origin, which changes the market-based Scope 2 number by 38%. Nobody catches it for another three weeks.
Week 6–7: Scope 3 horror. The analysts map your procurement ledger to EEIO emission factors (spend-based, category 1) because your suppliers won't share product-level data. For category 11, they sheepishly ask whether your products consume energy in use. You say yes, obviously, they're precision servo components that run 24/7 in customer factories. They didn't ask last year.
Week 8: Draft report. You find four errors — the PPA issue, a unit mix-up on refrigerant leakage (kg vs tonnes), a double-count of a leased warehouse in Scope 1 and Scope 3 category 8, and an outdated GWP value for R-410A. The consultants issue a revision. You sign off, tired.
Total time: 6 to 8 weeks of elapsed calendar, 60 to 100 person-days on the consultant side, plus roughly 25 of yours. Total cost: €100,000 to €300,000.
What you get
- A filed GHG inventory, probably defensible at limited assurance
- A PDF report you can hand to the CFO
- An email chain you can forward to the auditor
What you don't get
- Any of the calculation files in a form you can reproduce next year
- Year-over-year consistency that doesn't depend on the same individuals being free in FY2026
- A Scope 3 methodology that isn't 70% spend-based
Option B: Carbon Accounting Platform (The "We Bought Software" Approach)
After year two of consulting invoices, Anneke's CFO said the word "platform" in a board meeting and the IT team was asked to run an RFP. She evaluated five: Persefoni, Watershed, Plan A, Greenly, and Pulsora. The pitches blur together after the third demo. They all do Scope 1/2/3, all have emission factor libraries, all claim CDP-ready exports, all charge between €20,000 and €80,000 per year depending on entity count and module selection.
Here's what the annual cycle looks like after onboarding:
Week 1: Data refresh. You log in, the previous year's structure is still there — good. You re-upload the year's utility bills per site, fleet data, refrigerant logs, HR headcount for commuting, procurement spend by category. The platform ingests CSVs, has a few pre-built connectors (SAP, Workday, a handful of utility APIs), and auto-maps line items to the built-in factor library.
Week 2: Scope 3 slog. This is where the platform earns its keep and also where it breaks. Category 1 runs cleanly on spend-based factors — you click through. Category 4 (upstream transport) requires you to either upload carrier data or accept a distance-based estimate. Category 11 still needs you to model product use-phase assumptions manually in a calculator tab. Category 15 (investments) the platform flags as "not configured — contact your CSM."
Week 3: Review and validation. The platform flags anomalies — a spike in Stuttgart gas that turns out to be a December cold snap, not a data error. You fix unit mismatches the system couldn't parse (the German invoice with "m³" that was actually "Nm³"). You reconcile market-based Scope 2 manually by uploading the PPA contracts to a document library the platform can store but not actually read.
Week 4: Export. CSV for the auditor, dashboard for the board, CDP-formatted output for the June questionnaire.
Total time: 2 to 4 weeks, roughly 12 to 20 of your own days. Total cost: €20,000 to €80,000 per year, plus onboarding the first year (€15,000 to €40,000).
What you get
- Year-over-year continuity in one system
- A factor library you didn't have to build
- Decent dashboards for the CFO and the board
- CDP/ISSB export templates
What you don't get
- Reading the supplier PDFs for you. You or an analyst still parse them.
- Scope 3 categories 11 and 15 solved — most platforms punt these to manual calculators
- Cross-framework reuse — your CSRD filing still lives somewhere else, as does CBAM, and the same datapoints get retyped
- Someone to flag that your market-based Scope 2 is wrong because the PPA is unbundled GOs, not additional generation — the platform takes the field at face value
Option C: The Formist AI Agent (The "Drop the Folder In" Approach)
This is what Anneke tried in FY2025 after her CFO forwarded her a link to Formist from his LinkedIn feed. Formist is an AI-powered compliance platform built by WeCarbon. It works like a knowledgeable colleague sitting next to you — you talk to it, upload your documents, and it fills out the sustainability compliance forms for you. It supports GHG Protocol, CBAM, CSRD/ESRS, EU Taxonomy, CDP, ISSB, LCA, SBTi, and a dozen more frameworks on the same account.
Here's her afternoon:
Hour 1 — She opens a chat, picks the GHG Protocol skill, and drops in a folder: three zipped subfolders, one per plant, each containing 2025 utility invoices (PDFs, some scanned, two in Polish), refrigerant service logs, fleet fuel cards, the two PPA contracts, an HR headcount export, and the procurement ledger as a CSV. Formist acknowledges each document, extracts totals per site, asks two clarifying questions ("The Wrocław PPA — is this bundled with certificates, or unbundled GOs?" "The Stuttgart refrigerant log entry of 12.4 kg R-410A — is this a leak event or a top-up?") and builds up the Scope 1 and Scope 2 cards. Both location-based and market-based Scope 2 are computed in parallel. The agent cites each number to the source line in the PDF it pulled it from.
Hour 2 — Scope 3, which has historically been the multi-week slog. Formist walks her through all 15 categories explicitly. For category 1, it ingests the procurement CSV and proposes a hybrid approach: supplier-specific factors for the top 12 vendors (it has them on file from prior users and from CDP disclosures), spend-based EEIO for the long tail. For category 4, it reads the carrier invoices and derives tonne-kilometres. For category 11, it asks about product lifetime and typical duty cycle, proposes an assumption, and shows the sensitivity — "a 10% change in assumed use-phase hours moves category 11 by 8% of total footprint." For categories 8, 13, 15, it flags what data is missing rather than silently defaulting to zero.
Hour 3 — Validation. Formist runs an internal consistency check: Scope 1 + Scope 2 against last year (pulls FY2024 from the consultant's PDF she uploaded), flags a category 1 variance and explains it (procurement grew 14% YoY; emissions grew 9%, consistent with supplier decarbonisation in two top vendors). It asks her to confirm materiality thresholds, generates the full inventory workbook with source citations on every cell, and exports to Excel, CSV, and a CDP-ready JSON payload.
She hits submit at 16:40. Takes the dog out.
Total time: about half a day. Total cost: a subscription fraction — low four figures a year for a mid-cap, substantially less than year one of a platform and two orders of magnitude less than a consultant.
What you get
- Every Scope 3 category addressed explicitly, including 11 and 15, not silently dropped
- Market-based Scope 2 that understands the PPA contract language
- Source-document citations on every reported number
- The same account handling next month's CDP questionnaire, CBAM quarterly, and CSRD draft — with the same underlying datapoints
What you don't get
- A consulting partner's signature on the cover page (if your credit facility requires one, you still need that)
- Someone to blame at the golf club when the number changes year over year
The Full Comparison
| Dimension | Big Four / Boutique Consultant | Carbon Accounting Platform | Formist AI Agent |
|---|---|---|---|
| Filing time | 6 – 8 weeks elapsed | 2 – 4 weeks elapsed | Half a day to draft, ~1 day with verification |
| Expertise required | None in-house; full outsourcing | Medium — your team runs the tool and owns methodology | Low — Formist AI agent guides the methodology conversationally and justifies each choice |
| Labor cost | €100K – €300K annual fee, plus ~25 of your days | €20K – €80K/yr subscription, 12 – 20 of your days | Subscription (low thousands/yr), ~0.5 – 1 day of your time |
| Accuracy | Depends heavily on the individual associates staffed | Standardised library; weak on unstructured inputs (PDFs, contracts) | Reads source documents directly; flags cross-year variance and methodology sensitivity |
| Data traceability | Spreadsheets you didn't build; methodology notes in a 90-page PDF | Data stored in the platform; source documents attached but not parsed | Every cell links back to the invoice line or contract clause it came from |
| Risk management | Consultant's reputation is the backstop | Platform validation flags obvious errors; subtle methodology issues pass through | Agent catches PPA misclassifications, unit mix-ups, GWP version drift, and YoY inconsistencies |
What This Looks Like at Scale
Anneke's company has three plants, roughly 400 suppliers, and a product portfolio that spans six main SKU families. That's a normal mid-cap footprint. The scaling behaviour of each option is the part the pricing pages don't show.
Consultant: Costs roughly scale with plant count and Scope 3 ambition. Add a fourth plant — €30K to €60K more. Move from spend-based to supplier-specific category 1 — another €40K easily. Ask for ISSB S2 alignment on top of GHG Protocol — that's a separate workstream. The more seriously you take the inventory, the more the invoice grows.
Platform: Subscription scales with entity count and module count, often in 20% step-ups. But your team's time scales linearly with data volume: more suppliers, more invoices, more manual reconciliation. Platforms do not actually read your PDFs — they ingest the CSV you extracted from the PDF.
Formist AI agent: The time cost stays near-flat as scope grows, because the agent is the one doing the reading. Triple the supplier count and you're uploading a bigger ledger, not spending triple the hours. Add a plant and it's one more folder. Year two reuses the methodology choices from year one — the PPA classification, the category 11 lifetime assumption, the supplier-specific factor library — so the second annual inventory takes hours, not half a day.
The real compounding effect is cross-framework. The same Scope 1/2/3 numbers feed your CDP questionnaire in June, your CBAM quarterly in July, your CSRD E1 draft in Q4, and your SBTi progress report. On a platform, you often re-key the same datapoint into three modules. With consultants, you pay three separate engagements. Formist treats the datapoint as a datapoint and the frameworks as views on it.
Scope 3 Is Where the Year-Over-Year Lie Happens
If you've been doing this a few years, you've noticed something uncomfortable. Your Scope 1 and Scope 2 numbers move predictably year to year — fuel consumption, grid factor shifts, PPA additions. They're auditable. They make sense.
Your Scope 3 numbers jump around in ways that have nothing to do with your actual emissions.
Category 1 swings 20% because the consultant switched EEIO vintages. Category 11 appears in year three after being "immaterial" for two years, because a different analyst finally asked about use-phase. Category 15 shows up when your CFO mentions the new investment portfolio. The trend line is noise, not signal, and any decarbonisation target you set against it is meaningless.
Limited assurance in 2026, reasonable assurance coming down the pipeline for ESRS E1 in 2029, CDP's sharper methodology scoring, SBTi's FLAG rules — all of these reward year-over-year consistency and penalise the "new methodology every year" pattern that consulting engagements naturally produce. Every time the team staffed on your account rotates, you get a methodology drift. A platform locks the methodology but only as deep as its factor library goes; unstructured data decisions still bleed.
A Formist AI agent locks every methodology choice to an auditable prompt-and-source record and replays it next year with the same assumptions, flagging anything that changed in the data rather than in the method. That's the version of consistency an assurance provider actually wants to see.
From Here Through May
Anneke doesn't have to unwind her existing setup. She kept the platform subscription through Q1 and used Formist in parallel for the FY2025 inventory. The outputs reconciled within 1.4% on total Scope 1+2, and within 6% on Scope 3 — the Scope 3 gap was entirely in category 1, where Formist's hybrid supplier-specific factors were materially more accurate than the platform's pure EEIO defaults. She showed both numbers to the audit committee with the reconciliation. They picked Formist's.
Three practical moves for anyone in the same spot:
- Run Formist against last year's inventory before you run it against this year's. It takes an afternoon. If the numbers reconcile to within a few percent, you've de-risked the switch and given yourself a defensible comparison for the board.
- Don't cancel anything until you've closed one full annual cycle. Consultant contracts and platform subscriptions are sunk cost for their remaining term. Run parallel, compare, decide.
- Keep a specialist human on retainer for the narrow, high-stakes calls. Materiality disputes, sector guidance interpretation, a new-to-you Scope 3 category — these are worth a senior's time. Annual inventory bookkeeping isn't.
The GHG Protocol turned thirty years old this year. It was designed when carbon accounting was a niche corporate responsibility exercise. Today it's the substrate underneath CBAM, CSRD, ISSB, CDP, SBTi, and most of the supply-chain disclosure requirements tightening in 2026 and 2027. Treating it like a two-month consulting project, every year, forever, stopped making sense a while ago. An afternoon is enough.
Formist is built by WeCarbon, a climate-tech company with offices in Shanghai, Paris, and Dubai. The platform handles GHG Protocol, CBAM, CSRD/ESRS, EU Taxonomy, CDP, ISSB, SBTi, LCA screening, and 15+ other sustainability frameworks in a single account.