Trueness.ai: enterprise financial case
25 October 2026. USD excluding sales taxes; 36 operating months. This replaces the earlier project/subscription model. It implements the new-programme signing plans in business-plan section 8, the first-year engineering range in platform section 6, and the subsequent agreed per-cell audit and wave design. Demand, pricing, costs and renewals are unvalidated management assumptions. The workbook tests arithmetic, not investability.
Results: revenue recognition, not signed annual value
| Case | Y1 revenue | Y1 EBITDA | Y2 revenue | Y2 EBITDA | Y3 revenue | Y3 EBITDA | Y3 gross margin | Funding +20% |
|---|---|---|---|---|---|---|---|---|
| Low | $636,667 | $-1,510,833 | $2,725,833 | $-1,702,375 | $6,123,542 | $-2,403,156 | -8.6% | $8,576,838 |
| Mid | $1,267,500 | $-1,292,500 | $8,796,562 | $1,171,766 | $24,092,578 | $6,692,926 | 40.7% | $2,892,244 |
| High | $2,216,667 | $-1,040,833 | $22,251,667 | $10,318,917 | $62,896,667 | $33,801,100 | 60.8% | $2,176,760 |
The low case is a funded downside/stop scenario, not a recommended offer: its costs exceed its ACV before overhead. It is retained because the owner requested these assumptions. Do not commit its entire three-year funding requirement. Reprice, narrow or stop at the pilot gate. The mid and high cases require substantial sales and delivery organisations; neither is a solo operation.
Case definitions and interpretation
| Input | Low | Mid | High |
|---|---|---|---|
| New programmes signed Y1/Y2/Y3 | 3/6/10 | 5/14/28 | 8/24/50 |
| Annual programme value | $500,000 | $750,000 | $1,000,000 |
| New audits Y1/Y2/Y3 | 2/4/6 | 4/8/12 | 6/12/20 |
| Annual non-renewal | 25% | 15% | 10% |
| Initial sales cycle | 9 months | 6 months | 6 months |
| Fieldwork per active service year | $300,000 | $220,000 | $180,000 |
| Production AI per active service year | $20,000 | $15,000 | $10,000 |
| Incremental licence revenue from Y2, proxy to programme revenue | 0% | 10% | 25% |
| Fixed platform build Y1 | $600,000 | $750,000 | $900,000 |
Annual signings are new wins, not year-end customer totals. M1 contains one conditional anchor already signed; other Y1 programme wins follow the stated sales cycle. Two pilot audits sign in M3/M4; additional audit wins are spread through the year. Programmes activate two months after signature. Future-year signings assume an existing pipeline, with lead months shown in the sales-funnel CSV. No customers are actually signed by this model.
Wins are integers; retained revenue uses probability-weighted annual cohorts. Churn occurs on the service anniversary, not each month during a committed annual term. There is no automatic audit-to-programme revenue: any conversion is already inside the new-programme totals. The 25% opportunity win rate in the funnel is a capacity hypothesis, not measured conversion. An audit-derived funnel alone does not supply all projected programme wins; direct executive and channel selling must supply the balance.
Licence revenue is a provisional proxy for separate partner-delivered activity, additional to company-managed programme fees. It is not the same customer revenue counted twice. A 25% incremental delivery/review cost is charged. No partner pipeline or legal principal-versus-agent accounting decision has been established. The no-licence sensitivity is essential; replace the proxy with named partner contracts before spending against it.
Programme costs and margin reality
| Mature reference programme annual direct cost | Low | Mid | High |
|---|---|---|---|
| Fieldwork | $300,000 | $220,000 | $180,000 |
| Research lead, 0.5 FTE | $90,000 | $90,000 | $90,000 |
| External review, 10 days | $15,000 | $15,000 | $15,000 |
| Delivery engineering/operations, 0.2 FTE | $36,000 | $36,000 | $36,000 |
| Client/agency management, 0.2 FTE | $30,000 | $30,000 | $30,000 |
| Production AI | $20,000 | $15,000 | $10,000 |
| Production infrastructure | $12,000 | $12,000 | $12,000 |
| Included model-change fieldwork reserve, 10% | $30,000 | $22,000 | $18,000 |
| Total before integer staffing effects | $533,000 | $440,000 | $391,000 |
| Mature direct margin | -6.6% | 41.3% | 60.9% |
The $180,000 loaded research-lead cost and two-programme capacity come from the strategy. The $1,500 reviewer day, $180,000 delivery-engineering FTE, $150,000 client-management FTE and other allowances are explicit additions, awaiting quotes. Research leads are hired at signature, rounded up to whole people and retained across the horizon. Underutilisation and the two-month pre-service period reduce reported margins below mature unit margins. Audit researchers are separately costed contractors.
The reference scope is three audiences in two markets. The high-case cheaper fieldwork assumes volume terms, not cheaper evidence by assertion. A four-market/six-audience Enterprise programme is not covered at the same cost; its cadence and price are compared separately below. Group shared fieldwork has no assumed savings until shared estimands and rights justify them.
| ACV needed for mature direct margin | Low cost base | Mid cost base | High cost base |
|---|---|---|---|
| 55% | $1,184,444 | $977,778 | $868,889 |
| 60% | $1,332,500 | $1,100,000 | $977,500 |
| 65% | $1,522,857 | $1,257,143 | $1,117,143 |
These price requirements exclude corporate overhead and capacity rounding. A 55–65% gross margin is an aspiration, not a model input.
Audit economics: price and cost per cell
One cell is an audience × market × agreed task/instrument. Pricing hypothesis: $15,000 per cell, $75,000 minimum; ten cells cost $150,000; eighteen cost $270,000. Additional task families or scarce cohorts require a new quote. The $110,000 average in section 8 is retained as a blend of two-thirds seven-cell ($105,000) and one-third eight-cell ($120,000) audits. No individual contract contains a fractional cell.
Planning direct cost per cell: $8,000 fieldwork (240 sealed completes plus four discovery/context interviews), 20 research hours at $125, half a reviewer day at $1,500, $350 AI and $150 infrastructure: $11,750. Common audit preparation adds 40 research hours and two reviewer days: $8,000. This yields average direct cost $94,167, leaving about 14.4% before acquisition and corporate overhead at the $110,000 average. Supplier and review quotes could change this materially.
At three/five/ten/eighteen cells, direct costs are $43,250/$66,750/$125,500/$219,500. The $75,000 floor does not make all scope variations equally profitable. An eighteen-cell audit at the former $150,000 headline loses money; the stress test retains that error as a warning, not a proposal.
Full-wave cadence: cost the actual human sample
Per audience-market cell: first wave 240 sealed +160 development; later full waves 240 sealed +80 refresh-development. Foundation has quarterly full waves (1,200 completes per market initially, then 960). Twelve qualitative interviews per cell are budgeted once at onboarding. They are developmental, not part of the sealed quantitative sample.
Enterprise has six audiences × four markets. Both semiannual and quarterly full-wave options are costed, with 150-person market-level pulses in intervening months. Pulses do not validate six audience cells. Comparison assumptions: $22.50 per complete and $500 per interview, all-in fieldwork but excluding the separately costed research lead. These are unquoted allowances. Foundation first-year fieldwork computes to $219,600, rounded to $220,000 in the mid model.
| Scope/cadence and contract stage | Quant completes | Fieldwork | Total direct cost | Margin at stated price | Price for 60% |
|---|---|---|---|---|---|
| Foundation, first year | 8,160 | $219,600 | $439,560 | 26.7% | $1,098,900 |
| Foundation, later year | 7,680 | $172,800 | $388,080 | 35.3% | $970,200 |
| Enterprise semiannual 1.5m, first year | 23,280 | $667,800 | $1,183,080 | 21.1% | $2,957,700 |
| Enterprise semiannual 1.5m, later year | 21,360 | $480,600 | $977,160 | 34.9% | $2,442,900 |
| Enterprise quarterly 1.5m, first year | 37,440 | $986,400 | $1,743,540 | -16.2% | $4,358,850 |
| Enterprise quarterly 1.5m, later year | 35,520 | $799,200 | $1,537,620 | -2.5% | $3,844,050 |
Enterprise review/research staffing increases with scope: semiannual 1.25 research FTE and 25 review days; quarterly 2.25 FTE and 45 days, plus 0.4 delivery-engineering and 0.4 client-management FTE. These capacity assumptions need time measurement, not automatic acceptance. The cadence comparison and workbook sheet also show $1.2m pricing and the prices needed for 55% and 65%. The group-level 36-month forecast does not silently adopt these larger scopes; it remains a reference-programme scenario.
Engineering, commercial overhead and accounting
First-year platform build is $600,000/$750,000/$900,000, expensed over 12 months. This includes central development, security engineering and development/test model infrastructure. Per-client production AI, infrastructure and delivery engineering are separately charged in COGS; do not add supplier invoices to both pools. Security-assurance spend is external assessment/legal readiness, not another engineering team. No SOC 2 report or procurement acceptance is promised by a budget line.
Paid founder costs are $180k/$198k/$216k across years. Commercial capacity is one $180k loaded FTE per eight weighted annual wins (an audit counts half), minimum one, ahead of revenue. G&A, assurance and channel/content costs are explicit. There is no paid-media budget. Incremental executive selling/acquisition is $15k per programme and $5k per audit. $140k prelaunch covers legal/security setup and early commercial/admin work; it is charged at M1 with a prelaunch cash floor. Actual calendars are conditional, not a launch promise.
Programme revenue is recognised monthly over the active annual service term. Initial quarterly invoice is issued at signature; later quarters are invoiced at service month +3/+6/+9 and at renewal. Receipts lag invoices by two months. Audit invoices are 50% at signature/50% on delivery; audit revenue is recognised on delivery after two months. No full ACV is recognised on signature.
Fieldwork is paid one month before each wave. The first wave takes 41% of the initial annual budget, reflecting 2,400 completes and 72 onboarding interviews; the remaining 59% is divided across three refresh waves. Later years pay 25% per quarter. Fieldwork is expensed monthly over the service year, with the prepaid balance reconciled. This avoids understating the initial wave by spreading its cash evenly across quarters. Audit costs are paid half at signature and half next month, held as WIP until delivery. Other costs are paid as expensed. The model reconciles cash = cumulative EBITDA − AR − fieldwork prepayments − audit WIP + deferred revenue. No bad debt, tax, interest, financing or capitalised development is included. Remaining commitments and balances survive M36.
Six-operating-month anchor pilot
One $600,000 annual Foundation anchor signs at M1 and starts service M3. Two seven-cell audits at $105,000 each sign M3/M4 and deliver M5/M6. These are scoped pilot assumptions; no customer commitments exist.
| Pilot measure | USD |
|---|---|
| First-year contracted value | $810,000 |
| Revenue recognised within six operating months | $410,000 |
| Cash expenditure including prelaunch | $1,287,300 |
| Cash collected | $255,000 |
| Funding with 20% contingency | $1,238,760 |
| Closing receivables | $255,000 |
| Closing deferred revenue | $100,000 |
| Closing prepaid fieldwork | $60,133 |
Six-month pilot budget. It includes half the $750k first-year central build budget, a whole research lead, commercial capacity and enterprise setup. It is not a minimum-cost concierge experiment. Two full waves across two markets require 4,320 programme completes; two seven-cell audits add 3,360, for 7,680 total quantitative completes across separate engagements. No pooling of their validity claims. The second programme wave lands at M6; an annual renewal cannot be observed then. Measure a continuation decision and scheduled anniversary renewal separately.
Sensitivities and funding limits
| Mid-case stress | Y3 revenue | Y3 EBITDA | Funding +20% |
|---|---|---|---|
| base | $24,092,578 | $6,692,926 | $2,892,244 |
| prices minus 20pct | $19,274,062 | $1,977,922 | $5,379,791 |
| sales half | $12,046,289 | $2,340,963 | $3,264,417 |
| sales delay 6m | $15,261,812 | $3,229,245 | $3,852,969 |
| no licence | $22,022,344 | $5,140,250 | $3,023,400 |
| net90 | $24,092,578 | $6,692,926 | $3,793,191 |
| model change reserve 30pct | $24,092,578 | $5,478,388 | $3,170,544 |
| fieldwork plus 50pct | $24,092,578 | $3,352,948 | $5,123,297 |
| AI 5x | $24,092,578 | $5,036,738 | $3,271,744 |
| churn double capped | $22,780,312 | $6,147,203 | $2,898,413 |
| four market cost stress | $24,092,578 | $-3,296,077 | $16,575,929 |
| expanded audit 18 cells at 150k | $24,572,578 | $5,668,926 | $3,783,444 |
The six-month delayed-sales case keeps central spending underway and is deliberately severe. The four-market cost stress doubles fieldwork and research/reviewer capacity but does not substitute for the fully scoped cadence comparison. No-licence and slower-cash cases test two especially uncertain assumptions.
Editing and verification
Edit assumptions.json, then run from the repository root:
python3 -m pip install -r finance/requirements.txt
python3 finance/generate_model.py
python3 finance/build_narrative.py
python3 finance/test_model.py
The Excel workbook has editable blue inputs and monthly signing cells, formulas and cached values, including an editable cadence comparison. Signing totals are edited on monthly sheets; Python schedules derive from JSON annual totals. Changing one format does not update the other.
The eight test families retain cohort timing, cash identities, zero-sales/delay, same-month receipts/invalid inputs, USD/direct costs, sensitivity/tail, cached formulas and independent workbook recalculation. They were adapted to annual enterprise contracts, not removed to preserve a test count. No product/security/research validity test has been run by these finance checks.
Data: annual summary, monthly model, sensitivities, sales funnel, structured results.