Screening: Accessibility Drift (a11y-drift) — S0A
Prepared 2026-09-18. Claim tsk_20260918031508_8ht5m. Cap: $25 research/7 days (not approached — no new paid API calls made this pass; see cost note below).
Sources read: research/keyword-research-2026-09-07/REPORT.md (dated 7 Sept 2026, API data retrieved 8 Sept 2026 UTC), ventures/a11y-drift/brief.md, and a fresh cached Google Ads keyword pull already present at ventures/a11y-drift/research/google-ads/6bd7ca3c214786a3711f1312bf33d1886ffcf82d483b61734496e750738fc017.json (retrieved 2026-09-18T03:44:43Z UTC, US/English, Sept 2025–Aug 2026 window, 35 keywords, no additional cost incurred by this task — the file was already cached before this claim started). ventures/a11y-drift/metrics.json confirms zero behavioral evidence exists: 0 qualified exposures, 0 signups, 0 paying customers. This is a pre-S1 desk screen, not a measured-demand result.
Demand
Buyer-language and intent (US, English, Google Search, monthly averages, Sept 2025–Aug 2026, from the fresh cache):
| Keyword | Avg US searches/mo | Trend note |
|---|---|---|
| vpat | 5,400 | Broad; includes templates/services/vendor docs, not our product |
| vpat services | 720 | Spiked to 2,400 in Dec 2025, near-zero (50) by Mar–May 2026 — volatile |
| vpat template | 480 | Peaked 1,600 in Sept 2025 |
| accessibility audit | 720 | Peaked 1,300 in Mar 2026 |
| accessibility conformance report | 480 | Direct match to our proposed artifact name; spikes to 1,300 in Sept 2025 and Jan 2026 (likely fiscal-year/procurement-cycle timing) |
| axe devtools | 1,300 | Rising (880→1,900 over the window); brand/tool search, not a buyer of us |
| axe core | 720 | Rising; developer research, not commercial intent |
| european accessibility act | 390 | Declining (880→210 across the window) — awareness cooling, not building |
| en 301 549 | 260 | Flat |
| accessibility monitoring | 170 | Our closest commercial-intent term |
| accessibility remediation | 70 | High CPC ($67 avg) — service buyers, not tool buyers |
| accessibility dashboard | 70 | Low competition |
| accessibility statement generator | 110 | Adjacent, not our product |
| accessibility questionnaire | 10 | The named trigger event in the brief — smallest number on the page |
| accessibility procurement | 10 | Also the named trigger — negligible search volume |
| wcag compliance software | 10 | Negligible |
| website accessibility monitoring | 30 | Negligible |
| accessibility github action, accessibility regression testing, continuous accessibility testing, accessibility reporting tool, accessibility compliance monitoring, saas accessibility | no data | Zero measurable direct search demand for the CI/PR-diff framing at all |
This fresh pull confirms and slightly extends the 7 Sept report rather than changing its conclusion: it adds "accessibility conformance report" (480/mo, the one term that names our actual deliverable) and "accessibility remediation" (70/mo, high CPC), neither of which was in the original 9-keyword commercial table. Both are still small next to vpat (5,400) and axe devtools (1,300), and both terms are ambiguous — "conformance report" volume plausibly includes people searching for VPAT/EN 301 549 templates to fill in themselves, not a subscription trend service. The specific trigger named in the brief — "we got an accessibility questionnaire and don't have answers" — has no measurable search volume (10/mo, both accessibility questionnaire and accessibility procurement). People experiencing that moment are not searching Google for a generic product category; they need someone found some other way (SEO to a live procurement moment doesn't work if there's no search volume to catch).
Geography: only US, UK and Germany were sampled (7 Sept report), all in English. German-language search and other EU markets (which is most of the actual ICP — "EU-facing SaaS") were not measured. UK/Germany volumes for the commercial terms were smaller than the US in the earlier pull (e.g. accessibility monitoring: 170 US / 20 UK / 10 Germany). This is a meaningful gap: the ICP is explicitly EU-based buyers, and the best data we have is for a different market.
Market and acquisition
- TAM/niche/reachable is still not enumerated. No count exists of "B2B SaaS,
20–80 people, selling into EU enterprise/public sector" companies, nor of how many receive an accessibility question from a customer in a given 90-day window. This is the single largest missing input and it is required, not optional, per the brief's own "what would kill it" section.
- Google/English search cannot see this buyer population directly: the ICP is a
narrow B2B segment triggered by an internal procurement event, which is not reliably expressed as a public search query (evidenced by the near-zero volume on the two on-the-nose trigger terms above).
- Funnel and CAC estimates below are **illustrative sensitivity modeling
reused from the 7 Sept report's analysis.json**, anchored on a single representative term (accessibility monitoring, ~200/mo combined US+UK+DE), not a validated buyer count. They should not be read as a market-size claim.
- Competitive landscape unchanged from the brief: Deque axe DevTools Pro
($79–99/user/mo, $15–50K/yr contracts) and AccessProof ($0–199/mo) occupy the gap; W3C states automated tools cannot certify conformance (https://www.w3.org/WAI/test-evaluate/, cited 2026-09-04 in brief.md).
Durability
Unknown and untested. The 7 Sept report frames the central durability risk correctly: a buyer who needs one VPAT/questionnaire answered by Friday may not need a recurring subscription once the deal closes. Nothing in this pass adds evidence either way — there is no observed retention, no cohort data, and the brief's own S1 plan (test whether ongoing drift-tracking is valued after the initial ask) has not run yet. The declining european accessibility act search trend (880→210 over 12 months) is a mild negative signal against a growing, self-sustaining wave of procurement pressure, though it is an awareness metric, not a procurement-event count.
Economics (illustrative, not measured; reused from the 7 Sept sensitivity model)
Price assumption: $350/mo ($4,200/yr) — proposed in the brief, not approved by Yuval for display, per the SCREENING.md and charter constraints on pricing authority.
| Scenario | New customers/yr (anchor-based) | CAC incl. labor | Annual revenue/customer | Annual service cost/customer | Lifetime revenue/customer | Lifetime service cost/customer | Lifetime contribution after CAC |
|---|---|---|---|---|---|---|---|
| Conservative | 0.01 | $8,200 | $4,200 | $1,800 | $2,100 (6mo life) | $900 | -$7,000 |
| Base | 0.30 | $1,350 | $4,200 | $900 | $6,300 (18mo life) | $1,350 | +$3,600 |
| Optimistic | 3.60 | $350 | $4,200 | $420 | $10,500 (30mo life) | $1,050 | +$9,100 |
Base and optimistic scenarios are contribution-positive per customer; conservative is sharply negative. The 7 Sept report calculates that reaching a $50K/yr contribution benchmark at base assumptions needs ~21 active customers against ~9,334 monthly search events of equivalent quality — roughly 47x the sampled anchor term's actual volume (200/mo). No enumerated buyer population exists to say whether 21 EU procurement-triggered customers/year is a large or small share of the reachable market, because the reachable market itself is unknown.
Decision: REVISE
The riskiest, most falsifiable assumption — "an EU SaaS company facing an accessibility procurement question will pay $350/mo for a CI/PR drift tool" — remains untested by any customer behavior (metrics.json: 0 exposures, 0 signups, 0 payments). This screen adds one useful, mildly discouraging data point: the exact trigger phrase named in the brief (accessibility questionnaire, accessibility procurement) has almost no search volume, which means the acquisition channel proposed in the brief (SEO/content aimed at that moment) has little traffic to capture even if the offer is right. The nearer adjacent term that did show up, accessibility conformance report (480/mo, US-only), is closer to a VPAT/audit search than a subscription-trend search and should not be counted as validated demand for this product per the brief's own guidance.
Do not proceed to S1 spend on this evidence. The two blocking unknowns are:
- No enumeration of the reachable buyer population (EU SaaS 20–80 people with
a recent procurement accessibility ask) — required before computing a plausible required market share.
- No behavioral evidence that the named trigger event converts to willingness
to pay $350/mo, versus a one-time need satisfied by a free template or a one-off consultant.
What kills the idea if confirmed: that buyers who hit an accessibility questionnaire want a completed answer *this week*, not a recurring PR-diff dashboard — i.e., the offer and the trigger are still misaligned, exactly as flagged in brief.md's 2026-09-04 correction.
Cheapest next test (for a next-test task, not run here): a manually prepared example conformance record/VPAT answer, shown to a small number of real target-persona contacts (e.g. accessibility/eng leads at 20–80 person EU-facing SaaS) sourced from cssbed's existing audience or similar communities, asking whether they would pay for an ongoing trend service versus a one-time document — without building the GitHub App first. This tests the trigger/offer alignment risk directly and cheaply, per the brief's own suggestion.
No project stage, price, or spend was changed by this screening. This file and screening.json are evidence for the Steward review, not authority to advance.