The 2026 Annual Letter

Dear shareholders, customers and team —

Two years ago we set out to build the financial layer above every major ad platform. One payment in, every platform funded, every euro reconciled. In 2026 we crossed the threshold from useful tool to mission-critical infrastructure for the customers who depend on us. This letter is an honest accounting of how that happened, what it cost, and where we go next.

1. The year in one paragraph

We grew annualised revenue 3.4× to $11.2M, kept gross retention at 98%, moved $418M of advertising spend without a single lost transaction, and shipped 142 customer-visible improvements. We hired 18 people and lost 2. We were wrong about two strategic bets and right about three. Our NPS rose from 41 to 67. We remained profitable on a cash basis from Q2 onwards.

2. What we shipped

The Treasury reached general availability. Pacing accuracy across the customer base improved from a median 9.1% MAPE to 2.1% — meaningfully ahead of what platform-native pacing delivers. Autopilot went from labs to default-on for Growth and Enterprise customers, with the human override switch we promised: one toggle, twelve seconds to freeze every AI action across every platform.

On the boring-but-vital list: SOC 2 Type II renewed without findings, ISO 27001 added, EU AI Act self-assessment published, a CAIQ that procurement teams actually accept without follow-up. We moved primary banking to a second corridor so a single bank outage no longer disrupts customer top-ups.

3. What we got wrong

Spotify-first international rollout. We assumed audio ad spend would be a wedge in DACH and the Nordics. It wasn't. The wedge is rebate utilisation on Meta and Google. We pivoted in Q3 and recovered the quarter, but left two engineering quarters on the table.

Self-serve onboarding for enterprises. We thought the implementation runbook plus the academy could replace solutions engineering for $5M+ accounts. It cannot. We hired four solutions engineers in Q4 and added a fixed-fee implementation tier — see /implementation.

Mobile dashboard. We over-invested in a native mobile app that customers told us they did not want. CFOs review on desktop, react in Slack. We sunset the native build and replaced it with a polished PWA in six weeks.

4. What we are betting on next

Treasury becomes a balance sheet. In 2027 we are extending credit lines on platform-vetted run-rate, with Net 30 and Net 45 terms that customers can use to smooth seasonal swings without diluting equity. This is a regulated step; we have spent eighteen months getting the licensing in place.

AI that negotiates. Our pacing models already know when a platform is over-priced versus your historical baseline. We are wiring those signals into automated reallocation across platforms, with a human-approved guardrail. Customers in the pilot have recovered 11.4% of media cost without reducing volume.

The procurement-grade default. By the end of 2027 we want ADmetric AI to be the answer when a CFO asks "what's the safe pick?". That means deeper independent assurance, more public benchmarks, and a willingness to publish what isn't working.

5. To our customers

Thank you for letting us hold money that does not belong to us. We do not take that lightly. Every postmortem, every policy and every release note is written with one question in mind — would we be proud to show this to the customer whose money it touched? If you ever feel we have not lived up to that, write to me directly at ceo@admetricai.app.

Onwards,
The CEO & founding team