FAQ
Straight answers.
The questions an analytics lead actually asks before letting a new tool near the warehouse. If yours isn't here, ask us.
Is Nomos a dashboard?
No. Nomos publishes governed tables to your warehouse. You keep the BI tool you already use — Looker, Tableau, Power BI — and it reads clean, pinned data instead of a moving target.
Does it replace our BI?
No. It sits underneath it. Nomos makes the data your BI reads deterministic and defensible; your dashboards and decks stay where they are.
Does it replace our connector?
No. Think of Nomos as sitting under your connector output: it takes platform data, resolves it into one multi-client model, pins it in time, and writes defensible tables. It makes what you already pull trustworthy rather than ripping out your stack.
What does Nomos actually do?
Six things to every number, in order: Connect — OAuth ingestion from your ad platforms on a schedule you control. Classify — clients, campaigns, and markets resolved into one model across platforms. Pin — point-in-time locks so a dataset reads identically whenever you query it. Reconcile — platform delivery, ad-server cost, and vendor invoices matched against pinned numbers. Govern — your reporting standards evaluated continuously against live data. Publish — governed tables written to Snowflake, BigQuery, or Redshift. See how it works and the govern model.
What does “governance” actually mean here?
Nomos runs a three-step model on every dataset. Policy — you author the rules your data must obey: how clients and campaigns are named, how spend is counted, what a valid report looks like. Posture — Nomos evaluates your live paid-media data against that policy on every run and shows exactly where it holds and where it has drifted. Action plan — where reality breaks policy, you get a specific, ordered list of what to fix. Not a red cell in a dashboard nobody owns — a number that's defensible before anyone, or any agent, acts on it.
What does “defensible by construction” actually mean?
It means a number holds up under challenge because of how the layer is built, not because someone re-checked it. Nomos pins a dataset as-at a reporting date — a point-in-time lock that freezes exactly what a report saw when it ran, independent of later upstream revisions (like Meta's 7-day spend window). So the Tuesday report and the Thursday client call read the same as-at snapshot every time, and every figure traces back to the source event behind it.
Which platforms and warehouses do you support?
Core today: Google Ads and Meta in, and Snowflake, BigQuery, or Redshift out. Expanding: TikTok, LinkedIn, Microsoft Ads, DV360 and more, added continuously — the platform page has the current list. If you run something we haven't connected yet, tell us; connector priority follows real operations.
What does this actually save my team?
Do the math on your own operation — we won't invent a case study for you. Count the analyst hours spent re-pulling and re-reconciling numbers each week, the month-end days making finance and the platforms agree, and the client calls spent explaining why Tuesday's figure moved. One analyst day a week is roughly 400 hours a year. Nomos removes that work by construction: pinned datasets don't move, reconciliation runs against the same pinned source, and drift gets an ordered fix list instead of a hunt.
How does pricing work?
A percentage of your annual media under management, with marginal brackets so crossing a threshold doesn't spike your rate — calibrated to the connector and data-infrastructure line items you already carry. Details on the pricing page.
What data do you access, and do you ever resell it?
Read-only OAuth scopes on your ad platforms, with governed tables written into a warehouse you own. And no — no data resale, ever: no benchmarks, no syndicated datasets. More on the security page.
How do we know the numbers are right?
Proof by mechanism, not logos. A dataset pinned as-at a reporting date reads identically on Tuesday and Thursday — not because we assert it, but because the point-in-time lock is how the layer is built. Every figure walks back to its source event through full lineage: spend ← served impressions ← the contract version that governed the run. You can trace any number in a governed table back to the raw platform row it came from. That's the guarantee — auditable by construction, not by promise.