The pixel no longer sees everything
Browsers restrict it and most iOS users opt out of tracking. The conversions still happen; the ad platforms simply stop receiving them, and optimise on an incomplete picture.
20-40%
The only growth tool your brand would ever need.
Capture every conversion server-side, attribute it once, and feed it back to Meta and Google so their targeting and retargeting improve. One workspace. Your data.
Instrumentation health
Norvelle
| Env | Events | Last heartbeat | Helper |
|---|---|---|---|
| prod | 1.84M | 14s ago | v1.4.2 |
| stage | 42.1K | 6h ago | v1.3.9 |
| test | 8.9K | 52s ago | v1.4.2 |
Illustrative data
Why now
Your numbers stopped agreeing with each other.
Privacy changes didn't make growth harder to drive. They made it harder to see - and every tool you added to see it reports its own version.
Three tools, three ROAS numbers
Ads Manager, Google, and your attribution tool each report on their own terms. None of them is wrong, none of them reconciles, and budget decisions turn into a debate over which number to trust.
75%
The spreadsheet in the middle
Every week someone exports, reconciles, and re-keys the numbers by hand. At most brands that spreadsheet is the real source of truth - and it depends on one person.
~66%
How it works
One spine, from first click to the ad network and back.
Events are collected server-side, checked against a taxonomy you control, and stored in your own database. Attribution, dashboards, and agents all read from the same place - and conversions go back out to Meta and Google through a filter that knows what's sensitive.
Your apps
Web, mobile, and server code - per app, per environment.
SDK + JS helper
track · identify · alias, with UTMs, gclid and fbclid captured client-side.
Taxonomy gate
Validated against Active definitions. Unknown events blocked in prod.
Your event store
Append-only, environment-scoped, in a database that is yours.
Attribution
Touchpoints, identity stitching, five models side by side.
Decisions
Dashboards, funnels, and agent proposals you approve.
Egress filterAllowlisted, hashed identifiers only. Anything classified PHI has no path out.
Meta Conversions APIGoogle Enhanced Conversions
What makes it different
Agents are only as good as what they stand on.
Four foundations, built first. The fifth is what they make possible.
01
Own the spine
A dedicated workspace with a dedicated database. Your data, your schema, your deletion - not a row in someone else's table.
Learn more 02One number, not three
Deduplicated server-side conversions and five attribution models over one touchpoint table. The spread between models is the insight.
Learn more 03Governed, not guessed
A typed, versioned, sensitivity-classified taxonomy, and a health panel that proves the data is actually arriving.
Learn more 04Compliant by construction
Classification-gated egress means PHI can't leave. Per-environment retention. Person-level deletion as a primitive.
Learn more05
Meet the agents Agents you can audit
Agents for instrumentation, signal health, and media buying - acting only through guarded, logged paths, on data your brand owns.
In the product
The product, as your team uses it.
Instrumentation health
Instrumentation you can verify.
Every environment reports its own heartbeat, helper version, and origin match. Whether tracking is working is a fact on the screen, not an assumption.
Learn moreInstrumentation health
Norvelle
| Env | Events | Last heartbeat | Helper |
|---|---|---|---|
| prod | 1.84M | 14s ago | v1.4.2 |
| stage | 42.1K | 6h ago | v1.3.9 |
| test | 8.9K | 52s ago | v1.4.2 |
Illustrative data
Attribution
No model is the truth. The disagreement is the signal.
Five models over the same conversions. When first touch says scale and last touch says pause, the difference is the prospecting question - and now it can be answered with evidence.
Learn moreAttribution · Meta
$164,200 · last 30 days
| Model | Conversions | CPA | Reads as |
|---|---|---|---|
| First touch | 2,451 | $67 | Scale |
| Linear | 2,219 | $74 | Scale |
| Position based | 2,105 | $78 | Hold |
| Time decay | 2,027 | $81 | Hold |
| Last touch | 1,866 | $88 | Pause |
Same spend, same conversions, five ways of dividing the credit. The $21 spread between first and last touch is the prospecting question - not a number to average away.
Illustrative data
Conversion forwarding
More signal to Meta and Google. Nothing that should not leave.
Deduplicated conversions go back to the ad networks server-side. Every property carries a sensitivity classification, and the egress filter enforces it on every payload.
Learn moreEgress filter · purchase
→ Meta CAPI
- order_idforwarded
- order_valueforwarded
- emailPIIhashed
- gclidforwarded
- condition_codePHIblocked
A property classified PHI is not forwardable. Not by policy, not by a reviewer remembering - the filter has no path that lets it leave.
Illustrative data
Agents
An agent is only as good as the measurement under it.
Quantiv's agents read the spine your brand owns, propose what to do, and act only through paths that are guarded and logged. You decide how much latitude they have, and you can withdraw it at any time.
Proposal · media buying agent
awaiting review
Meta · Prospecting · Skincare US
Increase daily budget $2,400 → $3,000
First-touch and linear both put this set under blended target while last touch puts it over. The spread is prospecting credit, and the set has held CPA for 11 days.
- First touch CPA
- $67 → $71 projected
- Last touch CPA
- $88 → $92 projected
Within the $3,500/day cap · 1 of 2 daily changes used
ApproveRejectEdit
proposal #4182 · agent · awaiting review · nothing applied
Illustrative data
Design partners, under NDA.
- Series B consumer app
- DTC skincare
- Health-adjacent subscription
- Marketplace
48% → 72%
+15-30%
15-25%
Integrations
Plugs into the stack you already run.
Meta
Ads sync · Conversions API
Google Ads
Ads sync · Enhanced Conversions
Python SDK
track · identify · alias
JS helper
Click context · self-installing
Your BI
Direct database access
Questions
Common questions.
Isn't this a six-month data project?
No. The Python SDK installs in a few lines and injects the JS helper into server-rendered pages on its own; single-page apps get a snippet. The health panel tells you per environment when it's working. Most of the first two weeks is connecting Meta and Google and agreeing your event names.
We already have an agency.
Your agency stays. Operators receive free seats inside your workspace and work from the same data you see, with every change recorded in the audit log. The weekly reporting they assemble by hand is already on the dashboard.
Why would I let an agent touch my budget?
You set its authority. Every change begins as a proposal with the supporting attribution attached, and passes through the same guarded path as a human change: spend caps, per-action approval, and a complete audit log. Most teams begin with the agent as an assistant and extend its remit as confidence builds.
We handle health data. Can we use this at all?
Quantiv was designed with regulated data in mind. Every property is classified as none, PII, or PHI, and the egress filter will not forward anything classified PHI to an ad network. Retention is set per environment, and person-level deletion is a single operation.
See it on your own numbers.
A two-week pilot runs your conversions through Quantiv alongside what you report today. The difference between the two is the business case.