Technology
The signal starts on your domain.
From signal to spend, on servers we run.
Someone clicks an ad on a Monday and buys on the Tuesday of the following week, on a browser that has already cleared the visit. The platform that showed the ad never connects the two. If the person agreed to it, our server does. We collect the event under your own domain, check the consent that travels with it, count it once and send it server to server to the platforms you choose.
Our own server-side, first-party tracking
Browsers limit how long scripts can store data, strip identifiers from links and block known trackers. Apps ask permission before they track. People make consent choices. Each of these is legitimate, and each one thins the signal ad platforms use to measure and optimize. So we move collection to an endpoint under your own domain, and an event leaves only after it has been checked.
- 01 Action
A visit, a form, a purchase or a lead that changes stage, on the site, in the app, in a store or in the CRM.
- 02 Your domain
The browser or app sends one event to an address on your own domain, instead of one to every vendor.
- 03 Consent
The event carries the person's consent state, and the server decides what each destination may receive.
- 04 Clean and enrich
The schema is checked, identifiers are normalized and hashed where a destination asks for it, and the true order value comes from the order system.
- 05 Count once
When the browser reports the same action, a shared event ID makes sure it is counted once.
- 06 Deliver and log
Events go server to server to the ad platforms, analytics tools and data warehouse you choose. Every delivery is logged and retried only on temporary errors.
A first-party domain that is really yours
Collection runs on a path of your own site, or on a subdomain served from the same infrastructure as the site, so the browser treats it as your own. A subdomain that quietly forwards to someone else's server only looks first-party. Browsers increasingly recognize the pattern and cut what it can store, so we don't build it that way.
Every channel in the same event model
Web and app events, offline sales, CRM stages and store data land in the same event model, with a shared schema, across digital and traditional channels. Every change to the setup has a version, an author and a way back, so a broken tag is found and undone instead of argued about at month end.
The loop from collection to spend
Signal, decisions and spend run on the same data. What we collect feeds measurement, measurement feeds the plan, the plan turns into buys and payments, and what those produce comes back to be measured.
Signal
- 01CollectEvents from the site, the app, servers, the CRM, stores and offline files, under your domain.
- 02ConsentConsent state and opt-out signals attached to every event.
- 03IdentityIdentifiers normalized and protected, with exact matches and estimates labeled apart.
- 04ValidateSchema, deduplication and quality checks catch errors before they reach a platform or a report.
- 05StoreEvents, spend and results kept together, so any number can be queried and checked.
- 06DeliverThe right data to each platform, analytics tool and clean room, with every delivery logged.
Decisions
- 07MeasureAttribution, mix models and experiments on the same data, each with its limits stated.
- 08PlanForecasts and scenarios that come with a range.
- 09OrchestrateAgents work inside the limits and approvals people set.
Spend
- 10SpendBudgets, pacing, payments to the platforms and reconciliation.
Every source leaves a different trace
A click on a search ad, an ad in a streaming show, a radio spot and a coupon at the shelf leave different traces, and we collect each one the way it can be collected. Clicks and conversions arrive as server-side events. Ads in streams are read from the logs of the servers that insert them. App installs come through the aggregate attribution of the operating system where a person hasn't agreed to tracking, and through measurement SDKs with consent where they have. TV, radio and out of home are read from the audience measurement each medium trades on and the delivery data sellers provide, and print and coupons from dedicated codes and redemptions. Counts and estimates are labeled as such. How each family is measured, with clicks or without, is on Measurement, and every channel is on Channels.
What the AI gets from it
Platforms bid with their own models, and those models learn only from the events they can match. The event a campaign optimizes toward is the event its auction learns from, so it goes out server to server with its true value, and qualified leads follow from the CRM. Our forecasts and agents run on the same data. Events that are consented, counted once and carry true values are what make both worth listening to. How we use AI · How we buy digital
Questions we get
Does server-side tracking replace consent?
No. It changes where data is collected, not whether a person agreed. Consent is checked before any event is forwarded, and an opt-out preference signal applies to every destination.
Is a hashed email still personal data?
Yes. Hashing turns the email into a fixed code before it is shared, which protects the value in transit. The same email always gives the same code, and that is how a platform finds the match, so the code still points to one person and we treat it as personal data.
What about sales that close offline?
Signed contracts, store purchases and leads that become customers can be sent back from the CRM or the point of sale to the platforms that accept offline conversions, with the same consent checks as web events.
Does it slow the site down?
Collection happens after the page has loaded, and the conversion goes out once the server has confirmed the action. The person on the page never waits for a platform to answer.
See what your current setup is missing.
Tell us where your conversions are sent today, and we'll look at what gets lost, counted twice or sent without the consent it needs.