Your own data, turned into an audience you can actually buy
If your file has a geographic component — latitude and longitude, postal code, street address, county, or city — we can onboard it. That single field is enough to build a modeled look-alike audience and a live intent audience without cookies, device IDs, or an identity graph.
No match rates to negotiate. No per-record onboarding fees. No PII leaving your control in an activatable form.
What counts as usable data
Anything with a place attached. If one of the fields below exists somewhere in the file, the build can start. Everything else in the file is optional context, not a requirement.
| Field | Example | How we use it |
|---|---|---|
| Latitude / longitude | 38.9072, -77.0369 | Best case. Used directly, no geocoding step. |
| Postal / ZIP code | 20001, 20001-3456 | Maps cleanly to geography. Five-digit or full ZIP+4 both work. |
| Street address | 1600 Pennsylvania Ave NW, Washington DC | We geocode to a coordinate, then aggregate. Raw address is never activated. |
| City / town | Kansas City, MO | Coarser, still usable as a modeling input when paired with other fields. |
| County / FIPS | Jackson County, MO / 29095 | Standard geography, joins directly to census and behavioral layers. |
| State / region | MO | Useful as a constraint or weighting layer, not as a standalone seed. |
| Store / venue ID with location | Store 4471 — 39.10, -94.58 | Retail, clinic, branch and event location files are all valid seeds. |
| IP-free, ID-free records | No device IDs required | We do not need MAIDs, cookies, emails or hashed identifiers to build. |
The onboarding pipeline
First-party data loses value in the handoff — mismatched formats, duplicate records, sloppy handling. The pipeline is built to preserve signal end to end and to be re-run every time you add another file.
Ingest and normalize
Files arrive in whatever layout you already have — CSV, XLSX, a CRM export, one file per year. Every record is parsed and standardized into one consistent structure so rows line up across sources and time periods.
Resolve geography
Addresses are geocoded to coordinates. Postal codes, cities and counties are mapped to standard geographies. Every record ends up with a geographic anchor, which is the only thing the build actually requires.
Consolidate and de-duplicate
Records merge across sources and cohorts, repeat entries collapse. This cleans the file and simultaneously surfaces your highest-value segment: the people who show up more than once.
Aggregate to a privacy-safe seed
Individual rows roll up into geographic concentrations. Nothing person-level leaves the pipeline. The seed is a map of where your known audience actually is, not a list of who they are.
Model look-alikes
The seed is scored against census, behavioral and commercial layers to find the places that resemble your proven audience but are not yet in your records. That becomes your net-new reach audience.
Layer search listening
A custom search-listening audience is built around the topics and intent signals tied to your category, so the historical model is paired with people who are in-market right now.
Activate
Audiences are delivered into the DSP, CTV, DOOH or social platform holding the buy. No onboarding fees per record, no identity spine to maintain.
What gets built from the seed
Two audiences that answer two different questions, both grounded in the same first-party seed.
Look-alike audience
"Who else looks like the people we already have?" The seed models the defining traits of your proven audience and finds the places that share them but are not yet in your records.
The profile is learned from real behavior — who actually showed up, bought, or joined — instead of assumed from broad demographics.
Search-listening custom audience
"Who is showing intent right now?" A custom cluster built around the queries and topics tied to your category and moment, refreshed continuously.
Look-alikes extend reach to people who resemble proven customers. Search listening catches people signaling intent in the moment. You need both.
Turning three years of event data into high-precision audiences
The short version
A national association handed us three consecutive years of its own conference attendee data. Instead of letting that data sit in a CRM as a mailing list, we're onboarding it as the seed for a modeled targeting strategy — using verified past attendees to find look-alike prospects and to build a real-time, intent-based audience for the association's next flagship conference.
The result isn't a one-off campaign. It's a repeatable audience pipeline that gets sharper every year the client adds another year of data.
The problem most event marketers have
If you run a recurring conference, you already own the single most valuable targeting asset there is: the list of people who actually registered and attended. It tells you exactly who your real audience is — their roles, their organizations, where they came from — and the fact that they committed real budget and time to be there.
Almost nobody activates it. That data usually goes into an email tool and stops. So every new campaign cycle starts over from generic demographics and guesswork, ignoring the proof sitting in the client's own records.
This engagement was about fixing that: converting the client's own attendee history into audiences we can actually target against.
How the data drives targeting
The three-year attendee seed models look-alike audiences: prospects who share the defining traits of people who have genuinely attended, but who aren't yet in the client's own records. The profile is learned directly from real attendee behavior — role, organization type, professional context — instead of assumed from broad demographics.
Alongside the modeled look-alikes, the data informs a search-listening custom audience built around the intent signals and topics tied to the association's field and event. That captures people who are actively in-market right now — a live signal that complements the historical attendee model.
Where it stands
The first-party data has been received and is actively being ingested to build the look-alike and search-listening audiences ahead of the association's 2027 flagship conference. This is a live onboarding, engineered to keep enriching as each new year of attendee data is added.
- Year one — attendee registration list
- Year two — registrant listing
- Year three — final registration report
Three consecutive cohorts of verified, opted-in attendees, shared deliberately for activation. Because the files span consecutive years, they capture both the loyal core who return every year and the natural churn at the edges — which is what makes the data a strong modeling seed rather than just a contact list.
- Your own data outperforms anything off the shelf. It's a direct record of the behavior you're trying to reproduce, not a proxy for it.
- History compounds. One year is a list; three years is a pattern.
- Reach and timing from one foundation. Look-alike modeling scales reach; search listening adds in-market timing.
- It's a defined, repeatable process. Ingestion, de-duplication, seeding and audience build follow a documented workflow.
Anonymized case study. Client and partner identities generalized; methodology and status accurate as of preparation.
Files we onboard
Event data is one example. Any of these work the same way.
Event and conference attendees
Registration and attendance files across multiple years.
Customer and transaction records
Purchase history with a billing or shipping location.
Membership and subscriber rolls
Association members, subscribers, loyalty programs.
Store and clinic visit logs
Any file where a location and a date exist.
Lead and CRM exports
Form fills, inquiries, quote requests with an address field.
Distributor and dealer networks
Territory files used to shape and constrain delivery.
Nothing person-level is ever activated
Your file is used to locate concentration, not to target individuals. Records are aggregated to geography before anything is modeled or delivered.
No cookies, no device IDs, no identity graph, no hashed-email match. Nothing to maintain and nothing to deprecate later.
The activated audience contains no PII, which keeps the build workable under GDPR, CCPA and holding-company data governance review.
Send us the file you already have
Attendee list, customer export, membership roll — if it has a location column, we can tell you what it will build before you commit to anything.