Most First-Party Data Strategies Collect 40 Fields and Use 4

A first-party data strategy is a plan for collecting customer data through channels you own (checkout, forms, email, CRM), storing it under one customer ID, and sending it back to ad platforms and measurement tools. For most marketing teams, four fields do the bulk of the work: email, phone, a stable customer ID, and the ad click ID. Everything else is optional until those four are clean.

Most first-party data content you'll find is written by someone selling a customer data platform, so it tends to end with "unify every touchpoint" and a demo link. I think that framing gets the order backwards. Teams buy the storage layer first, collect forty fields because the tool has room for forty, and then quietly use four of them. The point of first-party data, at least for a paid media or growth team, is narrower and a lot more boring: feed the ad platforms better signal, stop paying to re-acquire your own customers, and give yourself a backend number to measure against.

This piece is part of our measurement series. If you've already read why every platform says your ROAS is great, you know the backend sum test depends on having clean order data you control. And the guide to running a free geo holdout assumes you can pull revenue by region from your own systems. This is the layer underneath both of those.

Four fields, and the boring reason they win

Ad platforms match your data to their users with a short list of identifiers. Meta's Event Match Quality score, the 0 to 10 grade it gives each conversion event, leans hardest on a few of them. Usercentrics' breakdown of how to improve Meta Event Match Quality names email, phone, and external_id (your own customer ID) as the strongest identifiers, with name and location adding confidence on top. Google's Customer Match works the same way, with email and phone as the primary keys.

So the collection priority list looks something like this:

1. Email. The single strongest match key on every major platform. Normalize it (lowercase, trimmed) before you hash anything.
2. Phone number. The second key, and the one most checkout flows make optional for no good reason. Store it in E.164 format (+15551234567), because formatting mismatches are one of the most common reasons records fail to match.
3. A stable customer ID. Whatever your store or CRM assigns. This is what lets you stitch a web event, an order, and a refund to the same person without guessing.
4. The click ID. gclid for Google, fbclid (which becomes the fbc parameter) for Meta. Capture it from the landing URL, keep it on the lead or order record, and you can tie offline outcomes back to the ad that started them.

Two things that aren't identifiers but probably belong on the same record: order value with margin (so you can optimize to profit later, not just revenue) and consent status. The second one is less exciting, but if you can't tell which records you're allowed to send to an ad platform, you can't safely send any of them.

Notice what's missing. Favorite color quizzes, scroll depth, "content affinity" scores. Not useless, exactly. But in a lot of setups, none of that ever leaves the tool it was collected in. If a field doesn't feed a platform, a suppression list, or a report someone actually reads, it's storage cost with a privacy risk attached.

The benchmark to check this week: open Meta Events Manager and look at Event Match Quality on your Purchase event. Usercentrics says pixel-only setups typically land between 3 and 5, and it describes 6.0 to 7.9 as "Good." If Purchase is under 6, your collection or your pass-through is broken somewhere, and fixing that will probably do more than any new audience you were planning to test.

You probably don't need a CDP yet (and the stats say you'd underuse it anyway)

According to Gartner's 2025 Marketing Technology Survey, marketers use just 49% of their martech stacks' capabilities, and CMSWire notes that's a rebound from a low of 33% in 2023. Buying a customer data platform before you know which four fields you need is a pretty reliable way to add to the unused half.

For a lot of teams, the system of record already exists. An ecommerce brand on Shopify has customers, orders, and refunds in one place, and Klaviyo or a similar email tool is already holding email engagement on top of it. A B2B team has its CRM. The storage question is less "which platform do we buy" and more "which system wins when two of them disagree about the same customer."

A rough way to think about it: one source of truth per field. Orders and revenue live in the store or billing system. Lead status lives in the CRM. Consent lives wherever the person gave it, and everything downstream reads from there. A warehouse starts to make sense when you have three or more systems that each hold a meaningful slice of the customer, plus someone who can maintain the pipelines. Before that, it's mostly a very expensive spreadsheet.

Anyway, the practical version of this section is a one-page audit, and it takes maybe an hour. List your four fields across the top. List every tool that touches customers down the side. Mark where each field is collected, where it's stored, and where it's sent. Most teams find at least one gap that explains a lot: phone collected at checkout but never passed to Meta, gclid captured on the form but dropped when the lead moves into the CRM, that kind of thing.

Collecting data you never send anywhere is just storage with extra risk.

Where first-party data actually pays: three jobs

On paper, first-party data is supposed to power personalization, segmentation, lifetime value modeling, and five other things. In practice, the payback for a performance team comes from three jobs, roughly in this order.

1. Better signal back to the ad platforms

This is the fastest return. Google's enhanced conversions sends hashed email and phone from your conversion page alongside the tag, and Meta's Conversions API sends events server-side with the same kind of identifiers attached. The evidence here is mostly case studies, which is worth keeping in mind. Workshop Digital ran enhanced conversions across five client accounts for about five months and reported a 16% average increase in tracked conversions, with results ranging from a 33% lift down to a 3% dip on one account. On Meta, Swiss department store Globus ran a four-week A/B test with Meta and fifty-five comparing pixel-only to pixel plus Conversions API, and reported 49% lower cost per order and 89% higher ROAS in the CAPI cell.

One caveat that I think gets skipped a lot. Part of any "lift" from better tracking is simply better counting: sales that were happening before and weren't being attributed. That's still useful, because the platform's bidding algorithm learns from what it can see. But it isn't the same as new revenue, and if you report it as new revenue to your CFO you'll have an awkward conversation in about two quarters.

2. Exclusions, so you stop buying your own customers

This one is underrated. Every "new customer" campaign that can't recognize existing customers is spending some share of its budget re-acquiring people who already bought. Meta has started doing some of this automatically; it now auto-excludes two years of buyers from new customer ad sets, but that only works for buyers it can actually match.

Which brings you back to match rates. Google says in its Customer Match best practices that most advertisers' match rates fall between 29% and 62%. That's a wide range, and where you land inside it is mostly a function of data hygiene and how many identifiers you send per row. Even the top of that range leaves plenty unmatched, so the goal is moving up the range, not hitting the ceiling. The same best practices page recommends refreshing lists at least once a day if you upload by API, or weekly on a fixed schedule if you're doing it by hand.

The benchmark: if your Customer Match rate sits under about 30%, check formatting before anything else. Missing phone numbers and un-normalized emails seem to explain most of it.

3. A backend number to measure against

This is the measurement-series reason to care. Every platform report is the platform grading its own homework. The only number that isn't is the one in your own order or CRM data. It's what the ROAS sum test compares against, what a geo holdout reads as its outcome, and what a marketing mix model needs two-plus years of before it says anything useful. If your first-party data can't produce weekly revenue by region and by new versus returning customer, that's the gap to fix first. Honestly, it matters more than any personalization project on the roadmap.

And to be fair, none of this is new. Direct mail teams were matching customer files and running holdouts decades before anyone said "first-party data." The tooling changed. The discipline is the same, and it feels a lot less optional now that browser tracking misses more than it used to.

A 30-day plan if you're starting from a messy spreadsheet

The sequence matters more than the speed. Something like this works for most small and mid-size teams:

• Week 1: audit. Build the one-page field map above. Pull current Event Match Quality for Purchase (or Lead) and your latest Customer Match rate. Write both numbers down; they're your baseline.
• Week 2: fix collection. Make phone a visible (not hidden) checkout or form field. Capture gclid and fbclid on landing and persist them to the lead or order record. Normalize email and phone formats at the point of entry, not later.
• Week 3: fix pass-through. Turn on Google enhanced conversions and Meta's Conversions API with deduplication against the pixel. Send every identifier you're permitted to send on every event, not just Purchase.
• Week 4: build the exclusion and measurement layer. Set up an automated customer list sync (daily if possible) for exclusions. Then build one weekly report from your own backend: revenue, orders, new versus returning, by region. That report becomes the denominator for everything in the GA4 cleanup and ROAS audit work.

Targets after 30 days: Purchase EMQ at 6 or above, Customer Match rate inside Google's 29% to 62% band (ideally the upper half), and a backend revenue number you'd be comfortable showing finance. If you hit those three, you've done more than most teams with a six-figure CDP contract, which is a slightly depressing thing to type.

My prediction, for what it's worth: within 18 months, I'd expect at least one of Google or Meta to start gating some automated campaign features behind a minimum match quality threshold, the same way some features already sit behind spend or history requirements. The platforms need your identifiers more than they need your creative briefs, and they've been getting less subtle about it.

FAQ

Do I need a CDP for a first-party data strategy?

Usually not at first. If your store, CRM, or email platform already holds customer records with a stable ID, start there. A CDP or warehouse earns its cost once you have several systems holding different slices of the same customer and someone to maintain the connections.

How often should I refresh Customer Match lists?

Google recommends at least daily for API uploads and a fixed weekly schedule for manual uploads. Stale lists hurt exclusions most, since recent buyers are exactly the people you don't want to pay to reach again.

I don't think the teams that come out ahead here will be the ones with the most data. It's more likely the ones whose four boring fields are actually clean, actually connected, and actually leaving the building.