What counts
First-party data is anything your company collects directly. Third-party data comes from outside: public conversation, search trends, reviews and research. Most brand teams have more of the first kind than they think. It’s just in pieces: sales with finance, email with the CRM team, social stats with a partner, each in a tool that doesn’t talk to the others.
A useful test: if your company could hand the data to an auditor and say “we recorded this ourselves,” it’s first-party. If you’d have to say “a platform told us” or “a vendor estimated it,” treat it with more care. Agency and partner reports sit in between. You received them directly, but they describe what the platforms saw, which is why they belong next to your own numbers.
| Data | What it can tell you | Where it usually lives |
|---|---|---|
| Sales, tickets, sign-ups, viewing | What happened, when, and through which channel | Finance, ticketing, retail partner reports, app or platform dashboards |
| Customers | Who buys again, and who your best customers are | Your CRM or customer list |
| Which messages drive results, and how many results came from email | Your email tool | |
| Your social accounts | Which posts get watched and which get shared | Platform exports or your social tool |
| Ad accounts | What ran, when, where, and what it cost | The ad platforms, run by your team or a partner |
| Agency reports | What each partner reports it drove | Decks and spreadsheets in your inbox |
Questions it can answer
- Which posts work? In a real Arcus content report for a hard seltzer brand, the brand’s own social export showed one type of post, a trick with the can you could copy tonight, got a median of 19,188 views on Instagram against an account average of 8,148. That’s 2.4 times the average, read from data the brand already had. See the content report.
- Did the campaign line up with results? Your own numbers are what every channel’s report should be lined up against.
- How much came from email? An ad platform often credits itself with a sale that came from email, because one of its ads was somewhere in the path. Your email tool knows better.
- Who are our best customers? Usually a small group drives a big share of results. Knowing who they are changes who you write for.
Notice what these questions have in common. Each one is about a decision: what to make, what to fund, what to stop, who to talk to. First-party data earns its keep when it’s pointed at a decision someone has to make this quarter. Pointed at nothing in particular, it turns into a dashboard nobody opens.
A worked example
Made up to show how it reads, not from a client.
Your sales are up 18% in week 3, and your team wants to know why. Your first-party data shows what happened: sales by day from finance, and which buyers were new from the CRM. It doesn’t show why. What people said online that week points to a reason: 1,240 TikTok and Reddit posts about one creator you never paid. The next step is to test it: pay that creator in half your regions, and compare sales with the other half.
That’s the job first-party data does best. It isn’t glamorous. It’s the list of what happened, which every other number should agree with.
Better together
On its own, first-party data tells you what your customers did. Third-party data tells you what people are saying. Lined up, they tell you whether the conversation turns into results.
In a real Arcus audience report for a streaming service, only 1.9% of 4,283 questions men asked each other on one large forum were about hobbies, games and entertainment. They didn’t ask each other what to watch. The talk happened under the shows instead: 78,682 comments under four trailers. That changes where a brand should listen, and which of its own numbers, like viewing by title, it should line the conversation up against. See the audience report and our guide to listening that counts.
The same pattern works for a drinks brand, a game or a fashion label. Count what people say about the product, the moment or the occasion, then line it up week by week with your own sales, sign-ups or streams. When a topic rises in conversation and your numbers rise with it, that’s worth testing. When it rises and your numbers don’t move, that’s worth knowing too.
Where to start: a checklist
- Write down one question you’ve been asked and couldn’t answer. “Is the creator program working?” is a good one.
- List the data that touches it: sales or sign-ups by day, creator codes, the creator recap, your email sends.
- Find who owns each piece. Finance, a partner, your own team.
- Check the dates line up. Same days, same time zone, same definition of a result.
- Put each report’s number next to your own number for the same dates.
- Write down what you still can’t answer. That list is your next step.
Common mistakes and limits
- Waiting until it’s all in one place. Start with the two or three sources one question needs.
- Letting the ad platforms be the record. They report what they saw. Your own numbers report what happened.
- Collecting more instead of reading what you have. Most teams already hold data nobody reads.
- Forgetting retail and partners. If most of your results come through someone else, their reports are part of your first-party picture.
Questions worth asking your team: Can we get every report’s results by day, so they line up with our own numbers? Which of our own channels, like email, could explain results an ad report counts? What’s missing from our data that would make the next answer easier?
What it doesn’t prove: first-party data tells you what happened and lets you see which reported results held up. On its own it can’t tell you why. A spike in sign-ups the week a campaign ran lines up with the campaign. Proving it caused them takes a holdout test.
If nobody on your team has time for the checklist, try it free with the question from step 1. We answer it with your own numbers and what people say online, then walk you through it in 30 minutes, so you see what your data can already tell you. After that, we connect every piece into one place you own, inside your own setup, and answer the next question with the whole picture.