Weekly Refreshed In-Market Audience Lists, Explained

A weekly refreshed in-market audience list is a segment of consumers predicted to be actively shopping a category right now, rebuilt on a recurring weekly cycle so the people in it are the ones still in-market, not the ones who were in-market last quarter. You pull it, push it to your ad platform or CRM, then pull it again the following week. The overlap between the two pulls is your carryover; the difference is your new entrants. The single most useful thing about a weekly list isn't its size. It's the churn: who left, who arrived, and how fast.

Last updated: September 2026

Why in-market lists decay faster than demographic ones

A demographic segment is stable. Someone who is 42, owns a home, and has two kids under ten is going to be all of those things next month too. You can build that audience once and run it for a quarter without it rotting underneath you.

In-market is the opposite. It's a claim about a temporary state, this person is shopping for a truck, a mortgage refinance, a new gym membership, right now. That state has a half-life measured in weeks. Some of it collapses even faster, because the outcome that ends the state is exactly the outcome you were targeting: they bought. The moment somebody buys the truck, they're the worst person in your audience. They're also still sitting in your list, absorbing impressions, unless something removes them.

If you've run an intent audience before and watched performance fall off a cliff by week three, that's why. The list didn't get worse at finding buyers; it kept the buyers it already found, long after they stopped being buyers. No wonder the numbers sag, you were paying to re-reach people who had already left the market.

That's the actual argument for refresh cadence, and it's worth being precise about it. Refreshing isn't about adding more people. It's about removing people whose in-market state has expired. Most people don't realize this cuts against how intent data is usually sold: vendors compete on list size at the moment of purchase, which is the one metric a stale file can fake indefinitely.

Strip it back: three things actually matter

This topic has accumulated a lot of vocabulary: signal decay curves, propensity half-lives, recency-weighted scoring. Almost none of it changes what you do on Monday morning. Here's the short version.

1. How often the list is rebuilt, and against what. "Weekly" can describe two entirely different things: how often the underlying consumer data is updated, or how often the in-market prediction is recomputed. Those are not the same, and a vendor can honestly say "weekly refresh" while meaning only the first. Ask which one they mean. The answer changes what a re-pull is worth to you.

2. Whether you can compute a delta. If you can't diff this week's pull against last week's, you don't have a refreshed list, you have a new list every week that you have to reconcile by hand. The delta is where the value lives, because new entrants are the people who moved into the category since your last campaign touched them.

3. Whether the identity underneath holds still. A refreshed list is only useful if "the same person" means the same person across pulls. If the matching is probabilistic, the same human can appear under two different records in consecutive weeks and your delta becomes noise. Exact Match matches deterministically, with every match verified against multiple identity anchors (name, email, phone, address) rather than statistically inferred, which is the property that makes week-over-week comparison mean anything at all.

That's it. Cadence, delta, stable identity. Everything else is commentary.

What a weekly pull looks like in practice

Run it on a fixed day. Tuesday is a reasonable default because it gives you a full working week to act on the file before it's a week old, but the specific day matters far less than the fixedness, you want a comparable interval, not an optimal one.

Then do three things with the file:

• Suppress the leavers. Anyone in last week's pull who isn't in this week's has probably exited the category. Move them out of the active campaign before they eat more budget.
• Treat new entrants as their own segment. They're the freshest signal you have. Giving them the same creative as someone who's been in-market for five weeks wastes the one advantage the refresh bought you.
• Leave the carryover alone. People persisting across pulls are in a longer consideration cycle. They need frequency, not urgency.

On budget: a weekly cadence tempts people into weekly creative changes, and that's usually a mistake. You'll destroy your own read on performance by changing two variables at once. Change the audience weekly; change the creative on your normal testing schedule.

Where Exact Match fits

Exact Match's in-market prediction lives in Predict ID, which uses Bayesian behavioral modeling to identify consumers actively in-market for a category in real time, not a static historical score. Audiences get built and exported through Audience ID, across 80,000+ targeting clusters spanning nine data domains (demographics, behavior, interests, financial attributes, and intent signals among them) over a base of 250M+ verified U.S. consumer profiles. Exports run as background jobs with CRM-specific column templates, so the file lands mapped rather than needing a hand-cleanup pass.

Here's the part worth being precise about, since this post is about refresh cadence specifically. Predict ID's behavioral signal, the thing actually powering the in-market call, updates daily, not weekly. The same daily cycle runs the identity graph underneath it: contact records, demographic attributes. So when you run a weekly pull, you're never looking at a signal that's more than a day old at the moment it was computed. The week is your operating cadence, not a ceiling the platform imposes on you.

The export itself isn't something that reruns on a schedule you set once and forget. An Audience ID export runs as an on-demand background job: you kick one off, it processes, you download the file. There's no recurring export that fires itself every Monday morning. Building a weekly cadence around it, a recurring calendar hold, a script that calls the export on a timer, is on you, not a switch you flip once inside the platform.

For the broader mechanics of building these segments, see in-market audience targeting and in-market buyer prediction. If the commercial terms are what's blocking you, audience data without annual contract covers that side.