Tracksuit.
Research & deep dives
Log in
How our studies work

Tracksuit campaign studies: what to promise, what not to

Tracksuit offers two campaign studies. Sell either one, or both on the same campaign. Every Campaign evaluation sits beside the customer's Tracksuit tracker.

Campaign evaluation is in pilot. Confirm availability and timing with the Tracksuit team before promising dates.

The two studies

One explains, one counts and compares.

Campaign diagnostic: why the campaign landed

Themes and quotes from people who saw it

  • About 15 in-depth interviews with people who recognise the ad.
  • You hear what they took from it, whether they knew it was yours, who they thought it was for, what put them off and what they say they did next.
  • You get themes, quotes and a landed / partly landed / didn't land read.
  • Use it to decide whether to renew, rework or stop.

Campaign evaluation: how far it reached

Read as "higher among people who saw it"

  • A fixed questionnaire in the campaign markets, at Small, Medium or Large.
  • It counts how many people noticed the campaign over and above false claims, how many knew it was yours, and how many got the message.
  • It shows whether awareness and consideration are higher among people who saw it than among comparable people who didn't.
  • Use it to decide budget, channels, branding or message.

Buying both: same creative, dates and markets, briefed once, two reports.

Alongside the tracker: every Campaign evaluation sits beside the customer's Tracksuit tracker. One campaign rarely moves tracked awareness or consideration by a meaningful amount, so the in-study comparison is where the campaign's effect shows up. The tracker gives the category context.

What each can and can't claim

Promise what's under "Can claim". Steer away from what's under "Can't claim".

Campaign diagnostic

Can claim

  • What people who recognise the ad took from it, in their own words.
  • Whether they linked it to the brand, and who got the credit instead.
  • What held them back, and what they say they did next.
  • Whether the campaign landed, partly landed or didn't land, as themes at the number of interviews fielded.

Can't claim

  • Any count or percentage.
  • Any comparison between people who saw it and people who didn't.
  • A change in brand awareness or consideration. That's the tracker's job.
  • A winner between ads, a score against norms, or a pre-launch test.

Campaign evaluation

Can claim

  • How many in the target audience recognised the campaign over and above the false-claim floor, by execution and channel.
  • How many named the brand without seeing the logo, and which brands got the credit instead.
  • How many took out the intended message, shown beside a decoy message.
  • Whether awareness and consideration were higher among people who saw it than among comparable people who didn't, with the smallest gap the study can detect.
  • Whether consideration was higher again among people who saw it on more channels, as directional supporting evidence only.
  • What people who saw it say they did next, labelled as claimed.
  • What holds people back, and the one thing that would change it.

Can't claim

  • That the campaign caused a change in awareness or consideration.
  • That the gap between people who saw it and people who didn't is the size of the campaign's effect.
  • That the channel pattern proves the campaign worked.
  • A difference smaller than the study can detect at its sample.
  • That recognition is exact. It's self-reported. We mix in made-up ads to catch false claims, but this doesn't catch people who forgot.
  • Sales, conversions or ROI.
  • A winner between ads, a score against norms, or a pre-launch test.

The line to use: "Consideration was higher among people who saw it." Never "the campaign increased consideration" and never "+X points".

How the options compare

Where Tracksuit fits beside the other ways customers and agencies read a campaign.

OptionStrengthsWeaknesses
Tracksuit Campaign evaluation and Campaign diagnosticReaches outdoor, BVOD and radio as well as digital, where platform holdouts can't. Made-up ads catch false claims, and ads shown without the logo test whether people remember the brand. In evaluation, brand questions come before any ad is shown. Evaluation counts and compares; diagnostic explains why.The comparison shows association, not cause, and probably reads on the high side. Recognition is self-reported. Diagnostic counts nothing. Neither measures sales or ROI or tests ads before they run. Low-incidence audiences are hard and expensive to field.
Tracksuit tracker readGives the category context for every Campaign evaluation, from tracker waves that already exist. Always present: every Campaign evaluation customer has a matching Tracksuit tracker.One campaign rarely moves tracked awareness or consideration by a meaningful amount, so the campaign's effect shows up in the in-study comparison, not the tracker.
Meta and Google holdout studiesRandomised: people who were eligible to see the ads but weren't shown them form the comparison group. The closest of these options to showing cause. The right choice for a digital-only campaign.Only measures that platform's own media. Needs large response counts and has budget and eligibility minimums. Short closed polls say little about why.
Listen Labs and Outset-style post-campaign surveysNot covered in our evidence. Ask the vendor.Questions worth asking: does the recognition set include made-up ads, is the logo masked, and are brand questions asked before the ad is shown? Without these, recognition and brand naming read high.
Agency ad trackingNot covered in our evidence for ongoing tracking. Where the agency runs a match-market or geo test, the design is flexible and gives market-level outcomes.Match-market design quality varies, and controls get contaminated when markets aren't clean. Any comparison of people who saw it with people who didn't carries the same high-side reading as ours.
Marketing mix modelling (MMM)Not covered in our evidence.Not covered in our evidence. If the question is sales or ROI, neither Tracksuit campaign study answers it.

Plain limits

Say these up front. They protect the customer and the result.

  • Association, not cause. People who notice ads were already different: more likely to own the brand, know it and use a lot of media. We balance the two groups on ownership before the campaign, media use, age and gender. We can't balance what we didn't measure.
  • Matched comparisons usually read on the high side. Checked against 663 randomised Facebook experiments, matched and machine-learning methods using over 5,000 user features still missed, mostly too high. Treat the gap as a sign of what the ad did, not a measure of its size.
  • Many honest studies find no clear gap. Ad effects are usually smaller than these studies can detect, so "no difference large enough to tell apart from chance" is a common, valid result.
  • Bias checks run at Medium and Large. We run the same comparison on competitors that weren't advertising and on a made-up brand. Those gaps should be near zero. If they show a gap of similar size, the report says the gap mostly reflects who notices ads and makes no headline claim.
  • Low incidence costs more. Every screened person is paid for. A brand few people notice, or a category few people are in the market for, means more screening and a higher price, and may mean a smaller tier.
  • Field soon after the campaign. Within about two weeks of it ending, four weeks at most, because a short burst is forgotten quickly.

Campaign evaluation tiers

Pick the tier from the decision. Each one carries its own caveat.

Tiers assume about a quarter of people screened recognise the ad. If expected reach is lower, we recommend a smaller tier or a lower target rather than set quotas we can't fill.

Small: about 40 people

  • Recognition as a count out of about 40 (12 of 40 carries about ±14 points).
  • Brand linkage and take-out as counts among the roughly 10 who recognise the ad, plus themes and quotes.
  • Caveat: directional only. No rates, no comparison between people who saw it and people who didn't, no channel splits, no bias checks.

Medium: about 150 people, from about 380 screened (the default)

  • About 100 people who saw the ad and about 50 who didn't.
  • Recognition rate to about ±4 points. Directional rates among people who saw it, about ±10 points.
  • Caveat: today the comparison detects gaps of about 21 points or more on consideration near 15%. This improves to about 14 points once legal clears using screening answers, which isn't live yet. Bias checks decide how strongly we word the result.

Large: about 300 people, from about 760 screened

  • About 200 people who saw the ad and about 100 who didn't.
  • Recognition rate to about ±3 points. Rates among people who saw it to about ±7 points.
  • Caveat: today detects gaps of about 15 points or more near 15% consideration; smaller differences can't be told apart from chance. This improves to about 10 points once legal clears using screening answers, which isn't live yet. The channel pattern stays directional. Bias checks are charted beside the brand. Large stops at about 300.

Tracksuit · Campaign diagnostic and Campaign evaluation · Draft for sales and partners · October 2026