Rules
Part of Paid media strategy: a governance framework
How to assess paid media strategy case studies
A practical framework for assessing paid media strategy case studies, checking claims, measurement, costs, concurrent changes, and whether a result transfers.
What to take away
- Uber Eats and TUI UK appear in Google Marketing Platform customer stories, so the numbers are publisher and customer claims, not independent audits.
- Read the mechanism firstreach and frequency control in one case, media and reporting consolidation in the other.
- Fill an evidence table before you quote any headline result to a client, a budget holder, or a board.
- Write a one-page transfer memo. It can legitimately recommend no test yet.
- Store the record with the decision and set the review date.
For paid media strategy case studies, the useful question is narrow: what do you need on the page before you copy anything from a published case?
Start with the claim, not the number
A vendor case study answers one question only: what did the publisher say happened. It cannot tell you what will happen in your account, your market, or your auction.
So begin with a claim record. Write down the sentence you would repeat, along with the publisher, the customer, the page date, and the product version. Attribute the figure every time you use it.
A wider set of published examples sits in paid media strategy examples, and it is worth reading beside these two cases because the reported setups differ more than the headlines suggest.
Assessing the Uber Eats frequency claim
Google Marketing Platform published a story about Uber Eats coordinating reach and frequency around football events with Display & Video 360 and Ads Data Hub. Put the reported change, the event window, and the date on one line.
Before you repeat that figure, capture four things: the baseline definition, the frequency caps applied, the audiences, and the markets covered. If the page omits cost or a holdout, treat the result as a lead for testing.
Case-study evidence checklist
- Who published it? Vendor, customer, agency, funder
- What changed? Media, data, creative, bids
- Compared with what? Baseline, control, dates
- How was success counted? Source, denominator, attribution
- What did it cost? Media, services, tools, labor
- Could it transfer? Market, category, scale, skills
Assessing TUI UK's stack-unification claim
A second Google Marketing Platform story describes TUI UK consolidating media buying and reporting across several of the same products. Multiple changes shipped together, so no single product should receive full credit.
Record what the reported savings cover. Media cost, working media, and total spend are three different numbers, and the page may name only one of them.
Case-study questions and the evidence to capture
| Question | What to record |
|---|---|
| Who published it, and when? | Publisher, customer, page date, product version |
| What was measured? | Delivery, attribution, or incrementality, plus the metric definition |
| What changed at once? | Media, creative, tracking, pricing, staffing, seasonality |
| What is missing? | Fees, internal costs, customer effects, selection criteria |
| Can it transfer? | Audience, data access, platform availability, budget, skills, outcome timing |
The paid media strategy checklist works as a second pass here, because it forces each row above to be checked against your own account rather than the publisher's.
Nine mechanisms worth turning into tests
- Frequency controlcap impressions per user per week so one household cannot absorb the whole budget.
- Channel consolidationmove spend into fewer platforms when reporting lag, not personal preference, drives the split.
- Audience exclusionsuppress recent purchasers or converters before the first impression, not after.
- Sequential creativeorder messages so the second ad assumes the first was seen, then test the order.
- Inventory-linked buyingtie placement to the job, such as video completion or pure reach.
- Local capacity routingsend leads only to locations that can serve them within the promised time.
- Matched-market testscompare two similar geographies, one exposed and one held, across the same weeks.
- Customer holdoutskeep a small share of eligible users unexposed and measure their behaviour.
- Offline-outcome reconciliationjoin CRM or call outcomes back to the campaign before you judge it.
Automated bidding changes how most of these behave. Maximize Conversions sets bids automatically to get the most conversions inside a budget, so the targets you feed it must come from your own economics.
Example: a transfer memo
Decision requested: test weekly frequency caps on the winter campaign. Claim under review: the published Uber Eats story reports reach and frequency improvement around football events. What must be true: the audience is large enough to cap, measurement can read reach, sales operations can absorb the same lead volume, and the placement supports caps. Not established by the case: our costs, our auction, our creative, and any holdout result. Test: two matched markets, four weeks, threshold written before launch. Owner and review date: named analyst, review on day 35.
The paid media strategy best practices page follows the same order, connecting a claim to economics and then to a controlled test.
Audit published evidence
- Check reasonable basis for objective claims
- Apply FTC review-integrity checks to named providers
- Separate claims, statements, records, findings
- Record criteria, relationships, version, population, method
- Reproduce one material claim with local data
- If method missing, treat as lead for testing
Applied substantiation for US and Canadian claims
Under the FTC advertising substantiation policy, an objective claim needs a reasonable basis before you publish it. That applies when a vendor figure moves into your own pitch deck or ads.
The FTC guidance on reviews and endorsements covers hidden relationships, incentives, and selective presentation. If you feature a partner's case study, disclose the relationship.
Canadian advertisers face a second layer. Competition Bureau rules govern the claims you make to Canadian buyers, and PIPEDA consent rules shape the tracking data behind every reach and frequency measurement you build.
Common questions
Are vendor case studies trustworthy enough to act on?
Only as a source of testable mechanisms. The figures belong to the publisher and the customer, so your own baseline, controls, and costs come first.
What is the minimum evidence before I quote a headline result?
Publisher, customer, date, product version, and the concurrent changes. Without those, present the claim as a lead rather than a result.
How do I transfer a frequency cap or a stack consolidation?
Run a matched-market or holdout test with the threshold written down before launch, then judge it at equal spend against your own baseline.
What changes if I advertise in Canada?
Consent rules shape tracking, and provincial rules add tax and language duties to the same campaign.
