Display retargeting saturation before a sitewide sale
An online apparel retailer spends $25,000 per month on Google Display. Its retargeting campaign uses a 30-day audience window, dynamic remarketing, a frequency cap of 8 impressions per day, and five creative assets that have not changed in five months. Over the last quarter, CTR fell 40%, conversion rate stayed flat at 2.1%, and reported CPA rose 28%. View-through conversions are included and make up 35% of reported conversions; the conversion window is 7-day click, 1-day view. The audience has shrunk from 310,000 to 180,000 users, and the team has added a broad 'similar audiences' layer to compensate. Finance questions whether retargeting still works. The retailer has healthy margins but cannot discount further. A 14-day sitewide sale begins in three weeks.
1. Is the CPA rise more likely creative fatigue, audience saturation, or measurement inflation? What checks would separate them?
Read the worked answer
The pattern points first to creative fatigue and audience saturation, with measurement inflation as a secondary driver. CTR fell 40% while conversion rate held at 2.1%; if the landing page or offer had broken, conversion rate would usually fall too. The audience shrank by about 42%, from 310,000 to 180,000, while the cap allows 8 impressions per day and creative has been unchanged for five months. That combination produces repeated exposure and declining response.
Separate the causes with these checks:
- Frequency and reach: pull average impressions per user, frequency distribution, and reach. A high share of users at 6–8+ impressions supports saturation.
- Creative decay: compare CTR by asset age and by asset. A steep decline across all assets suggests fatigue; one weak asset suggests rotation opportunity.
- Audience quality: compare click-through and conversion rate for the original retargeting list versus the broad similar-audience layer. A diluted layer can raise CPA without proving retargeting failed.
- Measurement: split click-based CPA from view-through CPA. If click CPA is stable and the reported CPA rise comes mainly from view-through conversions, the issue is reporting inflation, not necessarily performance. Check conversion lag and cross-device behavior.
A three-week sale deadline raises the cost of guessing, so run the checks before cutting the program.
2. Recommend changes to audience window, frequency cap, creative, and campaign structure for the sale. What would you test?
Read the worked answer
Restructure retargeting by recency instead of treating the full 30-day audience as one pool. Create 0–7-day, 8–14-day, and 15–30-day segments, and exclude recent purchasers where appropriate. For the sale, prioritize the 0–14-day segments and use a shorter conversion window, such as 14–21 days, so the campaign does not keep bidding on stale intent.
Lower the daily frequency cap from 8 to roughly 3–4 and monitor average frequency, not just the cap. Rotate in at least three new creative concepts: a direct sale message, a product-category dynamic ad, and a social-proof or urgency asset. Keep dynamic remarketing for catalog logic, but do not let one template carry all exposure. Move the broad similar-audience layer into a separate prospecting campaign so it cannot dilute retargeting CPA.
Test with a holdout: withhold 10–15% of the retargeting audience or use matched geos, then compare incremental conversions and profit, not reported CPA. Also test the shorter window versus the 30-day window while holding creative constant. Because margins are healthy but discounts are capped, judge success on profit after returns and on incremental conversions, not on the sale discount alone.
3. How should finance evaluate retargeting's incremental value given that view-through conversions are 35% of reported conversions?
Read the worked answer
Finance should not treat all reported view-through conversions as caused by Display. View-through conversions are useful for understanding upper-funnel influence, but they are especially vulnerable to claiming conversions that would have happened anyway. Separate reporting into click conversions, view conversions, and total, and set a consistent view window for comparison.
Use a conversion lift or geo holdout to measure incremental conversions and revenue among users exposed to retargeting versus a control group. Calculate incremental profit after product cost, returns, and any sale discount. For day-to-day bidding, use click-based CPA or a conservative blended CPA; for budget decisions, use the lift result.
A practical answer to finance is: retargeting may still work, but the current 35% view-through share is not evidence of 35% incremental value. Run the holdout through the sale period, then compare incremental profit per dollar against prospecting and against holding the budget. If the control group converts nearly as well, shift budget; if lift is positive and profitable, keep a capped retargeting program with refreshed creative and lower frequency.