Remarketing List Segmentation and Optimized Targeting Drift
A subscription coffee company runs Display remarketing with these lists: all site visitors 30 days (240,000 users), product page viewers 30 days (60,000), cart abandoners 14 days (7,500), active subscribers via Customer Match (45,000), and lapsed subscribers (30,000). There is a single remarketing campaign targeting 'all visitors' with a 540-day membership duration, a frequency cap of 4 impressions per day, and optimized targeting enabled. The bid strategy is Maximize Conversions with no target. After six weeks, 60% of impressions and 45% of conversions come from active subscribers, who already buy every month. Cart abandoners receive 1.1% of impressions. The team wants to raise the frequency cap to 10 to 'stay top of mind' and merge every list into the same campaign to 'consolidate spend'.
1. Diagnose the audience setup. How should the lists be segmented, and what membership durations make sense?
Read the worked answer
The setup collapses distinct intents into one pool. A 540-day membership on 'all visitors' is far too long for a coffee subscription with a short consideration cycle; it keeps stale users eligible and inflates the pool to 240,000, which then dominates delivery. Active subscribers are the largest, easiest-to-convert group, so they absorb 60% of impressions and 45% of conversions - conversions that mostly would have happened anyway.
Segment by intent and recency: cart abandoners 7–14 days (small but highest intent), product page viewers 30 days, all visitors 30–45 days, lapsed subscribers 90–180 days, and active subscribers used only for upsell or excluded entirely. Each segment gets its own ad group or campaign with messaging matched to the stage. Cart abandoners need dynamic remarketing with the viewed product; lapsed subscribers need a win-back offer; cold visitors need a value proposition, not a discount. Membership duration should reflect how long the intent remains plausible.
2. Explain why active subscribers dominate delivery and what optimized targeting does in this setup.
Read the worked answer
Delivery follows predicted conversion likelihood and list size. Active subscribers are 45,000 users with recent purchase behavior and high predicted conversion rates, so Maximize Conversions bids aggressively for them; they are also a substantial share of a 240,000 pool. Cart abandoners are only 7,500 and may have lower predicted value after a few days, so they receive 1.1% of impressions despite being the most valuable segment.
Optimized targeting expands beyond the selected audience to users predicted to convert, using conversion data as a signal. That can be useful for prospecting, but here it compounds the problem: the model learns from conversions that are dominated by existing subscribers and finds more people like them, not more cart abandoners. Optimized targeting should be used on prospecting campaigns with clean conversion signals, or with active subscribers excluded, not on a catch-all remarketing campaign where the easiest conversions define the audience.
3. Recommend a campaign structure, exclusions, and a way to measure whether the remarketing is incremental.
Read the worked answer
Build separate campaigns by stage: cart abandonment, product viewers, lapsed subscribers, and a prospecting campaign for cold visitors. Exclude active subscribers and recent purchasers (for example, 30-day purchasers) from all acquisition and remarketing campaigns except a dedicated upsell campaign. Keep frequency caps at 1–2 per day for remarketing; raising to 10 adds cost and annoyance without evidence of lift. Reserve optimized targeting for prospecting where subscriber contamination is removed.
Measure incrementality, not platform conversions. Use a holdout: suppress a random 10–20% of each list from seeing remarketing, or run a geo holdout in matched markets, and compare conversion rates and revenue between exposed and unexposed groups over four weeks. Track new-customer share and subscription starts, not just conversions. If cart abandoners convert at a much higher rate than the holdout, the campaign is incremental; if active subscribers convert at the same rate with or without ads, that spend is wasted.