Scaling a DTC Coffee Subscription Without Overlapping Audiences
Roast & Co. sells coffee subscriptions and spends $25,000 per month on Video action campaigns. It has a 42,000-address email list, 11,500 purchasers, and 3,200 purchasers in the last 90 days. The Google Ads tag has a 90-day window and captures add-to-cart, begin checkout, and purchase. Current targeting uses a custom intent segment built from coffee subscription keywords plus the affinity segment Food & Dining Enthusiasts. The campaign has a 2.2 target ROAS, but actual ROAS is 1.4. Frequency is 8.1 impressions per user per week, and 35% of spend goes to users who have purchased in the last 90 days. The team has not used Customer Match, similar segments, or placement exclusions. It wants to scale spend to $40,000 per month while improving ROAS, and it cannot increase discounting beyond 15%.
1. Build audience layers and exclusions for Roast & Co. that could improve ROAS while allowing spend to scale. Explain the role of each layer.
Read the worked answer
Start by separating prospecting, remarketing, and existing-customer treatment. For prospecting, build a Customer Match similar segment from the highest-value purchasers, not all purchasers, so the seed reflects subscribers who retain rather than one-time discount buyers. Add a custom intent segment around competitor subscription terms and coffee brewing terms, and test an in-market coffee segment alongside the broad affinity segment. The current affinity segment is too wide to carry most of the budget at a 2.2 ROAS target.
For remarketing, use the 90-day site audience split into cart abandoners, checkout abandoners, and product viewers, with separate messages and frequency caps. Exclude recent purchasers from prospecting; the setup shows 35% of spend reaching them, which is pure cannibalization. Use Customer Match to suppress the 3,200 recent purchasers and to build an upsell campaign for them if the margin supports it. Add placement exclusions for low-quality apps and channels that waste spend. Scaling should come from similar segments and expanded custom intent, not from removing all exclusions. Each layer needs its own budget and ROAS read so a strong remarketing layer does not hide a weak prospecting layer.
2. How should Roast & Co. handle frequency and overlap between prospecting and remarketing, and what risks come with scaling to $40,000?
Read the worked answer
Frequency at 8.1 per week is too high for a considered subscription purchase and suggests the same users are seeing the same message repeatedly. Set a lower frequency cap for prospecting, while allowing a higher cap for cart abandoners who need a reminder. Use audience exclusions so prospecting does not serve recent purchasers and so remarketing does not serve users who already converted. Put prospecting and remarketing in separate campaigns so budget and frequency can be controlled independently.
The main scaling risk is that the extra $15,000 buys more of the same inefficient affinity audience and depresses ROAS further. Scale in steps, for example 20% every two weeks, and judge each step on incremental ROAS after the conversion lag. Watch frequency, view rate, and new-customer share. If the similar segment is small, expand with custom intent and creator-led creative rather than loosening exclusions. The 15% discount ceiling means the team cannot buy its way out of poor targeting with deeper offers, so the scale plan must depend on audience quality and creative relevance.
3. What measurement plan would prove scaling is incremental rather than cannibalizing existing sales?
Read the worked answer
Separate new-customer revenue from returning-customer revenue and report ROAS for each. If the extra spend mostly reaches people who already bought, total ROAS may look acceptable while incremental revenue is flat. Use Customer Match to identify existing purchasers and compare exposed versus suppressed groups where feasible. A geo holdout or Google Ads experiment is stronger than pre-post reporting because it controls for seasonality and organic demand.
Set a decision rule before scaling: for example, require incremental ROAS above 1.6 on the added spend, new-customer share above 60%, and frequency below 5 per week in prospecting. Track view-through conversions separately from click conversions, but do not let them justify spend without a holdout check. Use a consistent attribution window and wait for the lag before judging each scale step. If the holdout shows no incremental revenue, the right move is to fix audience exclusions and creative rather than increase budget.