About the client
A regional food and grocery delivery service expanding into new cities: without a large user base, neither courier network effects nor the restaurant pool work. Marketing’s goal was app installs on a CPI model with quality control — downloads matter only if users stay through the first week.
GUGA MEDIA built the campaign around mobile audiences aged 18–35 in target geos. In parallel we aligned onboarding and a first-order promo with product — without that, even cheap CPI does not become revenue.
Strategy and tools
The focus was TikTok and Instagram Reels: visual “food on camera”, fast editing, a clear offer in the first seconds. We mixed macro and micro creators — the first for awareness spikes, the second for cheaper CPI through high local trust.
Every author received a checklist: show the app screen, enter address, confirm order, tight shot of the promo code. We banned fake discounts and hidden delivery terms — that reduced store negativity and post-install churn.
We used tracking links and partner pixels where available; daily checks against the client’s MMP on post-install events. Some creators had a quality-install bonus, not only a posting fee, to encourage creative iteration.
Challenges and fixes
Early on, some authors delivered “cosmetic” installs from irrelevant audiences — CPI was low but D7 retention collapsed. We set a retention floor and cut sources that failed the seven-day benchmark. TikTok moderation rejected overly aggressive offers; we kept alternate cuts without trigger words.
Seasonality and competition with aggregators meant many users already had two delivery apps. Creatives emphasised local restaurants and delivery time rather than abstract “food” to differentiate from global players.
Results in detail
The campaign generated 50,000 installs at an average CPI of 42₽. Day-seven retention was 28% on the weighted mix — above the product KPI and proof the traffic was not one-off.
Top creatives showed a real order “from screen to door”: slightly higher entry CPI but better first-order conversion and 30-day LTV. Pure memes without the product cheapened installs but killed retention.
The client received city and creator cohort cuts, geo expansion recommendations, and a creative library adaptable to paid social. We also documented common moderation rejections to speed future launches.
Marketing–support coordination was critical: install spikes drove chat volume; without ready answers on zones and promos, first-order conversion would have dipped in week two.
We cross-tested music and first frames: TikTok favoured trending audio with the app visible in second one; Reels performed better with calm food shots and voice-over — one creative cannot always be cloned across platforms.
Playbooks for new cities included delivery-zone checks in creatives, regional micro creator lists, and CPI/retention gates for a second budget wave.
We measured the first-order promo: without it CPI looked better but install-to-purchase in 72 hours fell; total paying-user cost was lower with a modest promo than with “clean” CPI plus expensive in-app retargeting.
Transparency with creators helped: when they knew we optimised D7 retention, not installs alone, they reshot more willingly — fewer one-and-done integrations and faster iteration without extra briefing rounds.
Bottom line: CPI stayed the buying KPI, but quality was the second cutoff — without it, scale would inflate churn in onboarding, support, and store ratings.