TL;DR
- None of this was a creative breakthrough: all three changes were operational, on a Meta budget above $5M a year where even small improvements compound.
- Audience grid – a fixed set of Meta target audiences, ranked by historical acquisition cost and locked for six months, so no segment is missed and no two overlap. Alongside it: no-code rules that cut losers and scaled winners without waiting for a manual audit, and a written protocol for creative tests, rotation and relaunch.
- CAC fell 27%, which let Meta spend double within 5 months at the ROI KPI; automation alone accounted for −16% on a statistically significant A/B test.
- The share of successful creatives rose 4×, to 25%, and FTE on Meta maintenance fell 13%.
Situation
Scentbird is a subscription-based fragrance service that allows customers to discover and try new perfumes and colognes monthly in the United States.
Scentbird utilizes the subscription model, and business growth is directly connected to active user base growth. Nevertheless, it's critical for business margins to optimize customer acquisition costs and focus on selling products with higher margins.
At the beginning of 2017, Scentbird was a startup; thus, it was critical to demonstrate steady growth with all the KPIs met, primarily including ROI.
In Scentbird, I began as a media buyer and was responsible for user acquisition on Meta.
Task
Meta was the major user acquisition channel for Scentbird, which brought the most valuable users from the ROI perspective. My tasks were to:
- optimize customer acquisition costs;
- scale the user acquisition meeting the KPI on ROI;
- automate usual processes and other operations to reduce the required FTE to maintain the channel.
Action
The main area of improvement here was the operation processes in both different user acquisition channels and in creative and analytics streams.
At the beginning of the journey, I stumbled upon the fact that Meta's media buying budget exceeds $5M annually. Keeping that in mind, I've concluded the following:
- such volume affords us to perform a decent number of A/B tests quickly and simultaneously;
- not only major optimizations are significant, since we would see effects of even minor improvements over time due to such volume;
- the routine procedures' automation would let us avoid mistakes and spare some FTE for analysis and further A/B tests' planning.
The systematic approach would let us see the bigger picture and determine the areas for A/B testing. Thus, I did the following:
- Implemented the audience grid: ranked the main target audiences by customer acquisition cost using the historical data; fixed (for a six-month period) the particular target audience sets; this approach let us ensure that we were not missing a segment; also, we were sure that the segments didn't overlap.
- Implemented the first version of no-code automation, which let us turn off ad sets immediately upon the occurrence of the predefined circumstances, and scale the ad set budget immediately upon the occurrence of the predefined circumstances; as a result, it lets us be sure that we cut losers and scaled winners at any given time rather than waiting for manual auditing.
- Built the day-to-day creative handling: coming up with benchmarks and hypotheses for creatives; rules of creative tests and rotation; rules for successful creatives' relaunch.
I formulated the following hypothesis:
- audience grid would decrease FTE on Meta media buying by at least 10%, since there would be no need for constant audience set-up pick and recheck;
- no-code automation would decrease customer acquisition cost by at least 15% due to timely budget reallocation between ad sets;
- transparent creative reporting implementation would decrease customer acquisition costs by at least 25%, since it would simplify the creative picking process and let us turn off the creative immediately after burnout.
Result
- Audience grid implementation decreased FTE for Meta maintenance by 13%, due to the opportunity to duplicate the audiences instead of picking from scratch and to decrease the time for results analysis.
- No-code automation decreased customer acquisition cost by 16% (based on statistically significant A/B test results).
- Increased the share of successful creatives (which led to Meta campaigns scaling at the target ROI) by 4 times (up to 25%).
- The measures mentioned above led to a customer acquisition cost decrease of 27%, which led to the opportunity to double spending on Meta in the first 5 months while still meeting the KPI on ROI.
- The business model, way of monetization, average retention rate, and other product and business metrics remained unchanged.
Results in numbers
| Metric | Target | Result | How measured |
|---|
| Customer acquisition cost, combined | – | −27% | Combined effect of all three measures; no combined hypothesis was set. ROI KPI held |
| CAC from no-code automation | ≥ 15% | −16% | Statistically significant A/B test |
| FTE on Meta maintenance | ≥ 10% | −13% | Time on audience setup and results analysis after the audience grid |
| CAC from creative reporting | ≥ 25% | Not isolated | The CAC effect of creative reporting alone was not measured separately; it is inside the combined −27% |
| Share of successful creatives | – | ×4, to 25% | Creatives that scaled Meta campaigns at target ROI, after the creative reporting |
| Meta spend | – | ×2 in 5 months | Annual Meta media buying budget above $5M at the start |
Table 7. Targets are the three hypotheses set before the work; results are as stated in this case.
"Maria is a great professional in the performance marketing field. She began working in my team at Scentbird as a Meta Ad Manager and grew to UA Group Head. She has successfully met her KPIs for three consecutive years with a top marketing budget under management of $2M/month. Maria has a solid mathematical and statistical background and a deep understanding of digital marketing channels and Ad Tech. She has a great ability to manage multiple projects and uses a very systematic approach to the team's benefit."
– Oleg Popov, VP of User Acquisition · Scentbird
Written by Maria Atamanova – Senior Growth & Performance Marketing Manager and applied mathematician; twelve years growing businesses – apps, e-commerce, EdTech, AI data platforms – on numbers, not gut feeling. All three measures are operational, not creative: what changed was how audiences, budgets and creative tests were handled day to day.
More cases
