How to Build a Unit Economy Model to Improve Media Buying Strategy
44% acquisition budget cut · revenue unchanged
Toloka is a crowdsourcing platform built inside Yandex – spun out in 2025 with Nebius Group (ex-Yandex) as its majority shareholder – that connects businesses and researchers with a distributed crowd of workers worldwide to complete micro-tasks such as data labeling, image recognition, and content moderation.
Toloka's revenue depends on client request fulfillment performed by workers. Two variables are critical for margins: the number of tasks completed per day, and how long a worker stays active on the platform.
As Senior Marketing Manager, B2C, I was responsible for acquiring, engaging, and retaining workers.
The goal was to increase Toloka's margins by optimizing the worker acquisition budget – specifically, reducing acquisition costs without sacrificing revenue.
I focused on three areas to achieve this:
This case covers area 1. Areas 2 and 3 are covered in separate cases.
To enable ROI-based acquisition decisions, I needed proxy metrics that could predict worker quality early – before full performance data was available.
We defined worker quality along two dimensions: relevance (acceptance rate of completed tasks) and productivity (number of tasks completed while active). The goal was to pay more for high-quality workers and less for low-quality ones, rather than applying a flat acquisition cost across all cohorts.
Working with the Analytics team, we identified money-per-hour (mph) as the best efficiency metric – the average income a worker earns per hour, factoring in time spent on tasks and access to higher-paying work. The higher the mph, the higher the ROI.
To put this into practice, I mapped average acquisition cost against average mph for each cohort, segmented by language and country. This gave us a ranking system to reallocate budget toward the most efficient segments.
To execute the project, I aligned the User Acquisition, Analytics, and Attribution teams around a shared hypothesis:
If we acquire workers based on the mph proxy metric, we can pay more for high-efficiency workers and less for low-efficiency ones – reducing the acquisition budget by at least 30% while holding revenue flat.
| Metric | Target | Result | How measured |
|---|---|---|---|
| Worker acquisition budget | ≥ 30% reduction | −44% | Budget against revenue; revenue held flat |
| Share of relevant, high-quality workers | – | +21% | Cohort quality after the switch from fixed to variable acquisition cost |
| Cohort ranking metric | – | Money-per-hour (mph) | Average income per active hour, mapped against acquisition cost by language and country |
Table 3. Targets are the hypothesis agreed with the User Acquisition, Analytics and Attribution teams before the work; results are as stated in this case.
"Maria achieved great results in the acquisition, retention, and engagement of ML data makers thanks to her strategic thinking, data analysis skills, project management skills, understanding of marketing tools and technologies, strong work ethic and attention to detail, and continuous improvement mindset."
– Dmitry Stepanov, Founder & GP at AAL VC · Forbes 30 Under 30 · Senior to Maria at Toloka