AI agent vs manual Ozon ads: 6.7% ACOS
We connected our own AI agent to a partner store's Ozon advertising and ran it alongside campaigns managed by experienced specialists on the same products. Over the first 8 days the agent held ad cost share (ACOS) at 6.7% versus 10.1% for manual campaigns, on a 3.4x smaller budget.
The task
We wanted to test whether an AI agent could manage Ozon advertising more efficiently than an experienced specialist while staying within a set budget and target ad cost share (ACOS).
What we did
- Set the agent clear goals and limits: target and maximum ACOS, daily, weekly and total test budgets, and the products to advertise
- Let the agent review campaign stats, sales and stock several times an hour and adjust bids and budgets around the clock
- Gave new products and ad formats small test budgets, scaling what sold and switching off what did not
- Built independent safeguards: hard spend ceilings, a separate watchdog that can stop ads, self-shutdown, a stock-based pause and a one-click stop button
- Ran manual campaigns by experienced specialists on the same products over the same period for a fair comparison
Results
| Manual campaigns | AI agent | |
|---|---|---|
| Ad cost share (ACOS) | 10.1% | 6.7% |
| Cost per order | $4.60 | $2.90 |
| Cost per click | $0.17 | $0.06 |
| Ad spend | $670 | $200 |
| Sales from ads | $6,650 | $2,920 |
| Orders from ads | 145 | 67 |
The agent ran on a 3.4x smaller budget. At the agent's ACOS, the same sales volume would have cost about $450 in ads instead of $670, roughly $220 saved in 8 days on one product line.
Takeaways
- New Ozon campaigns need a warm-up: the agent's ACOS fell from 21% on day one to 3% on day five, so do not judge a campaign by a single bad day
- More budget does not mean more orders: each product has a demand ceiling, so scale only while ACOS stays within target
- Test new products and ad formats with small budgets: the agent spent about $50 to find out what did not work instead of a month of budget
- An agent cannot fix a weak product: results still depend on the product card, price and stock, but the agent stops ads from eating profit
Amounts are shown in US dollars, converted at a fixed rate.
More case studies
How we run our agency with AI agents
A group of AI agents runs our website and six YouTube channels: finds news, selects what matters, writes, fact-checks, voices, edits, publishes and reports. The same work done by hand would cost $5,500-$6,300 a month.
Read the case study →
Moving a 700-page website from Tilda to our own server
We moved more than 700 pages and 650 blog articles from a site builder to a private server, kept every address, replaced forms and payment with our own, and made automatic content publishing possible.
Read the case study →
How SEO became our cheapest source of customers
We built a steady flow of customers from search engines for our own agency. SEO now pays back all the investment and brings customers 1.5-2 times cheaper than paid and social ads, with 1-2 hours a month of management time.
Read the case study →Want results like this?
Request a free consultation. Tell us about your project and we will say honestly what we can do.
- Free, no obligation
- Reply within one business day
- Straight answer, even if the answer is "you do not need us"