Picking an Ozon niche and a first product: what the analytics actually tell you

This comes from nine Russian-language videos on niche and product selection. Two are from Ozon’s own training channel — the category comparison report, and a webinar with the Sell Monitor analytics service — the rest are independent sellers describing their own practice. These are not Ozon’s official documents, and the figures belong to 2024–2025. Tariffs, subscription tiers and dashboard screens change: treat thresholds as reference points, not rules.

The shared starting point: arithmetic, not instinct

All nine agree on what not to do. Ivan Babalo says the sellers who come to him with no sales explain their choice with “I just felt like it”; his counter is that product selection is arithmetic at school level, and the hard part is knowing which numbers to pull. Farit Garifullin tells the same story: at the start they asked traders at the Lyublino wholesale market what moved in volume, bought 50-unit test batches, then struggled with what they had bought. “I do not recommend this method.” The webinar adds that there is no ideal niche today — collect 5–15 candidates rather than one perfect category, because unit economics will kill most.

The screens sellers actually open

Third-party tools named: MPStats, MoneyPlace, Mayak, Sell Monitor, SalesFinder. Two caveats — Roma Raid notes that external analytics are built on FBO, so sellers shipping otherwise (Chinese sellers on Ozon Global) are often invisible to them; and Vyacheslav Malykh stresses their yearly figure is ordered revenue, not money collected: “a compass.”

Criteria and the thresholds they name

On certification, to be honest: none of the nine treats certificates or declarations as a screening criterion. The only direct mention is Garifullin’s, on earphones — yes, a declaration or certificate is needed, “but that is a solvable question.” Work to be done, not a veto; no cost or lead time is given anywhere.

Red flags

• Judging on 28 days. In Ozon’s own video, belt-bag revenue is down 19% for the month and the seller is ready to drop the product. Over a year it is up 117%: low from August to January, then a sharp climb. The call — do not exit, ship less now, increase supply in January.
• A long-term slide in the average price. A red flag outright in the webinar: if a category’s average price erodes year over year, unit economics that work today fail in six months, because you get forced down to competitor prices. On Malykh’s chart it is where the units curve separates from the revenue curve.
• More listings than sales, and niches needing hundreds of SKUs. The webinar cites a shop whose revenue is spread across 1,500 listings — a newcomer cannot build that assortment.
• Reading a competitor’s stock as proof of demand. The Q&A answer is flatly no: a snapshot says nothing. Watch the stock chart over time — restocked, how fast it drained, restocked again.
• Entering a season late. Barbecue sales climb from early April, so the product must be live by the end of March — the search starts two to three months earlier. Arrive late and competitors already hold reviews and organic traffic.
• Following ready-made “best niches” lists. Malykh argues a niche is good or bad only relative to your goal. Notably, Roma Raid attaches the same warning to his own list video: do not buy what I showed — first unit economics, then competitor analysis, then sourcing in China.

The order of steps they recommend

Three recipes, different sequences, the same ending.

1. Ozon webinar with Sell Monitor: first decide the product type — steady demand, seasonal, trend-driven, digital. Then three steps: search queries (is there demand), categories (supply, monopoly, seasonality, price distribution), products (competitor cards, which variant sells). Write what you rule out into a brief; collect 5–15 niches and let unit economics cut.
2. Dasha Syomina, on SalesFinder: rank categories by revenue, filter (revenue under 15 million ₽, under 3,000 products, average price 500–900 ₽, under 200 sellers), drill to the lowest sub-category and find the narrow sub-product inside it. Her example: a phone-parts category holds 1,707 products, but only 80 cards have “glue” in the name — the real competitor count. Each candidate goes into a sheet, then unit economics.
3. Vyacheslav Malykh (2025 guide; demos mostly on another marketplace): three passes. (a) Switch your brain off and collect five numbers per niche — yearly capitalisation, two buy-out rates, revenue and units of the top 100 — then score and discard the weak. (b) On the survivors: seasonality, dominators, trend dynamics. (c) The tenth criterion, your own interest or edge. He searches top-down from a category vector and bottom-up from a product to its sibling “neighbours”.

One newcomer test can be added on top. Roma Raid filters a niche to cards with under 30 reviews and checks whether they earn: electric sharpeners return 1 million ₽ last month with 63 of 88 items selling; e-scooter spare parts, 2.6 million ₽ across 535 items. If newcomers are earning, the niche is open to them.

In short

• Choose on numbers, not instinct — the one method all nine reject is buying on a trader’s tip.
• The free category comparison report covers demand, dynamics, top-5 share, price segments and FBO/FBS in one screen.
• Never judge on a short window: −19% over 28 days can be +117% over a year.
• Health test: sales outnumber listings, the leader’s share is low, revenue declines smoothly.
• Oversized, fragile and very cheap goods, plus clothing, are ruled out as a first product.
• List seasonal goods one to two months before the season; search two to three months ahead.
• The last word belongs to unit economics — most of the 10–15 collected niches die there.
• Certification burden is not treated as a measured criterion in any of them.