Nine stores joined in August 2025. A year later, five are still running — here are their numbers

Ukrainian cases without the showcase: we take everyone who joined in a single month a year ago. Four improved their return, one stopped advertising. With methodology and honest caveats.

In short: what you’ll take away

Who it’s for: owners of Ukrainian stores deciding whether to hand product advertising to someone else.

What’s inside: a year of numbers from five stores — before and after, in the same season, with the calculation methodology.

The main point: four out of five improved ROAS. But there is no single “lever” that worked for all of them — and we’ll show why.

1. Why this article didn’t exist before

Every fourth person who talks to us asks the same question: “Do you have Ukrainian examples?” For seven months we answered in general terms. The reason is simple and awkward: showing a handful of pretty stories is a showcase, and nobody believes showcases. Especially in advertising, where everyone has seen decks with charts going up.

So we did it differently. We took one calendar month — August 2025 — and looked at everyone who joined then. Not the best ones. Everyone.

How many we were and how many are left

In August 2025, nine Ukrainian stores joined us. A year later, five are still running.

Two left right after the first month. One more after five months, another after nine. Why exactly they left — honestly: we have it recorded nowhere, so we won’t invent reasons.

Slightly more than half survive. That is the real picture, and we think it’s more honest to show it than to pick five winners out of two years of work.

2. How we calculated

The methodology matters more than the numbers themselves, because that’s where most of the pretty lying in advertising cases hides.

1. We compare the same season. May–July 2025 (before joining) against May–July 2026 (a year of work). E-commerce lives in seasons: compare December with February and you can “prove” anything. So the months are identical.

2. The baseline is almost three months before the start, not a week. A short baseline catches a random spike rather than the real level. We take 80 days before joining.

3. We drop the unfinished month. August 2026 hasn’t ended yet, so it isn’t included — otherwise we’d manufacture an artificial “decline”.

All amounts are rounded; store names and domains are not disclosed. Niches are given by subcategory so it’s clear what type of assortment we’re talking about.

3. Five stores over a year

Store	Catalogue	ROAS before → after	What’s visible
Car diagnostic equipment	400+	4.2 → 5.5	revenue +45% on the back of a ×1.7 order value
Smart home and video surveillance	850+	decline stopped	conversions ×1.8, click 14% cheaper
Access control systems	~9K	5.6 → 9.8	revenue ×2.7, order value ×2.2
Women’s clothing and vyshyvankas	~200K	6.5 → 9.7	advertising stopped from June 2026
Car accessories	~30K	4.4 → 5.9	revenue +81%, selling products ×1.9

Car diagnostic equipment: order value doubled

The smallest catalogue in the group — around 400 products. Spend grew moderately, while revenue for the same season rose by roughly 45%. The main change was the average order value: from ~860 to ~1,490 UAH.

An important detail that’s easy to hide: there were fewer orders — roughly 400 against 470 a year earlier, i.e. 16% fewer. Revenue was carried by order value, not by the number of sales. For the store that’s good (less handling for more money), but if what you need is growth in the number of orders — this example isn’t about you.

The catalogue structure barely changed: the share of budget going to products with no sales at all was around 7–8% and stayed there. There was nothing to clean up here — the store arrived already tidy.

Smart home: the most important thing is visible only in the trend

This is the case where a single number lies. Compare “before” and “after” head-on and ROAS appears to have dropped: 17.5 against 7.0. It looks like a failure — until you look at what was happening before us.

The trend before joining

May 2025 — ROAS 27.0. June — 13.1. July — 7.3. The metric fell three months in a row, halving every month. After joining it settled at 7–11 and has held there for a year.

So the “17.5” in the baseline is an average across months that included an anomalous May with an average order value six times higher than usual. Comparing against that is dishonest. What actually happened over the year: the number of orders nearly doubled (from ~380 to ~690 per season), the click got cheaper, from 5.3 to 4.5 UAH, and the share of budget on products with no sales shrank from 25% to 16%.

Stopping a decline is also a result — it just doesn’t look impressive on a slide.

Access control systems: the strongest growth and the biggest volatility

Revenue for the same season grew roughly threefold, ROAS from 5.6 to 9.8, average order value from ~4,050 to ~9,000 UAH. This is a B2B profile: few orders, a high amount on each.

That’s exactly why the swings here are the largest. There were months with ROAS of 20, 30 and even 65 — and that’s not a reason for pride but a feature of the niche: one large order moves the month’s figure. We show the season average, not the record month.

An honest fly in the ointment

Since February 2026 the number of orders at this store has been falling: 116 in February against 30 in July, at the same spend. ROAS is high thanks to order value, but the volume trend is going down. We see it and we’re looking into it — writing only the good parts about a case like this would be dishonest.

Women’s clothing: the largest catalogue and a stop

Around 200 thousand products — the largest assortment in the group. Over a year of work ROAS grew from 6.5 to 9.7, and May 2026 became the best month: revenue roughly 60% higher than May 2025.

And then, in June 2026, advertising practically stopped — spend fell more than tenfold and did not recover. Why — we don’t know; it isn’t in the data, and guesses here would be invention. But removing this store from the article would mean doing exactly what we promised not to do.

Car accessories: more products started selling

Revenue for the season grew by roughly 81%, ROAS from 4.4 to 5.9. The most interesting part is in the structure: the number of products that actually bring sales grew from ~250 to ~490, i.e. almost double. The store started earning from a wider part of its catalogue rather than from a few positions.

At the same time its share of budget on products without sales grew — from 63% to 78%. That looks contradictory, and we won’t pretend we have a ready explanation: as catalogue coverage expands, the share of those that haven’t sold yet inevitably grows too. The overall result is positive, but “we cleaned up the budget” is not what happened here.

4. What they have in common — and what they don’t

One thing is common: in four stores out of five the return on advertising grew over the year; in the fifth it grew too, for as long as it was running.

And now what you won’t see in most cases. There is no single mechanism that worked for all five.

In one, average order value grew while the catalogue structure didn’t change at all.
In the second, the main thing was a stopped decline and a cheaper click.
In the third, threefold revenue on a handful of large orders.
In the fourth, twice as many products that sell.

We deliberately tested one convenient hypothesis: does the share of budget going to products with no sales shrink during the first month? That would have made a nice common storyline. The data did not confirm it — in two stores the metric improved, in two it didn’t change, in one it got worse. That’s why we don’t tell that story.

What to take from this

If you’re promised one universal lever that will work specifically in your store — that’s a sale, not an analysis. Stores differ, and a year of work shows it plainly.

5. Three checks you can run yourself

You don’t have to hire anyone to see the state of your advertising. These three numbers can be calculated in your own Google Ads reports.

1. What share of budget goes to products with no sales at all? Take the product report for a year, sort by spend and look at those with zero conversions. Across our five stores this share ranged from 7% to 82%. If yours is over half — that’s the most expensive thing you can fix.

2. How many products actually sell? Not how many are in the catalogue, but how many brought at least one order over the year. Compare that with the number that received budget. The gap between those two numbers is your room to work.

3. Where has the metric been heading for the last three months? One number for one month says nothing. Three points in a row show direction — and direction matters more than the absolute value. The smart home store is an example of a decline visible only in the trend.

Compare identical months across different years. February against December will “show” a catastrophe even in a healthy store.

6. Summary

Nine stores started, five have been running for a year. Slightly more than half survive — that’s the real figure, not a showcase.
In four out of five the return grew — from +31% to +75% depending on the store.
There is no common mechanism. For one it was order value, for another a stopped decline, for a third a wider selling catalogue.
A single number lies. Without the pre-intervention trend, the smart home store’s “ROAS drop” would have looked like a failure, when it was in fact stabilisation.

🔍 See the same numbers for your own store

Shopping X-Ray connects to Google Ads in read-only mode and in 30 minutes shows your breakdown: how much budget goes to products with no sales, how many products actually earn, and where the metric is heading. Free, no commitments.

X-ray your catalogue →