Case study: a women's handbag online store — ROAS up 2.3× in 3 months
A Ukrainian eco-leather accessories brand (StarBags) was burning its ad budget almost to zero: the ads broke even at ×2.1, orders came in expensive, and there was no end-to-end analytics at all. We rebuilt the tracking and the campaign structure — and in three months ROAS rose from ×2.1 to ×4.8, while the cost per order nearly halved, with no budget increase. Here's exactly how.
The situation "before"
An online store selling women's handbags, totes and belt bags made from eco-leather. Their own production in Ukraine since 2020, on the market since 2008, with over 10 000 buyers a month. Advertising was already running, but it hovered at the edge of profitability — every dollar invested barely came back.
| Niche | E-commerce · women's eco-leather accessories (handbags, totes, belt bags) |
|---|---|
| Geo | Ukraine |
| Period | February – April 2026 (3 months) |
| Budget | $1800–2000 / mo |
| Channels | Meta Ads (Facebook/Instagram), Google Ads |
| Task | Lift ROAS and cut the cost per order with no budget increase |
Problems at the start
- ROAS held at ×2.1–2.3 — at the brand's margin the business was breaking even or running a small loss.
- High cost per order (CPA): the budget was going to non-targeted traffic.
- No end-to-end analytics — it was unclear which channels actually drove sales.
- The Meta and Google algorithms "couldn't see" purchases because of misconfigured events.
The situation "after"
Three months on, the ads stopped just breaking even: every dollar invested started returning nearly $5 in revenue (ROAS ×4.8). Instead of "feast or famine" — a predictable flow of orders at a clear price, and the owner can finally see where every hryvnia of the budget goes. Most importantly: the budget didn't grow in the process — only the efficiency of how it was spent changed.
What we did in plain words
No technical details — here are the four steps that lifted the return, and what each one gave the business.
1. First — honest analytics (CAPI)
We started with the essentials — letting the algorithms see real purchases. We set up server-to-server delivery of order events (Conversions API for Meta and Google via GTM), in real time, with deduplication — no losses or double counts. Simply because the algorithms received the right data, the visible efficiency of the ads grew by +15–20%.
2. We rebuilt the campaign structure
Before us, everything sat in a single campaign — and the budget was spent inefficiently. We split it: "cold" traffic (brand discovery) separated from "hot" retargeting, a split by category (everyday handbags, evening models, totes and belt bags), and A/B creative tests. On Google — Performance Max, search on brand queries and Shopping ads.
3. We recovered abandoned carts
We launched dynamic retargeting: anyone who added an item to the cart but didn't buy saw those exact handbags for the next 7–14 days — along with a separate discount offer. Abandoned carts are the most underrated source of sales: thanks to this, the share of repeat sales grew to 18% of turnover.
4. Creatives that sell
Instead of "grey" product cards we moved to video (Reels and Shorts — unboxing, in-motion reviews, styling with outfits), UGC content (photos of real buyers) and seasonal time-limited offers. That more than doubled the click-through rate.
Under the hood — Conversions API and transaction deduplication, a product feed in Merchant Center and passing order statuses from the CRM back to the algorithms, so the ads learn from real buyers. I'll gladly walk you through the technical details on a call.
Results over 3 months
| Metric | Before | After | Change |
|---|---|---|---|
| ROAS (return on ad spend) | ×2.1 | ×4.8 | ×2,3 |
| Cost per order (CPA) | $34 | $14 | −59% |
| Cost per 1000 impressions (CPM) | ~$14 | ~$11 | −21% |
| Click-through rate (CTR) | 1.2% | 2.8% | +133% |
| Cart → order | 35% | 52% | +49% |
| Average order value (AOV) | ~$71 | ~$67 | stable ~$70 |
An important detail: the average order value stayed stable (~$70). In other words, the higher return came from the efficiency of the ads, not from inflating product prices — the business earns more from the same range and the same budget.
| Month | Budget | ROAS | CPA | AOV |
|---|---|---|---|---|
| February (start) | $1800 | ×2.1 | $34 | ~$71 |
| March (CAPI) | $1900 | ×3.4 | $22 | ~$75 |
| April (optimization) | $2000 | ×4.8 | $14 | ~$67 |
How ROAS grew
What it gave the business
For the owner
The ads stopped just breaking even — every dollar invested brings nearly $5 in revenue. Instead of chaotic spikes — a predictable flow of orders. And full transparency: you can see where every hryvnia of the budget goes and which channel actually drives sales.
For the team
The sales team is unburdened — the inquiries are now targeted and "hot". A path to scaling has opened up: as the budget grows, ROAS doesn't drop, because the system is trained on real buyers, not "browsers".
Tools and technology
- Meta Ads (Facebook/Instagram) — Conversions API, Advantage+ Shopping.
- Google Ads — Performance Max, search, Shopping ads.
- Google Tag Manager + GA4 — end-to-end analytics and transaction delivery.
- Google Merchant Center — product feed.
- CRM integration — passing order statuses back to the algorithms.
Takeaways
The main growth levers for an accessories online store come down to three things. Correct analytics: without CAPI and GA4 the algorithms work blind — you pay for traffic, not for buyers. The right structure: a single campaign for everything spends the budget inefficiently, while a split by category and funnel stage lifts ROAS several times over. Cart retargeting: abandoned carts are the most underrated source of sales. At a budget from $1000/mo this approach pays off as early as the 2nd–3rd week and steadily holds ×4+ ROAS in the fashion-retail niche.