Meesho ads ROI model: set a target ROI, let it spend.

The ROI model asks you for one number instead of a hundred bids. That is a real simplification — but the one number carries all the weight, so it is worth getting right.

How the ROI model actually works

Under the old system, running Meesho ads meant behaving like a bid manager: you picked keywords, chose a cost per click or per thousand impressions, and constantly nudged those bids up and down to chase visibility without overpaying. It rewarded time and skill, which quietly favoured larger sellers who could afford a dedicated ad person. The ROI-target model removes that machinery. You declare a target ROI — the revenue you want back for every rupee of ad spend — and set a daily budget ceiling. From there, Meesho’s system takes over the spending decision. It decides how much of your ceiling to actually use, which placements and audiences to buy, and when, all judged against whether the resulting orders are likely to clear the ROI you asked for.

The reason this works is data. A single seller sees only their own listings and a sliver of search behaviour. Meesho’s system sees conversion patterns across the whole marketplace — which placements convert for which price bands, at which times, for which audiences. When you hand it a target ROI, it is effectively using that far larger picture to place your money better than manual keyword bids usually could. This is the honest case for the new model: for most sellers, especially small ones, the system will allocate spend more efficiently than they would by hand, because it is optimising against information they simply do not have.

How to choose and set your target ROI

The single most common mistake is treating the target as a wish rather than a constraint. Sellers type in a high number — “I want eight rupees back for every one I spend” — and then wonder why nothing happens. The target ROI is not a goal you are hoping the system reaches; it is the efficiency bar every placement must clear before the system is willing to buy it. Set it too high and you have effectively told the system that almost no placement is good enough, so it spends nothing. The right anchor is your own economics: roughly, the return that keeps an ad-driven order profitable after your product cost, shipping and expected returns. If you keep about a third of an order’s value as contribution margin, then a target around that break-even multiple is a reasonable starting point — profitable if you clear it, but low enough that the system is actually allowed to work.

In practice, start at or near the suggested default Meesho offers for your category, give the campaign a budget ceiling large enough to gather real order data, and then leave it alone long enough to learn. Read the results over a week, not an afternoon, because daily numbers are noisy and will tempt you into over-correcting. When you do adjust, move the target in small steps: nudge it down a little if spend is stuck at zero, nudge it up a little if the orders coming in are too thin on margin. Big swings whipsaw the system and reset its learning; small, patient moves let it settle into the working band. The full Meesho ads guide walks through campaign setup, creatives and product selection around this target.

How to read spend, revenue and realised ROI

Once a campaign is running, three numbers matter and they only mean something together. The first is spend — how much of your ceiling the system actually used. The second is attributed revenue — the sales the system credits to those ads. The third is realised ROI, which is simply revenue divided by spend, and is the figure the dashboard leads with. The trap is stopping there. That headline ROI is a gross, pre-return number. It does not know that a slice of those ad-driven orders will come back as returns or RTO, and it does not net out the ad deductions that land in your settlement. Your true ROI — the money you actually keep — is always lower, sometimes dramatically so in high-return categories like fashion. A campaign showing a comfortable dashboard ROI can be running at break-even or worse once reality is counted, and you would never see it from the ads panel alone.

So the discipline is to read the dashboard ROI as a signal, then confirm it against your settlement. Subtract the returns and RTO on the ad-attributed orders, subtract the ad deductions Meesho charged, and only then judge whether the target is set right. If true ROI is comfortably above break-even, you may have room to lower the target and buy more volume. If true ROI is at or below break-even while the dashboard looked fine, the target is too low and the system is buying orders that lose money after returns. This gap between the pretty number and the kept number is exactly where sellers quietly lose margin on ads.

The two traps: target too high, target too low

The two failure modes sit on opposite ends of the same dial. Set the target ROI too high and the system spends nothing: you demanded a return no placement can realistically deliver at your price and conversion rate, so the model, correctly, refuses to buy. Sellers often misread this as the campaign being broken, when it is doing exactly what it was told. The fix is to lower the target in steps until spend unlocks. Set the target too low and the opposite happens: almost every placement clears the easy bar, the system spends freely, and a chunk of that spend buys orders that were barely profitable or outright unprofitable once returns and deductions are counted. This one is more dangerous because it hides — the campaign looks busy and successful on the dashboard while your real, post-return ROI sinks. The working setting lives in the band between the two, and finding it is the whole job.

Sources & further reading

Meesho updates its ad products and their defaults over time; confirm the current campaign options and suggested ROI against the official supplier tools.

• Meesho Supplier Hub — ads & campaign tools
• Meesho Supplier Hub — seller dashboard and settlement
• The Media Ant — background on ROI and ad-efficiency models