Setting Target ROAS Without Guessing

Target ROAS is a Google Ads bid strategy in which you tell Google the return on ad spend — conversion value divided by what you paid for it — that you want across a campaign, and Google raises or lowers each bid to hit it. The setting is simple. The number is where accounts go wrong, because the history most owners base it on is flattered by three things the dashboard does not separate out.

Google’s own About Target ROAS bidding page says to set the target from your business goals, with historical ROAS as a reference. The guides ranking for this say roughly the same, and one suggests going slightly above your historical average. None of them asks whether the historical average is true. In a store account it usually is not, and the fix takes about fifteen minutes. If you would rather hand the account over, this is how I run Google Ads for ecommerce brands.

What is target ROAS?

It is an instruction to Google’s automated bidding to chase conversion value at a set ratio to cost, rather than chase clicks or conversions. Google writes it as a percentage: $5 of sales for every $1 of spend is a 500% target, which is the same thing as a ROAS of 5.0.

That percentage format trips people up more than it should. A store with a break-even ROAS of 2.5 needs to type 250%, not 2.5% and not 25%. Google also expects enough data before it can bid this way: the same help page lists at least 15 conversions in the past 30 days for Search and Shopping campaigns. Below that, the strategy is guessing on your behalf.

What should my target ROAS be?

Somewhere between a floor set by your margin and a ceiling set by what your account has genuinely returned without help from brand searches, returns or late conversions. Start close to the ceiling, then move towards the floor only if you want more volume.

Everything in the rest of this post is about getting those two numbers right, because the setting itself is a single field.

The floor: what your margin allows

Your floor is one divided by your gross margin as a decimal, then netted for returns. At a 40% margin the floor is 2.5, entered as 250%.

Gross margin means what is left of a sale after the product, shipping, packaging and card fees. The derivation, with a table for margins from 25% to 70%, is in is Google Ads worth it for a Shopify store. The adjustment that page does not dwell on is returns. Google counts the order value at checkout. If one order in ten comes back for a refund, the revenue behind every reported ROAS is 10% lower than it looks, so divide your floor by 0.9. A 2.5 floor becomes 2.78, entered as 278%. Skip that step and you set a target that breaks even on paper and loses money in the bank.

The ceiling: what your account has actually done

Your ceiling is the ROAS your non-brand campaigns returned over a period whose conversions have finished arriving. Three things inflate the headline number, and each one has to come out.

Branded search

People who type your brand name convert far better than strangers, and in most store accounts brand campaigns sit inside the same average. Suppose brand spent $1,000 and returned $12,000, while everything else spent $4,000 and returned $13,000. The account reports $25,000 on $5,000, a ROAS of 5.0. The non-brand half, the part actually finding new customers, returned 3.25. Set a 500% target on a non-brand campaign and you have asked it for a return it has never achieved. If you run Performance Max, the same trap applies inside one campaign; more on that in Performance Max for ecommerce.

Returns

The same 10% correction as the floor, applied the other way: a clean 3.25 with one order in ten refunded is really about 2.9. Apply it to both numbers or neither, or the gap between them is an illusion.

Conversion lag

Orders that happen days after the click are credited back to the day of the click, so the most recent week always looks worse than it will once they arrive. Leave the latest days out of the window you read, or you will lower a target that was working. How long to leave depends on how long your customers take to buy; the time lag reports in the account will show it.

Why did my sales drop when I set a target ROAS?

Because the target was above what the campaign had really been achieving, so Google stopped bidding on everything except the surest auctions. Google’s documentation says it plainly: a target that is too high may limit the traffic your ads get.

This is the most common call I get about target ROAS, and it is almost always the brand problem from the previous section. The owner looked at a blended 5.0, typed 500%, and the non-brand campaigns, which had been living at 3.25, went quiet. Lower the target to the clean ceiling and the volume usually comes back. Then move it in small steps, one change at a time, and give each change a couple of weeks before judging it.

Two tools help here. A portfolio bid strategy lets several small campaigns share one target and pool their conversion data, which matters when no single campaign reaches the minimum. And for a short sale, seasonality adjustments tell Smart Bidding to expect a jump in conversion rate; Google says they suit events of one to seven days and may not work as well beyond 14. Use them instead of rewriting the target for a weekend.

What if the floor is above the ceiling?

Then no target ROAS will make the campaign profitable, and the problem is not the bid strategy. If your margin needs 2.8 and the clean non-brand history is 2.2, no target satisfies both.

You have four honest options, and none of them lives in the bidding settings:

1. Raise the margin. Price, cost of goods, shipping thresholds or product mix. Each point of margin lowers the floor.
2. Raise order value. Bundles and sensible free-shipping thresholds lift conversion value without lifting cost per click.
3. Narrow the campaign. Cut it back to the products and queries that do clear the floor, and accept less volume.
4. Accept a first-order loss on purpose. Only if repeat-purchase data shows customers come back unprompted. Write the payback period down before you start, not after.

What does not work is setting the target above the ceiling and hoping. That just produces the volume collapse from the section above, at a ROAS that still does not cover the margin.

How do I know if it’s working?

It is working when clean non-brand ROAS holds above your floor and total gross profit rises. The ratio alone is not the test, because a higher ratio on half the revenue can mean less money.

Returns fall as spend rises in almost every account, and that is fine so long as the floor holds. At The Lifeguard Store, the blended return was 22.6x in year one and 14.5x in year two while monthly spend grew from $1,513 to $14,597. The ratio fell and the business was better for it, because the margin left room. The reverse also happens: at Petticoat Lane, spend was down 32% year over year at a better return. Neither direction is the goal. The target is a dial between volume and efficiency, and the margin tells you where the dial may go.

Measure it in money, not only in the ratio. The unit I use across channels is customer acquisition cost, which your accountant will recognise.

How long before I see anything?

A couple of weeks after each change, longer if your customers take days to decide. Google needs time to relearn the auctions after a new target, and the late conversions need time to land.

Judging a new target after three days is the most expensive habit in automated bidding. The first days after a change are noisy, and the conversions from them are not all in yet. Make one change, write down the date, and read the result once the lag window has passed. More on how I approach bidding and budgets across everything I write about Google Ads for ecommerce.

Questions people ask