How to Set a First Target ROAS Without Guessing a Number

Your starting tROAS is not a goal you invent, it is a number you read off your own conversion history and then loosen on purpose.

· QueryCut

The short version

  • Set your first Target ROAS at or slightly below your trailing achieved ROAS, not at your profit-math goal.
  • Treat the target as a throttle on the auction, not a thermostat: higher target removes volume, it does not improve a fixed pool.
  • Wait for roughly 30 to 50 conversions of value history before trusting an achieved ratio, and exclude sale periods from the window.
  • A high target that shrinks the account while raising the ratio is a loss disguised as a win; judge it on total conversion value.
  • After launch, leave the target alone for 1 to 2 weeks, then tighten in 10 to 15 percent steps with a week between each.

To set a first Target ROAS, do not pick the number your business wishes it hit. Pull the ROAS the campaign actually achieved over the last 30 to 60 days under Maximize Conversion Value, and set your first target at or slightly below that achieved figure. The target is a constraint on the bidding algorithm, not a profit aspiration, and starting it above what you have proven you can do is the single most common way to choke a campaign on day one.

The answer: your first target is your trailing achieved ROAS, minus a buffer

Run the campaign on Maximize Conversion Value (no target) for at least 30 days, or long enough to collect roughly 30 conversions. Then go to the campaign, open the date range to the last 30 to 60 days, and divide Conv. value by Cost. That ratio is your achieved ROAS. If you spent $4,000 and generated $12,000 in conversion value, your achieved ROAS is 3.0, or 300 percent.

Your first Target ROAS should be that achieved number, or about 10 to 15 percent below it. If achieved is 3.0, set your first target at 2.6 to 3.0. You are not trying to improve efficiency in week one. You are handing the algorithm a target it already knows how to hit, so the transition from volume-maximizing to value-maximizing does not crater your traffic.

The mistake almost everyone makes is reaching for the profit-math number instead. You calculate that your margins require a 5.0 ROAS to be profitable, so you type in 500 percent. If the campaign has only ever delivered 3.0, you have just told Smart Bidding to find a world that does not exist in its data, and it will respond by bidding down hard on nearly every auction to protect the ratio.

Why the target is a throttle, not a thermostat

Target ROAS works by predicting, auction by auction, the conversion value a given click is likely to return, then bidding so that predicted value divided by predicted cost lands near your target. Raise the target and you raise the bar a query must clear to get a bid at all. Queries the system predicts will return below-target value get bid down or dropped entirely.

This is why a target set too high does not gently nudge efficiency up. It removes the entire bottom and middle of your eligible auction pool. Impressions fall, clicks fall, and conversions fall faster than spend, because you have amputated the volume that was subsidizing your average. The achieved ROAS on the survivors may even look great, which fools people into thinking the high target worked. It did not. It just shrank the account.

Understanding this lets you predict behavior in cases nobody described to you. Double the target and expect roughly a step-change drop in volume, not a smooth curve. Set a target below achieved and expect the opposite: the system loosens, chases more marginal auctions, and volume climbs while efficiency drifts down toward your target. The target is the floor you let the algorithm fall to, not the ceiling you push it toward.

The conditions where this pays, and the data you actually need

This approach pays when three things are true. First, the campaign has real conversion-value history, meaning you are importing actual revenue or value figures, not a flat $1 per conversion (if every conversion is worth the same, tROAS degenerates into Target CPA and you should just use tCPA). Second, you have enough conversions that the achieved ratio is not one lucky order. Third, your conversion tracking is stable and not mid-change.

On volume: wait for roughly 30 conversions in the measurement window before trusting an achieved ROAS, and closer to 50 if your order values are spiky (a handful of large B2B deals can make a 30-conversion average meaningless). The reasoning is simple. Below that, a single unusually large or small conversion swings the ratio enough to set your target in the wrong place, and the cost of waiting another two weeks is almost always smaller than the cost of launching a target that strangles the campaign.

A cleaner alternative to eyeballing the average: Google Ads often shows a recommended target when you switch bid strategies, and the Bid Strategy Report (Campaigns > the bid strategy column > View bid strategy report) shows achieved ROAS over time. Use these to sanity-check your own division, but trust your own trailing math over the suggestion if the suggestion is built on a different date range than you intend to run.

Where it backfires: the traps that look like good decisions

Trap one: setting the target off a window that includes a sale or seasonal spike. If Black Friday is in your 60-day lookback, your achieved ROAS is inflated by demand you will not see again, and a target anchored to it will suffocate January traffic. Exclude obvious anomaly periods before you compute the baseline.

Trap two: thin conversion data dressed up as a number. If you had 11 conversions last month, the achieved ROAS is a coin flip wearing a decimal point. Setting a target off it is still guessing, you have just laundered the guess through a division. In that case, stay on Maximize Conversion Value longer, or run Target CPA until value data accumulates.

Trap three: the high target that looks like it worked. You set tROAS to 6.0 against a 3.0 history, volume collapses 70 percent, and the remaining conversions average 6.5 ROAS. The dashboard says you won. You did not. You are now paying fixed account overhead and labor against a fraction of the revenue. Judge a tROAS change on total conversion value at an acceptable efficiency, not on the ratio alone. If value dropped more than spend, the target is too tight.

The first two weeks after you flip it on

Expect a learning period of roughly one to two weeks or 30 to 40 conversions, whichever is longer, during which performance is noisy and not representative. Do not touch the target during this window. Every change restarts the learning the algorithm is doing to calibrate its value predictions, and a jumpy manager who adjusts twice a week never lets any target prove itself.

After the learning period, read two things: did total conversion value hold, and did achieved ROAS land near your target? If value held and ROAS is near target, you have a working baseline. Now you can tighten. Move the target up in increments of about 10 to 15 percent, wait a full week per step, and watch whether volume falls faster than efficiency rises. The moment total value starts dropping meaningfully, you have found the ceiling, so step back down one notch.

One thing you genuinely cannot know in advance: how much headroom exists between your achieved ROAS and your true profit-maximizing target. Google does not expose the shape of your value-vs-volume curve. You discover it by walking the target up in small steps and watching, which is why starting at a known-safe baseline matters so much. You cannot climb a staircase you blew up on the first step. Cleaning out junk search terms before you start (Campaigns > Insights and reports > Search terms) also helps, because wasted spend on irrelevant queries drags your achieved baseline down and makes every target you set harder to hit.

Questions people also ask

What if my campaign has never run on a value-based strategy?

Start on Maximize Conversion Value with no target and let it run until you have around 30 conversions of real value data. Then compute achieved ROAS from that period and use it as your first target. Launching tROAS cold with no value history means the algorithm has nothing to calibrate against, and your first target is a pure guess.

Should I use the Target ROAS number Google recommends when I switch strategies?

Use it as a sanity check, not gospel. Google's suggestion is built on its own date range and assumptions, which may not match the window you intend to run. If your own trailing division disagrees with it, trust your math, especially if the suggestion was inflated by a recent spike you want to exclude.

How is Target ROAS different from Target CPA if all my conversions are worth the same?

If every conversion carries the same value, tROAS collapses into tCPA mathematically, because value divided by cost reduces to a fixed number over cost. In that case just use Target CPA, it is simpler to reason about. Target ROAS only earns its keep when conversion values genuinely vary, like ecommerce order totals.

How high can I eventually push the target?

You find out by walking it up, not by calculating it. Raise the target in 10 to 15 percent steps, wait a week each time, and watch total conversion value. When value starts dropping faster than efficiency improves, you have hit the practical ceiling for your current demand and should step back one notch.

My ROAS looks great but conversions dropped after I raised the target. Is that good?

Usually no. A rising ratio with falling volume often means you amputated the mid-tier auctions that were still profitable in aggregate. Compare total conversion value before and after: if value fell more than spend, the target is too tight and you are leaving profitable volume on the table.

More from the blog

Landing Page Experience: The Changes That Actually Move ItImpression Share Lost to Rank vs Budget: What Each One MeansHow Many Campaigns Is Too Many for the Budget You Have?Primary vs Secondary Conversion Actions: What Each One Actually Changes