Optimized Targeting

Optimized Targeting works by taking the audiences you have explicitly selected in an ad group - whether that is a remarketing list, an in-market segment, a custom segment, or a combination - and using them purely as a starting seed rather than a hard boundary. From that seed, plus live conversion signal flowing back from your Google Ads tag, the system looks for additional users outside your selected audiences who share behavioral patterns with the people who are actually converting, and begins showing them ads too. It is switched on per ad group with a single toggle, sitting inside Display, Demand Gen, and Video campaign types, and Google has been steadily nudging it toward being the default-on setting in newer ad group flows.

It is explicitly the intended successor to Similar Segments (formerly Similar Audiences), which Google sunset in 2023 because it worked from static list-similarity math rather than live conversion feedback. Optimized Targeting is meant to do the same underlying job - find people who look like your best audience but aren't on your list yet - using a more direct, conversion-driven method instead of a periodically refreshed similarity score.

The setting it depends on most heavily is conversion tracking quality, full stop. Optimized Targeting has nothing useful to learn from if the conversion action it is optimizing toward is misconfigured, delayed, or too sparse - a campaign with only a handful of conversions a month gives the system almost no signal to identify who actually converts, and it will end up guessing broadly instead of targeting precisely. This is the same conversion-volume dependency that governs Smart Bidding generally, and the two settings compound: Optimized Targeting under Maximize Conversions or Target CPA bidding, fed by clean, sufficient conversion data, is a genuinely different product than the same toggle switched on in an account with thin or noisy conversion tracking.

It matters most on a campaign that already has a reasonably well-performing, clearly defined seed audience and healthy conversion volume, where the goal is incremental reach beyond that list without hand-building yet another custom segment. It matters least, or actively backfires, in an account with unreliable conversion tracking, since the practitioner ends up unable to tell whether a performance dip came from the expanded reach or from the underlying measurement problem the toggle inherited. The recurring practitioner mistake is confusing Optimized Targeting with straightforward audience expansion - a setting or mental model that simply widens reach for its own sake. Optimized Targeting is meant to widen reach specifically toward people likely to convert, using your conversion goal as the filter; if you switch it on without a working conversion goal attached to the campaign, you get expansion without the optimization half of the name doing anything.

In reporting, judge Optimized Targeting by watching what happens to cost per conversion and conversion volume together after toggling it on, not by watching reach or impression volume alone, since a reach increase with a flat or worsening cost per conversion means the expansion is not finding genuinely better-fit users. Give it a real evaluation window - at least a couple of weeks of stable spend - before judging, since like other machine-learning-driven settings it needs time and volume to calibrate, and a same-day before/after comparison usually just reflects normal day-to-day noise rather than the setting's actual effect.