Broad Match

Mechanically, broad match works by running your keyword through Google's query-understanding layer rather than matching text. Google looks at the words in the keyword, the ad copy and landing page in that ad group, the other keywords sharing the group, and signals from your account history (what has converted before, what audiences have engaged with your ads) to decide whether a given search qualifies. That means the same broad keyword can pull in a different mix of queries in two different accounts, or even in the same account after you swap the landing page, because the model is reading context, not just the string you typed.

The setting it interacts with most is bid strategy. Broad match on Manual CPC is a bad combination for most accounts because a fixed bid has no way to pay less for a loosely related query and more for a tightly related one - you are paying the same price whether Google guessed well or poorly. Broad match earns its keep under Smart Bidding (Maximize Conversions, Target CPA, Target ROAS) because the algorithm can see the same conversion signals it used to widen the match and can bid each auction individually, cutting the bid on weak-fit queries automatically. This is also why Google's own guidance nudges advertisers toward broad plus Smart Bidding as a pair rather than recommending either alone.

It matters most when an account has enough conversion volume and a stable enough offer that the bidding algorithm has something to learn from - a campaign converting only a handful of times a month gives Smart Bidding almost nothing to work with, while one converting steadily every day gives it a real signal to price auctions against. It matters least, and is actively risky, on a brand-new account with no conversion history, on a campaign with a hard budget cap where a few wasted clicks meaningfully dent daily spend, or on a niche B2B offer where an irrelevant click and a qualified click can both look plausible to the matching model but have wildly different value to the business.

The recurring mistake is treating broad match as a volume lever without treating negative keywords as its companion, not an afterthought. Because broad match takes into account the whole ad group and account context, a broad keyword can drift over time as you add new keywords or conversions accumulate in a new direction - what performed cleanly for two months can start pulling in adjacent queries after account history shifts. The fix is not to abandon broad match but to check the search terms report on a schedule, not just when something looks wrong.

In reporting, broad match keywords should be read alongside the search terms report, never on their own. A broad keyword's own row in the keyword report tells you aggregate CPC, conversions, and cost, but it hides which specific queries drove those numbers. Segment by search term (or open the report directly) to see whether spend is concentrated in a handful of close variants or spread thin across dozens of loosely related queries - the latter is the signal that your negative list needs work or that the ad group's landing page or other keywords are giving Google too loose a hint.