Agency Echelon
Targeted Digital Advertising

Broad Match Is Spending Your Money on Questions You Never Asked

A magnifying glass resting on thick stacks of paper, high-contrast black and white

Google's recommended account structure gets simpler every year, and every simplification points the same direction: broad match everything, let smart bidding sort it out, stop looking at the details. Sometimes that genuinely works. Broad match paired with good conversion data can find query territory no keyword list would have imagined, and I have watched it do so. I have also opened the search terms report on accounts run this way and found a third of the budget answering questions nobody at the company would recognize as their business.

Both experiences are real, and the difference between them is never the match type. It is whether anyone was reading. The recommendation to go broad arrives bundled with an implicit second recommendation, stop supervising, and the second one is the dangerous one. Google's incentive here deserves naming plainly: broader matching expands the auctions your budget can enter, which expands Google's revenue, and the same interface that recommends broad match has been steadily burying the reporting you would use to check its work. The search terms report shows fewer queries than it did five years ago, hidden behind privacy thresholds that conveniently obscure exactly the low-quality matches you would object to. None of this makes broad match a scam. It makes it a tool sold by a counterparty, which is a category of tool you inspect.

The report is the discipline

The search terms report is the whole discipline. Not a monthly glance. A standing ritual with money attached. Every query in that report is a sentence a real person typed, and reading them in bulk tells you three things no dashboard summarizes: what the algorithm thinks you sell, where its guesses are drifting, and which drift is cheap exploration versus expensive delusion. I have found a staffing client's budget matching to queries about a similarly named software product, a healthcare client bidding on a celebrity's medical news, and a B2B client whose broad match had decided, with total confidence, that it sold a consumer version of the product that does not exist. Each discovery took ten minutes of reading. Each had been spending for months.

An account that stops reading its own queries has outsourced its definition of relevance to a counterparty that gets paid either way. Write that sentence into the account's operating doc, because every new hire and every new automation feature will pressure the team toward not looking.

Negatives are how you talk back

A maintained negative list is the fence that makes broad match safe to run, and the accounts where broad match earns its reputation are always the ones with hundreds of accumulated negatives doing quiet work. The mechanics matter: negatives layered at the account level for the categorical exclusions, campaign level for the strategic ones, reviewed on a calendar, with the search terms reading feeding new entries every cycle. Building that fence is unglamorous, which is why it is skipped, which is why the horror stories exist. Here is the operating rhythm that works: weekly reading while a campaign is new or recently changed, biweekly at steady state, and a hard rule that every reading session must either add negatives or affirmatively conclude none were needed. The rule matters because it converts read the report from an intention into a checkbox someone can miss visibly.

There is also a positive discipline hiding in the same report, and it is the part the horror-story crowd misses: broad match is running a continuous, free query-research program on your behalf. The odd match that converts is a keyword you did not know you had, ready to be promoted into its own exact-match presence with its own ad and page. The best search accounts I have run treat broad match as the scout and exact match as the settler, a pipeline from discovery to ownership, and the search terms report is the border crossing between them.

The deeper point is about automation generally, and it is the same one I made about Performance Max: the systems are good, and they are not on your side. They are on the side of the objective you fed them, pursued literally, with your money. Broad match with clean signals and a tended negative list is a powerful instrument. Broad match set and forgotten is a standing order to buy whatever the machine finds interesting. Read the stack. It is your money in there.

Quick answers

Should I use broad match keywords?

Only with the guardrails that make it safe: value-based smart bidding, dense negative lists, and weekly search-term hygiene. Without those, broad match is a permission slip for the auction to spend on questions you never asked.

How do I keep broad match under control?

Feed the bidding real conversion values, review search terms on a schedule, promote winners to exact, and let negatives accumulate like the budget defense they are. Control is a routine, not a setting.

Also worth reading