There’s a particular frustration that comes with reviewing your search terms report and finding a long list of queries you never intended to show for. Queries that are adjacent to what you sell, but not quite right. Queries where the person clearly wasn’t looking for your product. Queries where you’ve paid for a click you had no business winning.

If your campaigns use exact match keywords, this shouldn’t be happening. That’s the whole point of exact match — to show your ads only when someone searches for precisely that term. Except that’s not how exact match works anymore.

What Exact Match Used to Mean

When exact match keywords were introduced, they meant exactly what the name suggested. If you added [leather wallet] as an exact match keyword, your ad would only show when someone searched for those two words in that order. Nothing more, nothing less.

This gave advertisers precise control over which searches triggered their ads. You could confidently build campaigns knowing that your keywords and your search terms would align closely. The search terms report held few surprises.

That era is over.

What Exact Match Actually Means Now

Google has progressively expanded exact match to include what it calls “close variants.” The official definition has broadened over the years, and at this point it encompasses:

  • Plurals and singulars (searching “leather wallets” can trigger [leather wallet])
  • Common misspellings (“lether wallet” can trigger [leather wallet])
  • Reordered words where intent is judged the same (“wallet leather” can trigger [leather wallet])
  • Implied words (“men’s slim wallet” can trigger [leather wallet] if Google judges the intent is close)
  • Semantically similar queries where Google decides the meaning is equivalent

That last one is where it gets contentious. “Semantically similar” is a judgement call made by Google’s AI — and Google’s AI doesn’t always agree with your commercial judgement about what’s relevant to your business.

What This Looks Like in Practice

Here are realistic examples of what you might see in a search terms report if you’re running exact match [leather wallet]:

  • “genuine leather wallet” (reasonable, you probably want this)
  • “slim leather wallet” (possibly reasonable, depends on your range)
  • “best leather wallet under £30” (informational intent — they’re researching, not necessarily ready to buy)
  • “brown leather wallet men” (probably fine if you stock that)
  • “vegan leather wallet” (this is a completely different product)

That last one illustrates the problem clearly. Vegan leather is not leather. A customer searching for a vegan leather wallet doesn’t want what you’re selling. But if Google’s AI decides that “vegan leather wallet” is semantically similar to [leather wallet] — because both are wallets made from a leather-like material — your ad shows and you pay for a click from someone who can’t convert.

This isn’t hypothetical. These kinds of expansions appear in real accounts regularly.

The Business Impact

For most e-commerce businesses, the net effect of close variants is mixed.

The genuine plurals, misspellings, and reordered-word matches are usually fine and often beneficial. You’d want to show for those searches anyway, and having exact match capture them automatically saves you from having to manually list every variation.

The semantic expansion is where budgets leak. When Google decides that two queries have “the same intent” based on its model of searcher behaviour, it doesn’t necessarily share your knowledge of what’s actually relevant to your specific products. Irrelevant traffic from over-broad semantic matching adds up — particularly in categories with adjacent terms that sound similar but mean very different things.

The subtler problem is what happens to your data. If your exact match keywords are capturing a wider range of queries than you intended, your keyword performance data becomes harder to interpret. A keyword with a 2% conversion rate might be a mix of a 4% rate on genuinely relevant searches and a 0.5% rate on borderline ones. You can’t see that breakdown in the keyword report — only in the search terms report.

What to Do About It

Check your search terms report weekly. This is not optional hygiene anymore — it’s how you actually manage what your exact match keywords are doing. Filter for the last 7-14 days, sort by spend, and look at anything spending money without generating conversions.

Add negatives for genuinely irrelevant matches. When you find search terms that are clearly off-target — different product types, different materials, different audiences — add them as negative keywords at the campaign or account level. Be specific: if you don’t sell vegan leather wallets, “vegan leather” is a negative keyword candidate, not just “vegan.”

Don’t rely on exact match as a complete control mechanism. If precise control over which queries trigger your ads is critical to your strategy — for example, you’re separating brand terms from non-brand terms, or segmenting by product category for margin-based bidding — phrase match with tight negative keyword lists gives you more predictable behaviour than exact match alone.

Pay attention to keyword reports vs search term reports. Your keyword report shows you how your keywords are performing in aggregate. Your search terms report shows you the actual queries. For understanding what Google is actually doing with your exact match keywords, the search terms report is the document that matters.

The Deeper Shift This Represents

The expansion of exact match is part of a broader move by Google away from keyword-matching and towards intent-matching. Google’s position is that they can infer what a searcher wants more accurately than an advertiser can specify via keywords — and in many cases, they’re right.

But intent-matching works best when it aligns with what you’re actually selling. When the AI’s model of intent diverges from your commercial reality, you end up paying for traffic that can’t convert. The search terms report is your window into where those divergences are happening.

Exact match is still worth using — it still provides more control than phrase or broad match. But treating it as a precise filter set and then walking away is no longer a safe approach. The search terms report has become the real management interface for keyword campaigns.

If your Google Ads account hasn’t had a thorough search term review recently, a Google Ads audit from Roksys typically surfaces several hundred pounds a month in wasted spend from exactly this kind of keyword expansion.