Your Google Ads Budget Is Aimed at the Wrong Words

Look at your Google Ads account and find the keyword you spend the most on. Odds are it is one or two words, it has the highest cost per click in the account, and it has been in there since the day the account was built. Data published by Search Engine Land on September 11, 2026 says that keyword is quietly losing the argument: across the Google Ads dataset Jason Tabeling analyzed, one and two word queries fell from 62% of conversions in January 2025 to 52% in August 2026, while three and four word queries climbed from 20% to 46%.

Every article ranking for this topic ends in the same instruction: go build a bigger long-tail keyword list. I think that advice is a year out of date. With broad match and AI Max deciding which searches your ads show on, adding more keywords no longer controls much of anything, and the accounts that win are the ones mining the search terms report for the long conversational queries that are already converting, then pointing budget, negative keywords and ad copy at those. That is the whole article. What follows is the exact audit I run to find them, in an order a business owner can do without an analyst.

The words carrying your conversions are not the words on your bill

Two numbers moved at once in that dataset, and the gap between them is where the money leaks. Impression share for one and two word queries dropped from 42% to 24%, meaning your head terms are simply being typed less. At the same time their conversion share only fell from 62% to 52%. Fewer people search that way, but the ones who do still convert, so the terms keep looking healthy in a report while the auction underneath gets thinner and more expensive per useful click.

Ten years of running Google Ads accounts through Google's automation shifts is why I read that as an account problem and not a trend piece. I watched exact match stop meaning exact, watched phrase match absorb close variants, watched broad match get a smart bidding chaperone, and now watch AI Max choose queries nobody in the account ever wrote down. Each time, the owners who lost money were the ones still managing the keyword list instead of the query list.

Picture an emergency plumber in Tampa. The account was built around "plumber Tampa" because that is what the first agency bid on, and it still eats about 40% of the monthly budget at the highest cost per click on the list. Meanwhile the search terms report, which is the tab showing the actual phrases people typed before they clicked, is full of things like "burst pipe under sink who to call now" and "emergency plumber open sunday near me." Those are converting at a fraction of the cost and getting a fraction of the budget, because nobody ever wrote them down as keywords, so nobody ever looked at them as a budget line.

Bidding on the category is bidding against everyone who sells anything like you

A one word search is a person who has not decided anything yet. They have not chosen urgency, location, price range, or brand. NP Digital's study of 40 companies running paid search, published May 2025 with branded terms excluded, put one-word keywords at a 0.17% conversion rate and four-word keywords at 1.61%. That is roughly nine times the conversion rate, and it means a four-word click at the same kind of price turns into a customer roughly nine times as often. Six-word terms came in highest at 1.94%.

Owners stay on the head term for a reason that is not stupid. It is legible. "Plumber Tampa" is the word you would use to describe your business at a networking event, so seeing it in the account feels like the account is doing the right job. Volume reports reinforce it, since the head term always shows the most impressions and clicks. A four-word phrase looks like a rounding error next to it, right up until you sort the same report by cost per conversion instead of by clicks.

The second reason is structural. Google's own interface nudges you toward adding keywords, raising bids and accepting recommendations, and I have written before about how Google's account recommendations are built to make you spend more, not to make you spend better. None of those prompts ever say "your best-performing query is not a keyword in this account and it is underfunded." That observation only comes from a human reading the search terms report.

Google already picks the queries, so stop pretending the keyword list does

Broad match means Google can show your ad on searches it judges related to your keyword, not just the keyword itself. AI Max goes further and lets Google's systems expand matching and even rewrite ad copy on top of that. Search Engine Land reported on September 7, 2026 that Google is also testing traditional Search campaigns appearing inside AI Mode, which pushes ad delivery further into long conversational phrasing that no keyword list anticipated. The mechanics of that expansion, and the conditions under which I turn it off, are in my breakdown of why AI Max for Search is broad match wearing a new badge. Today's post assumes it is on and asks what to do with the results.

Once you accept that Google chooses the queries, the long-tail keyword list becomes busywork. You are writing down guesses about phrases that the system was going to match anyway. What you control is different and more powerful: which queries you fund, which you block, and what your ad says when one of the good ones fires.

Tabeling is blunt about the cost of ignoring it. He writes that "continuing to pour budget into broad, high-cost head terms that are showing diminishing conversion share is a recipe for inefficiency," and his prescription is that "the most successful advertisers will be those who audit their performance through the lens of query length and reallocate their budgets toward the more cost-effective, higher-converting long-tail segments." I agree with the audit and part ways on the execution. He and most of the ranking articles land on building out long-tail keywords. I would not add a single one. Sort what Google is already sending you, then move money and message toward the winners.

For historical context, this pattern is old. Seer Interactive's study by Allie O'Brien in August 2012 found that across 12 clients, short-tail terms pulled 11 times more traffic while long-tail conversion rates ran 4.15% higher on average. That was organic search data and it was fourteen years ago. What changed is not the direction of the effect, it is that the volume has now moved too, so the head term no longer even buys you the reach that used to justify its price.

The audit takes one afternoon and needs no agency

Run this yourself with a laptop and a spreadsheet. Everything in it lives in the standard Google Ads interface, no scripts, no tools, no third-party platform.

Step five is the one that changes the conversation, because it converts a vague complaint into a line item. If that audit shows, say, a two-word term that spent three months of budget and produced nothing, you are no longer arguing about whether the ads work. A vague complaint gets managed. A number gets fixed.

What to do with the ten phrases you just wrote down

A negative keyword is an instruction telling Google never to show your ad on searches containing that word. It is the cheapest lever in the account and the most neglected, because adding negatives feels like shrinking and owners are trained to want growth.

Waste-cutting and budget-pointing are different jobs and they run in that order. If the account is also leaking on click fraud, irrelevant placements and duplicate audiences, sort that first with the wasted spend audit that finds where your account is leaking, then come back here to decide where the surviving budget points. Cleaning a pipe and aiming it are not the same task.

Measure the split, not the total, and ignore impression share

Rerun the word-count audit every quarter, not every week. Google's bidding needs enough conversion volume to learn from, and a small account that gets re-cut every Monday never accumulates the data to make the next decision well. Four readings a year is enough to see the drift.

The metric that matters is your bucket split: what percentage of spend and what percentage of conversions each query-length group holds, tracked across quarters. If conversions keep migrating toward the longer phrases while spend does not follow, the account is misaligned no matter how good the headline numbers look.

Two traps are worth naming. The first is impression share on head terms, which is the percentage of available impressions you captured. Agencies love the chart because losing it feels like losing ground. Given that one and two word impressions in the Search Engine Land dataset dropped from 42% to 24% of the total, winning a bigger slice of a shrinking pie is an expensive way to stand still. The second is total click volume. Long queries produce fewer clicks by definition, so an account that is correctly rebalanced will often show fewer clicks and more customers in the same month. If your reporting rewards clicks, it will punish the right decision.

Tabeling's dataset is a slice of Google Ads accounts, not your industry, and NP Digital's 40 companies are an average your business may sit either side of. That is why step two of the audit matters. The published numbers tell you which direction to look. Your own search terms report tells you whether it is true where you spend.

Frequently Asked Questions

Should I delete my one-word and two-word keywords?

Not immediately. Run the 90-day audit first and look at whether those terms produced conversions at an acceptable cost. If a head term produced zero conversions over 90 days, pause it and move the budget. If it converts but expensively, lower its budget share rather than removing it, because the head term still feeds the bidding system data it learns from.

How do I find the search terms report in Google Ads?

In the Google Ads interface, open Campaigns, then Insights and reports, then Search terms. Set the date range to the last 90 days and add the columns for cost and conversions. Download it to a spreadsheet, because the word-count analysis is far easier there than in the interface. The report only shows searches that met Google's privacy thresholds, so your totals will not perfectly match the account totals.

Is this worth doing on a small budget of a few thousand a month?

It matters more on a small budget, not less. A large account can absorb one expensive head term across many campaigns, while a small account can have most of its monthly spend sitting on two or three short keywords. When NP Digital's May 2025 study put one-word keywords at 0.17% conversion and four-word at 1.61%, that gap decides whether a modest budget produces leads at all. The audit costs an afternoon.

The keyword you are proudest of is often the one costing you customers. Nineteen months moved nearly half of all conversions into phrases nobody in your account ever typed into a keyword field, and the accounts that noticed did it by reading their own search terms report, not by buying a tool. Your customers have been telling you exactly what they want in four and five word sentences this whole time. The question is whether your budget has been listening.

If you want a second set of eyes on your own query-length split before you move money around, book a call and we will pull the 90-day report together and name the number your head terms are costing you.