Open almost any Google Ads audit template and you will find a line for "number of negative keywords," scored as if more is better. The instinct is intuitive: negatives block waste, so a bigger list must block more waste. It is also, as far as the available data goes, wrong. A practitioner analysis of 31 accounts covering roughly $133M in spend found that the correlation between negative-list size and zero-conversion spend was near zero — about r = -0.10, which the author described as close to random. A 2,000-line list can sit on top of exactly as much wasted spend as a 50-line one.
That finding should change what you put on a scorecard. List length is seductive because it is trivial to measure — one number, always available, always goes up — but a metric that does not move with the thing it is supposed to represent is a vanity metric, and negative-list count is one. This post is about what to measure instead: waste concentration, the share of your zero-conversion spend that lives in your worst terms. It is almost as easy to read as a list count, it actually tracks the money, and it tells you in two minutes whether your last cleanup did anything.
The short answer: length barely tracks waste
The direct claim is that negative-list length and wasted spend are nearly uncorrelated across real accounts, so counting entries tells you almost nothing about whether an account is clean. In the 31-account study the relationship was statistically indistinguishable from no relationship at all. That does not mean negatives are useless — a well-aimed negative stops a real leak. It means the number of them is not the quantity to track, because two accounts with wildly different list sizes can be equally wasteful or equally clean. The quantity that varied with waste was not length; it was which specific terms and words the account was leaking through.
The same analysis found the leak was concentrated: a small number of recurring words accounted for most of the zero-conversion spend, each burning large sums while the rest of the list contributed little. That is the real shape of waste in a Google Ads account — not evenly spread across thousands of terms that a long list dutifully covers, but piled into a short head of expensive terms. A cleanup that adds a hundred low-impact negatives while missing the handful of terms carrying the leak grows the list and leaves the waste in place, which is precisely how an account can score well on "negative count" and badly on the P&L.
Why list length fails as a signal
List length fails for two structural reasons. The first is that negatives block strings, not meanings. A negative keyword does not expand to close variants, plurals, or synonyms the way a positive keyword does, so a list is only ever as current as the exact wordings someone has manually added. The queries reaching your account are not static — broad match and AI Max keep introducing fresh phrasings, misspellings, and paraphrases — so a large list built in the past is a record of old leaks, not a guarantee against new ones. Length measures accumulated effort, which is not the same as current coverage.
The second reason is that length is blind to aim. Fifty negatives aimed at the exact terms bleeding money will beat two thousand aimed at low-cost or hypothetical queries every time, because waste is concentrated and only the aimed negatives hit the concentration. A scorecard that rewards count actively encourages the wrong behaviour: it pays people to pad lists with borrowed "universal" negatives and speculative blocks, which inflate the number, increase the risk of blocking good traffic, and do nothing for the leaks that matter. The metric creates the dysfunction it then fails to detect.
What to measure instead: waste concentration
Replace the count with a concentration figure: of all the money you spent on terms that produced zero conversions, what share sits in the top 10% of those terms by cost? This single number captures the thing list length misses. A high concentration — most of your zero-conversion spend piled into a few expensive terms — means the leak is both large and fixable, because a short list of well-aimed negatives can close most of it. A low, flat concentration across many cheap terms is a different problem that negatives alone will not solve, and knowing which situation you are in tells you where to spend your time.
The reason concentration beats count is that it moves when you fix things. Negate the terms carrying the leak and the concentration figure drops at your next check, because the expensive zero-conversion terms are gone and what remains is smaller and more even. A list count, by contrast, only ever rises, whether or not you touched the actual waste. Concentration also pairs naturally with method: once it tells you the leak is concentrated, an n-gram pass over those terms shows you the shared words so you can close many leaks with few, well-chosen negatives rather than one entry per term.
How to compute the concentration check in the UI
You can do this in the search terms report without any extra tooling. Open the search terms report at the account or campaign level, set a window long enough that a zero in the Conversions column is meaningful — 30 to 90 days for most accounts — and sort by Cost descending. Add Conversions (and Conv. value if you bid to value) to the view. Now identify the terms with zero conversions, total their cost, and compare the cost of the top tenth of those terms to the whole. That ratio is your concentration. Write it down, because the number only means something as a trend: today's figure is a baseline, and the next reading after a cleanup is the result.
Two honesty caveats keep the figure trustworthy. The search terms report is privacy-thresholded, so low-volume terms do not appear even when they took spend — your true waste is somewhat higher than the visible rows, though the concentration ratio is fairly robust to that because the missing terms are by definition low-volume and sit in the tail, not the head. And a zero-conversion term is not automatically waste: it may have assisted a conversion credited elsewhere, or convert after the window closed, so read the head terms before you negate rather than blanket-blocking every zero. If you want a fuller instrument panel around this single ratio, the wasted-spend metrics scorecard sets out the thresholds that turn a reading into a verdict.
The counter-argument, honestly
There is a respectable objection worth stating fairly. Some agencies report that the first hundred or so negatives capture the large majority of recoverable waste — that there is a steep early return to negating, and the uncorrelated-with-length finding is really about the long tail past that point. That is consistent with the concentration story, not against it: if waste is concentrated, then a small number of well-aimed negatives does capture most of it, which is exactly why the first batch pays off and the next thousand entries do not move the needle. The early return is real; what is not supported is the extrapolation that more entries keep buying more savings.
So the two views reconcile cleanly. Negatives work, and the first well-aimed ones work best — but that is an argument for aim and for stopping, not for length. It is the same shape as the point where negative-keyword mining stops paying off: the curve flattens, and past the flattening the marginal negative costs maintenance and traffic risk while returning nothing. Grade the account by whether waste concentration is falling toward that floor, not by how many lines the list has reached. The cleanest accounts are not the ones with the longest lists — they are the ones where the few terms that leak money keep getting found and fixed, and the number of negatives it took to do that was never the point.