Back to all questions
Google Ads Strategy
We turned off all our broad match and nothing happened. What does that tell us?
J
Jenna Krause
Asked Apr 21, 2026
3
91
Genuine puzzle rather than a humblebrag.
Six weeks ago we paused every broad match keyword in the account (about 40 percent of the
keyword count, roughly 25 percent of spend). Expected: volume drop, CPA improvement, an
argument about which mattered more.
Actual: total conversions within 2 percent of before, total spend within 4 percent, CPA
essentially flat. The phrase and exact keywords simply absorbed the traffic.
Three readings, and I cannot distinguish between them:
1) The broad keywords were matching almost entirely to queries our phrase keywords also
matched, so they were redundant all along.
2) The budget was the binding constraint, so removing supply just redistributed the same
money (in which case we learned nothing about match types at all).
3) Six weeks isn't long enough and something will show up in the seasonal tail.
Is there a way to tell these apart after the fact? I have the search terms for both periods,
which feels like it should be enough, and I can't work out the actual test.
Replies (3)
G
Greg Pallister
·
Apr 21, 2026
Reading 2 is testable and it's the one to rule out first. Were the campaigns hitting their
daily budgets before the change? If yes, you didn't run a match type experiment, you ran
nothing at all - the budget decided the outcome and the keywords were never the constraint.
That's the single most common way this experiment gets misread, and half the case studies
you see published haven't checked it...
M
Marta Silva
·
Apr 22, 2026
For reading 1 the search terms comparison does work, and here is the shape of it.
Take the search terms from the six weeks before and the six weeks after. Look at unique
terms, not spend. If the set of queries you appeared for is essentially the same in both
periods, the broad keywords were redundant - they were bringing you traffic you were
already reachable for.
If the after period is missing a chunk of distinct queries but total volume held, then
broad was reaching genuinely different searches and something else compensated. That's a
different and more interesting finding.
The distinction matters because in the second case you have lost reach you'll feel later,
exactly as your reading 3 suggests.
P
Priya Deshmukh
·
Apr 23, 2026
Adding one thing to check that's not in your three readings: close variants.
Your phrase and exact keywords match far more loosely than their names imply. It is
entirely possible that they absorbed the broad traffic because they are now doing what
broad used to do. If that's what happened, the finding isn't "broad was redundant" but
"match types have converged", which has quite different implications for what you do next.
The search terms comparison Marta describes will show you this too - look at whether your
exact keywords are now matching things you would not have called exact two years ago.