GOOGLE ADS KEYWORD LIMIT: WHEN YOU GET CLOSE AND HOW I MANAGE IT

In five years of managing accounts I have hit a hard google ads keyword limit exactly twice, and neither time was because the account had too many keywords in any meaningful sense. Both times it was structure — a bulk upload dumping tens of thousands of terms into a handful of ad groups, and an experiment quietly duplicating a campaign that was already close to the edge.

That's the pattern I want to describe here, because the question "how many keywords can I have" has a boring answer that almost nobody needs, and a useful answer that almost nobody publishes. The boring answer is a number. The useful answer is that you will hit a manageability wall at maybe 5% of the documented ceiling, and the accounts that genuinely approach the technical one have a structural problem, not a volume one.

So the practical framing is: the limit is not a budget to spend down. It's a tripwire that tells you something upstream went wrong.

Where the keyword limit google ads enforces actually sits

Two ceilings matter, and they behave completely differently.

The account-level one is enormous: millions of ad group targeting items across the whole account. I have never seen a single advertiser account get within an order of magnitude of it, and I don't plan around it. Google's own official account limits page lists it alongside everything else, and it reads more like a systems safeguard than an advertising constraint.

The per-ad-group one is the one that bites: twenty thousand targeting items in a single ad group. Two things about that number matter more than the number itself. First, it counts targeting items, not keywords — placements, audience lists and other criteria share the same pool, so a Display ad group with heavy audience layering is spending that allowance on things you don't think of as keywords at all. Second, twenty thousand keywords in one ad group is already so far past sane structure that if you're near it, the limit is the least of your problems.

Negative keywords have their own separate allowances — per campaign, per list — which is why a client can be nowhere near the positive ceiling and still get blocked mid-cleanup. If you're building exclusion lists at scale, the negative keyword fundamentals matter more than the positive-side math, because those caps are much lower and much easier to reach.

Who actually gets close

From the accounts I've worked on, three profiles approach the ceiling and one never does.

Large e-commerce catalogues built programmatically

A retailer generating keywords from a product feed — brand × model × modifier — hits six figures across the account without anyone deciding to. Each combination looks reasonable in isolation. The generation logic doesn't know what a reasonable total is, and nobody reviews the output because the whole point was automation. This is the profile I most often find already over the manageability line.

Aggregators and marketplaces with long-tail strategies

Job boards, real estate portals, ticketing. The strategy is explicitly to own the tail: city + category + qualifier, multiplied across a country. Volume is intentional here, which makes it healthier than the e-commerce case — but the per-ad-group ceiling comes into play if the generation script groups too coarsely.

Legacy accounts nobody has pruned

The most common and the least defensible. Ten years of additions with no removals. When I audit one of these, the majority of keywords have zero impressions over twelve months. They are not near a limit because the strategy demands scale; they are near it because deletion never happened.

The profile that never gets close

Lead gen with a few hundred terms — and increasingly, anything running on broad match plus Smart Bidding. That's the inference I'd emphasise: the platform's own direction of travel has moved the ceiling further away. Broad match and automation replace ten thousand explicit variants with a few hundred well-chosen ones, so the accounts I build today have a fraction of the keyword count of the accounts I inherited five years ago. Anyone asking about the limit for a modern Search build is almost certainly solving a problem they don't have — the real question is how many keywords a campaign should actually hold, which is a strategy question, not a capacity one.

The signals I watch before an error appears

By the time the interface returns a limit error, you're in remediation. These are what I look at first.

Editor gets slow. Google Ads Editor downloading a single account in fifteen minutes rather than two is the earliest signal I get, and it shows up long before anything else.

Bulk uploads fail without an obvious reason. A partial-failure message on a sheet that validated fine is often a resource-count issue reported vaguely.

Ad group counts with a suspiciously round shape. When I sort ad groups by keyword count and the top of the list clusters near a round number, that's not organic growth, that's a script that stopped.

Anyone mentions running an experiment. This is the one that catches people. When you launch an experiment, Google duplicates the keywords and targets from the base campaign and those duplicates count toward your account totals — a base campaign near the line becomes two campaigns over it. Both times I hit a wall, this was a factor. If you plan to test, you need headroom reserved for the copy, not just the original.

How I manage headroom in large accounts

My working rule: stay under half the per-ad-group ceiling, and treat anything above a few thousand keywords in one ad group as a structural bug regardless of what the limit says.

Quarterly zero-impression sweep

Twelve months, zero impressions, pause then remove after a season. On a legacy account this alone typically clears 30-50% of the count. Nothing else I do reclaims capacity that cheaply.

Deduplicate before you scale

Before any bulk addition I run a duplicate pass, because the same term across match types and ad groups inflates the count without adding reach. I've written up the full method in my walkthrough of cleaning 3,000 redundant keywords out of one account — the short version is that duplicates cost you legibility and headroom, not money.

Cap generation at the source

For programmatic accounts I put the ceiling in the script, not in the review step: a hard cap per ad group and a minimum expected search volume per generated term. Reviewing 80,000 generated keywords after the fact never happens. Constraining the generator does.

Split across accounts only when there's a real reason

People reach for a second account far too early. It costs you shared conversion history, shared budgets and shared negative lists, and it fragments the signal Smart Bidding learns from. I split by market or business unit when reporting demands it — not to dodge a keyword count.

Reserve experiment headroom

Any campaign I expect to test gets planned at half its allowance, so the duplicate fits. That's the cheapest fix on this list and the one I adopted only after being caught by it.

The limit that actually constrains you

Here's the inference that matters more than any number Google publishes: I hit a human ceiling around 10,000-15,000 active keywords per account, and it arrives at roughly 5% of the technical one.

Past that point I can't hold the account in my head. I can't scan a search terms report and notice the anomaly. Bid adjustments become batch operations I can't sanity-check. Every diagnosis takes an hour instead of ten minutes. The account keeps running — the platform is fine — but my ability to improve it degrades, and that degradation costs far more than any capacity constraint ever would.

The corollary is that a keyword limit google ads exposes is a diagnostic, not a wall. If an account is approaching it, either the generation logic is unconstrained or nothing has been removed in years. Both are worth fixing on their own terms. And with automation absorbing more of the matching work, keyword count is becoming a weaker proxy for coverage each year — which is exactly why I plan capacity around what I can manage, not what the platform permits.

The practical takeaway

Don't plan around the google ads keyword limit. Plan around the number you can actually manage, keep ad groups well under the per-ad-group ceiling, sweep zero-impression terms quarterly, cap generation in the script rather than in review, and reserve headroom before you run experiments.

If you're currently near a ceiling, the fix is almost never a second account. It's a pruning pass — and the account will be easier to work in afterwards regardless of whether the limit was ever a real risk.