A Search campaign with 800 keywords on Broad match is not a larger strategy. It is the same budget asked to learn 800 things at once, plus every close variant Google decides is related. Catalog stores feel this as a thin spread: a few obvious terms spend, hundreds of ad groups sit at zero impressions, and the search term report fills with queries nobody wrote.

The operating rule we use is stricter than Google's default. Keywords stay in Phrase and Exact. A campaign holds at most 150 terms that have already survived a volume check. Empty ad groups are deleted. The list is sized to the daily budget so each term can actually be seen.

Why 150 is a budget decision, not a magic number

Google will accept far more keywords than your budget can teach. If a campaign can spend $80 a day and the average click in that category is $2, you have on the order of 40 clicks a day to allocate. A list of 600 terms does not get a fair read. It gets a handful of auctions concentrated on whatever is cheapest or most famous, and silence everywhere else.

One hundred and fifty is the cap we hold so a campaign remains inspectable: you can still say which ad group owns which idea, and you can still read the search terms. It is not a claim that 149 is profitable and 151 is not. If the budget is smaller, the list should be shorter. If a term has no volume after a real Keyword Planner check, it does not get to occupy one of the slots. That check is the point of running Keyword Planner twice.

Phrase and Exact are the match types that stay

Broad match is a different product. It will spend on queries you did not list, and the negatives have to do more of the work. Some accounts test it later, on purpose, with a separate budget and a person watching search terms. It should not be the way a 1,000 SKU store "covers the catalog" in week one.

Phrase and Exact still are not literal. Close variants exist. That is why look-alike spellings need their own volume check, and why a shorter keyword can already cover a longer one. Those two controls sit beside the cap: split spellings and plurals, and stop ad groups bidding on the same idea.

Delete the empty structure

An ad group with no keywords after the sieve should not stay in the campaign as a placeholder. It confuses reporting and invites someone to paste Broad match back in "so it serves." Delete it. If the product still matters and truly has no search demand under its current name, the next step is a different query, a different page, or acceptance that Search is the wrong channel for that SKU. Silence is data.

How Kopa applies the sieve

In Kopa, Smart Campaigns keeps the draft on Phrase and Exact, caps the campaign at 150 terms that cleared the volume checks, drops empty ad groups, and fits the list to the daily budget. You see the cut list before anything is published.

What good looks like

  • The live campaign is Phrase and Exact only, unless a separate test says otherwise in writing.
  • Keyword count stays at or under 150, and under that when the budget cannot support 150.
  • Terms with no verified volume are not in the campaign.
  • Empty ad groups are removed, not parked.
  • Broad match is a deliberate later test with its own watch, not the default coverage method.