The First Thing to Do in a New Google Ads Account: Freeze, Don't Fix

One tactic, taken apart: why a search-term read-only audit beats rushing to restructure an inherited account in week one.

· QueryCut

The short version

  • Week one is read-only: export 90 days of search terms before touching structure or bids.
  • Negatives are reversible and preserve learning; restructures reset it, so negate first and rebuild later.
  • Broad-match-heavy accounts with thin negative lists yield the most waste per hour; exact-match accounts yield little.
  • Use a spend rule (2x to 3x target CPA, zero conversions, off-intent) for performance cuts, but negate wrong-intent terms on sight.
  • Do not mass-add negatives on a Smart Bidding campaign; surgical cuts in batches avoid starving the bid strategy.

When you inherit a Google Ads account, your first move should be a read-only search term audit, not a restructure. Pull 90 days of search terms, find where broad and phrase match are matching to junk, and stage negatives before you change a single bid or campaign. The mechanism matters: structural changes reset Smart Bidding's learning and destroy the historical read you need to make good decisions, while negatives are reversible and non-destructive.

The tactic: a read-only search term pass before anything else

The specific move is this. In the first week, go to Campaigns > Insights and reports > Search terms, set the date range to the last 90 days, and export everything. Do not touch bids, budgets, match types, or campaign structure yet. You are reading, not editing. The deliverable from week one is a list of search terms the account paid for, sorted by cost, with your judgment on each: keep, negate, or watch.

Why search terms first and not, say, the account structure or the conversion tracking? Because search terms are the only place where you see what the account actually did, in the advertiser's own money, independent of how the previous manager labeled things. Campaign names lie. Match types mislead. A campaign called 'Exact Match Brand' can be quietly hemorrhaging budget on broad-matched competitor terms if someone changed the keywords and never renamed it. The search term report is ground truth.

This is also the fastest path to a win you can show the client inside 30 days. Negatives cut waste immediately and they are completely reversible. If you negate a term and it turns out to matter, you add it back and you have lost a few days, not a quarter of learning data.

The mechanism: why negatives are safe and restructures are not

Here is the distinction that makes this tactic work. Adding a negative keyword changes which queries enter the auction. It does not reset anything. Smart Bidding keeps its conversion history, your Quality Score signals stay intact, and the campaign keeps running on the data it already has. You are trimming inputs, not rebuilding the machine.

Restructuring is the opposite. When you move keywords into new ad groups, split campaigns, or swap bid strategies, you detach the performance history from the entity that was learning. A Target CPA or Target ROAS strategy in a new or materially changed campaign re-enters a learning period, and during that window its bids are less reliable. If you restructure on day two, you have blinded yourself and the algorithm at the exact moment you least understand the account.

This is why the sequence is freeze, read, trim, then (only later) rebuild. Negatives during the read phase are the one aggressive action that carries almost no structural risk. You get to improve efficiency and buy yourself time to understand the account before you make the irreversible calls.

Where this pays the most: broad-match-heavy accounts

The accounts where a search term audit returns the most waste per hour of effort are the ones leaning hard on broad match, especially broad match paired with Smart Bidding and a thin or sloppy negative list. The reason is mechanical. Broad match deliberately expands beyond the keyword text to anything Google's models judge related to the query intent and the landing page. The looser the match type and the weaker the negative coverage, the wider the net, and the more off-intent queries you will find in the report.

An exact-match or tightly-phrased account that has been actively managed will surface far fewer negatives, because the matching layer was never allowed to wander in the first place. If you inherit one of those, the search term pass is still worth doing, but temper your expectations: the win will be a handful of terms, not pages of them. The volume of recommended negatives is roughly proportional to how much latitude the match types and bid strategy were given to explore.

A practical tell: before you even export, look at the match type distribution. If most spend sits on broad match keywords and the negative list is short, expect a long negative list and plan time for it. If spend is on exact match with a mature negative list, scan for anomalies instead and move on.

The thresholds: what to cut, what to leave alone

Do not negate a term just because it has zero conversions. A single expensive query with no conversions can be noise. Use a spend-based rule tied to the account's target CPA. A workable heuristic: if a search term has spent more than 2x to 3x the target CPA with zero conversions, and it is clearly off-intent (wrong product, wrong intent, a free-seeker query like 'free' or 'jobs' or 'DIY'), negate it now. If it is only slightly over and plausibly relevant, flag it to watch.

Volume matters for the zero-conversion read. Wait for roughly 100 clicks on a term before you trust a zero-conversion verdict as a performance signal, because below that a single missed conversion swings the picture. The exception: intent-based cuts need no click threshold at all. If the term is 'free resume template' and you sell paid software, you negate it on sight regardless of clicks, because the problem is relevance, not statistics.

Remember the mechanical limit: a single negative keyword can be at most 10 words. If you are tempted to negate a 12-word query verbatim, you cannot, and you should not want to. Negate the shared junk token or phrase instead ('free', 'salary', the name of a product you do not sell), which catches the whole family of queries rather than one exact string.

The failure mode: strangling a learning account with negatives

Here is where the standard 'audit and cut waste' advice backfires. A broad-match-plus-Smart-Bidding campaign uses the breadth of queries it is allowed to match as exploration fuel. If you walk in and add 200 negatives in week one, you can starve the bid strategy of the signal it was using to find converters, and performance can get worse, not better, even though every individual negative looked defensible.

The trap is treating every irrelevant-looking query as pure waste. Some off-core queries are how the system discovers adjacent intent that converts. The safe version of the tactic is surgical: negate the clearly wrong (wrong product, free-seekers, job seekers, competitor-by-name if the client does not want them) and leave the merely-adjacent to the watch list. Cut in batches, then wait a week or two and re-read, rather than dumping a giant list in one action.

The honest limit: you cannot know from the search term report alone whether an adjacent term is a hidden converter with a long lag or genuine waste. Conversion lag, cross-device paths, and view-through all hide in data the report does not fully expose. When a term is ambiguous, the conservative move is to watch it for one more conversion window before cutting, because re-adding a converter you killed costs far more than the few dollars you spend waiting.

Questions people also ask

How do I audit a Google Ads account I inherited?

Start with the search term report under Campaigns > Insights and reports > Search terms, set to the last 90 days, and export it before changing anything. Search terms are ground truth about where the money actually went, independent of how campaigns were named or structured. Identify clear waste, stage negatives, and only after you understand the account should you consider restructuring.

Should I pause or restructure campaigns in the first week?

No. Restructuring detaches performance history and pushes Smart Bidding strategies back into a learning period, which blinds both you and the algorithm right when you understand the account least. Do your trimming with negative keywords first, since they improve efficiency without resetting anything, and save structural changes for after your 30-day read.

How many clicks before I trust a zero-conversion search term?

Roughly 100 clicks before you treat zero conversions as a real performance signal, because below that a single missed conversion changes the read. That threshold only applies to performance-based cuts. If a term is clearly the wrong intent for the business, negate it immediately regardless of click count, since the issue is relevance, not sample size.

Why does one account produce way more negatives than another?

The volume of negatives tracks how much latitude the match types and bid strategy were given. Broad match deliberately expands to related queries, so broad-match-heavy accounts with thin negative lists surface long negative lists. Tightly managed exact-match accounts have already constrained the matching layer, so they surface far fewer.

Can mass-adding negatives hurt performance?

Yes. Smart Bidding on broad match uses query breadth as exploration fuel to find converters, so dumping hundreds of negatives at once can starve it of signal and worsen results even when each negative looked justified. Cut the clearly-wrong terms, watch the merely-adjacent ones, and add negatives in batches with a week or two between passes.

More from the blog

Landing Page Experience: The Changes That Actually Move ItImpression Share Lost to Rank vs Budget: What Each One MeansHow Many Campaigns Is Too Many for the Budget You Have?Primary vs Secondary Conversion Actions: What Each One Actually Changes