If a new agency’s first move is to rebuild your entire Google Ads account, that is not proof it knows what it is doing. It is demolition dressed up as strategy.

I spent years managing accounts by hand for ecommerce shops and home services companies, mining search terms at 1am and watching performance fall apart because someone changed a landing page without telling me. I have also inherited messy accounts and wanted to tear them down just to make them mine. I used to tell clients a clean structure would fix everything. I was wrong. Clean feels good to the person doing the work. It does not tell you whether the new structure will buy a single qualified lead.

Here is my rule when you change agencies: keep the account, protect campaigns with proven results, and make the new provider earn every structural change with evidence. A proposal that starts with demolition before diagnosis should lose.

A takeover changes the manager, not the account’s owner

Your campaigns, keywords, ads and conversion data live in an account you own. If you hold Admin access, switching agencies does not require a new account. Freeze a baseline before the new team touches anything: what spent money, what produced qualified leads, and what changed recently. The handover checklist is a useful reminder to handle access deliberately rather than treating a new login as a fresh start. Unlinking the old manager ID preserves history inside your account, though shared remarketing lists and manager-owned invoicing do not come with it (what survives an unlink).

That is why the full-rebuild pitch should make you suspicious. It gives a new team visible work to do in week one. It lets them apply their preferred template before they have explained what is wrong with yours. If performance dips, they can point to the transition and the prior agency. What the deck calls a fresh start, I call two people getting paid to build the same campaigns.

This does not mean the existing account is sacred. It means the burden of proof belongs to the person proposing the change. If a campaign produces qualified leads at an acceptable cost, tell me what replacing it will accomplish that repairing it cannot. If no one can answer, leave it running while you investigate the rest.

A takeover plan should begin with the account you have, not the account the new agency would rather show in a sales presentation.

The Learning label is not permission to start over

Google’s visible “Learning” status is a bidding condition, not an onboarding ritual. Its Smart Bidding help says that status lasts no more than seven days; ads keep serving during that window. The documented triggers include a new or reactivated bid strategy, changes to the composition of campaigns, ad groups or keywords, and changes to conversion goals (the Learning triggers). Giving a new manager access is not one of them.

An agency that says it must rebuild your account “to get out of learning” owes you an explanation. What, specifically, is learning? Which change caused it? Why would replacing campaigns solve the underlying problem? A clean naming convention is not an answer. Neither is a screenshot of the status column.

The label also does not tell you when performance has settled. Google advises judging results over one to two conversion cycles or two weeks, whichever is longer, because conversion volume and lag matter after the visible status clears (why volume and lag outlast the label). Say you spend $20k a month and close 30 SQLs with a 21-day sales cycle. You cannot judge a major bidding change by whether a label disappears next Tuesday. You need enough clean conversion data to tell whether the change helped.

That is the cost a demolition pitch tends to skip. A rebuild within the same account does not erase its historical data. But replacing campaigns, changing goals and moving budgets can interrupt the continuity you need to evaluate bidding and business results. The account may look tidier long before you know whether the new setup performs better.

I audit tracking before I touch structure for exactly that reason. One account audit found conversion actions that had been deleted or had never recorded a result. No bid strategy or brilliant new campaign layout fixes an empty measurement feed. If your search terms show spend but your reporting cannot distinguish a booked call from a worthless form fill, fix measurement and establish a baseline first. Otherwise, the new structure has a more polished way to chase the wrong thing.

Ask what signal the rebuild improves, not how quickly the new team can build it.

Fix the bleed now; make the rebuild wait for evidence

Some changes should happen on day one. Broken conversion tracking, search terms burning spend on irrelevant queries, or a Performance Max campaign consuming 80% of the budget without producing SQLs do not need a month of ceremony. Pause the bleed, record what changed, and keep the rest running while you establish a baseline you trust. That is triage. It is not a license to replace every campaign in the account.

Claymation wrecking ball poised over a thriving miniature ad account

Preservation is where a takeover either protects the business or quietly makes it pay for the same work twice. One practitioner framework for a $100k-a-month account starts with a six-to-twelve-month performance audit and change-history review before deciding what to preserve or rebuild. Another account of the cost of a hard reset describes two to four weeks of relearning and pipeline exposure for accounts spending $10k a month or more. The sensible sequence is familiar: audit what exists, keep what works, fix tracking, then rebuild what cannot be repaired (why takeover plans say to keep what works).

I do order full rebuilds. Just rarely, and for a reason I can put in a sentence. If one Performance Max campaign lumps brand, competitor and cold prospecting together and gives you no way to control their CPA targets separately, a bid tweak will not create that control. If an account optimizes toward page views, or counts every form fill like a qualified sales call, changing bids alone will not turn its goal into pipeline. Those are problems with what the structure asks the system to do.

I saw the other side of this with a SaaS account that changed its offer every quarter. The old build had 40 ad groups pointing at pages that no longer existed, three duplicate conversion actions firing on the same thank-you page, and a shared budget starving the one campaign that produced SQLs. Patching that setup would have taken longer than rebuilding it clean. The case for a rebuild was not that the account looked ugly. Its structure and measurement blocked the business goal.

Rebuild when the container blocks measurement or budget control, never because a new logo wants its own naming convention. Before anyone starts, they should name the defect, the proposed replacement, the result that would justify the disruption, and the point at which they would roll back. That is a structural decision. “We always set accounts up this way” is a preference.

Put the takeover proposals next to each other

When you compare agencies, ask for first-month plans you can place side by side. The strongest proposal should name what stays untouched, not just what changes. Ask each provider to identify the campaigns, keywords and conversion actions it would preserve for the first 30 days, with a reason tied to performance. For example: this campaign produced 42 SQLs at $310 CPA over the last 90 days, so it stays while we inspect the rest. An agency that can audit and then manage the account should be able to show its diagnosis first. That is the standard I use in how to actually audit an account.

Then make each provider answer the same questions:

  1. What changes in week one, and what waits until week four? Which finding triggers each move?
  2. What gets protected while you test? Name the campaigns and the business result you are unwilling to put at risk without a reason.
  3. What forces a rollback? Set a CPA or ROAS threshold and say how quickly you would act.
  4. What requires a structural reset? Identify the defect, the proposed fix and why the expected business result justifies the disruption.

This is not paperwork for its own sake. It prevents a provider from calling every change an improvement after the fact. It also exposes the opposite bad pitch: the agency that keeps everything running, changes two ad titles a month and sends a polished PDF about impression share. That is babysitting with better fonts. If your ads spend money without bringing in qualified leads, preservation without diagnosis is surrender. A campaign that produced zero SQLs in 90 days does not deserve protection. It deserves an autopsy with dates, search terms and conversion paths attached.

Surgeon repairing one part of a running ad machine while an unused wrecking ball waits nearby

What I want instead is simple: stabilize, then test in isolation. Keep campaigns with proven SQLs live. Fix tracking and irrelevant traffic immediately. Give a structural experiment its own campaign, budget and CPA target where the account allows it, so you can see what changed without sacrificing everything else at once.

That is the operating order behind groas as a fully autonomous growth engine: continuous execution on bids, keywords and budgets, with a named strategist accountable for direction and results. Changes are logged with reasoning you can read. The flat fee replaces a percentage-of-spend or billable-hours incentive. None of that makes an audit optional. It makes the audit the first test of whether the provider understands the account it is about to manage.

Buy a defensible transition, not a clean slate

If your ads spend every week and bring back no real leads, you do not need a new structure first. You need someone to show you where the money went. In my takeovers, that meant pulling 90 days of search terms, tying conversions back to CRM stages and looking for familiar leaks: broad match buying job seekers and DIY researchers; brand mixed with cold prospecting so brand hides the waste; micro-conversions counted equally so Smart Bidding chases brochure downloads. A generic score will not trace those paths, which is why automated audit grades do not fix campaigns.

Take an account spending $20k a month for 15 junk form fills. Before anyone proposes new campaigns, I want to know which queries bought those forms, which actions bidding treated as conversions, and whether any existing campaign already produces pipeline. Cutting wasteful ad groups and correcting the conversion goal may be more useful than rebuilding the campaigns that work. If it is not, the investigation should show why.

Judge companies that offer both an audit and ongoing management by the audit itself. A useful one names what stays, what pauses on day one and what evidence would force a rebuild. A serious mid-market transition playbook puts dates, baselines and change history ahead of a first-month plan with rollback rules. groas is built for that order: it preserves working history, executes continuously, and pairs that execution with a named strategist and readable action logs. Its flat-fee arrangement is listed on pricing. The point is not to buy a prettier account. It is to buy decisions someone can defend.

Hire the demolition pitch anyway and you may pay twice: once for the rebuild, in fees and meetings and new naming conventions, and again while changed campaigns relearn and you wait to see whether the new setup brings in qualified leads. I have watched that movie from both sides of the agency relationship. The familiar ending is a clean account, a dirty CPA and another explanation about Learning.

Get the takeover plan in writing before you sign. Ask for keep, kill and rebuild lists, with the CPA or other business result each protected campaign earns. If an agency cannot tell you what it will leave alone and what evidence would change its mind, it has not done the diagnosis yet. Walk away until it has.