When I take over an account, I don't fix it first. I ask if it should exist.
The first thing I do with a new account isn’t a fix. It’s deciding whether the whole structure deserves to survive, and most of the time it doesn’t.
That sounds harsh, so let me explain where it comes from. Almost every account I inherit arrives the same way: someone hands me the login, points at the campaigns that stopped spending, and asks me to get them bidding again. The obvious move is to go in and fix the visible thing. Raise the cap, repair the bid strategy, swap the broken ad. The account starts moving, the client sees life on the graph, everyone’s happy for a week.
I’ve learned to distrust that week.
Here’s the one I think about. An account came to me where the campaigns had stopped bidding entirely, and the previous fix had been a clean API change that nudged the bid targets back into range. Technically it worked. The campaigns bid again, spend resumed, the line went up. But the moment I actually read the architecture instead of the symptom, the picture fell apart. The brand campaign was running on a target-CPA strategy when brand should sit on impression share, holding 95% of the absolute top. The keywords were a soup of broad, exact, and old broad-match-modifier syntax, the kind with the plus signs that Google quietly converted to plain phrase match back in February 2021, so it hadn’t meant what the previous manager thought it meant for years. There was no location extension on a business that lives or dies on being local. Two sitelinks where there should be at least four. A phone number in the account that no longer rang the right desk.
None of that shows up when you fix the bid. The account bids beautifully into a structure that was wrong the whole time.
A fix inside a broken architecture isn’t a fix, it’s a faster way to spend money the wrong way.
So I run a different first pass on anything inherited, and it’s deliberately not a quick-win hunt. Before I touch a single mutation I work through the same questions. Does the bid strategy match the campaign’s actual role, or is brand sitting on tCPA because someone copied a template in 2019? Do the match types still mean what they say, given the BMM-to-phrase change? When the company name is also the service name (think of any “Plumber Rotterdam” type business), is the brand-versus-generic line drawn by match type and negatives, or has it collapsed into one undifferentiated bucket? Are the ad groups tight enough that I could write a specific ad for each, or are they too broad to say anything sharp? And is the account even asset-complete, four sitelinks and up, six highlights and up, a location extension where the business is locally served, a phone number I’ve checked against the site and the Google Business Profile, a price asset where the site shows fixed prices?
Only after that do I let myself think about optimizations. The order matters more than the individual checks. If I let myself start with quick wins, I’d polish the symptoms of a structure I’ve already decided shouldn’t exist, and I’d feel productive doing it.
One caveat I’d offer, because I don’t want this to read as “rip everything out on day one.” This only applies to accounts I’m inheriting. For the accounts I’ve built myself, the structure is the starting point, and incremental tuning is exactly the right altitude, going architecture-first there would be me re-litigating my own decisions for no reason. The takeover mindset is specifically a takeover mindset.
If you manage other people’s accounts for a living, I’d encourage you to separate the two moves in your own head. Fixing what’s broken inside a structure, and deciding whether the structure earns the right to keep existing, are not the same job, and they don’t happen in the same week. Structure first. Symptoms can wait.