The account you just took over is a record of decisions you did not make. Your first 90 days are worth more spent reading those decisions than overwriting them. Most of what looks broken is a signal someone chose on purpose, not a mistake you can safely delete.
You are new in the seat. There is pressure to prove you were the right hire, and the fastest way to look decisive is to restructure in week one. That instinct is the expensive one.
The dashboard does not explain the history behind it. The account holds months of calibration you cannot see, and the platform has already learned from it. Change it before you can read it, and you throw away the one asset that took the longest to build.
The short version:
- An inherited account carries decisions and constraints you did not set. Read them before you change them.
- Smart Bidding runs on the conversion actions marked in the Conversions column. The most common inherited fault is optimizing to the wrong signal.
- Structural changes trigger a new learning period. Adjusting a target does not. Know the difference before you touch anything.
- The real growth constraint is often the landing page, the tracking, or the offer, not the ad platform.
- The first 90 days are a diagnosis, not a rebuild. Change second.
Start with what the account is optimizing toward, not what it looks like
A dashboard can look healthy while your pipeline stalls. Those are two different questions, and the gap between them is where inherited accounts hide their problems.
Smart Bidding optimizes toward the conversion actions you put in the Conversions column, and nothing else. That is Google's own account of how Target CPA bidding works: including an action in that column is what tells the algorithm to buy more of it.
So the real question is what the account counts as a conversion. Often the counted action is a form-start or a content download, something a lukewarm visitor does without intent to buy. Engagement events like scroll depth and time on site misdirect Smart Bidding and do not belong in the core conversion actions that drive it, as Ameet Khabra wrote in Search Engine Journal in July 2025.
Across the accounts we run on Google and LinkedIn, the most common first fault in an inherited account is the wrong optimization signal. Reported numbers look healthy. Pipeline does not move. The platform has spent months buying the cheapest, lowest-intent action it was told to value.
Emily Wood makes the same point about LinkedIn in her 30-minute LinkedIn Ads audit. "I don't trust performance until I know the tracking is working," she writes. Page views or clicks counted as conversions are a red flag, and, as she puts it, "Whatever's spending the most is where problems cost the most."
The difference between adjusting an account and resetting it
Adjusting an account and resetting it are different actions with different costs, and most people conflate them.
Adding or removing campaigns, ad groups, or keywords, or switching your bid strategy type, moves the account into a Learning status. That status can take up to three weeks, or one to two conversion cycles, to clear, per Google.
Changing the target value on a CPA or ROAS strategy behaves differently. A target change does not trigger a learning status and does not reset anything Smart Bidding has already learned about your account. Google states that directly.
That single distinction should govern your first month.
Source: Google Ads Help, learning-period documentation.
The inheriting owner's mistake is to restructure in week one to clean things up. That erases weeks of calibration exactly when you understand the account least.
Read the account in order
There is a reading order, and each step depends on the one before it. Skip ahead and you will make confident changes on top of a misread.
The conversion signal first
Start with what the account counts. Look at every conversion action, then find which ones are marked primary and feed bidding. Separate real buyer intent, a demo booked, a trial started, an opportunity created, from engagement that only looks like progress. If the primary signal is engagement, the account has been optimizing away from pipeline the whole time.
Then the intent map
Next, read how the account expresses buyer intent. Check the campaign structure, the match types, and how brand, non-brand, and competitor traffic are separated. Watch for MOFU and BOFU blur, where an ad promises a demo and the click lands on a PDF. Watch for fragmentation that starves learning. Google recommends at least 30 conversions a month, per ad group, for consistent performance from a target. That is a recommendation, not an on-off switch. Below it, Google says the learning period can run up to four weeks and performance volatility is medium to high, up to 100%. Split one funding source across too many campaigns and each one drops under that line.
Then the constraint
Only now look for the real limit. It is often the landing page, the tracking, or the offer, not the ad platform. A page that loads slowly or contradicts its ad will cap performance no matter how well the bids are set. The same is true of an offer that asks for too much too early in the buyer's journey. Bidding cannot fix a promise the page does not keep. Name the constraint before you touch a single bid.
What is normal, and what actually slows you down
Some of what alarms a new owner is just how paid media works.
Most experiments do not work. Across the accounts we run, only about two in ten move the needle, and you cannot know which two in advance. A flat stretch is not proof the account is broken. It is often the cost of learning.
Accounts left on set-and-forget decay quietly. Auctions shift, competitors change their bids, and creative fatigues while the dashboard still reads fine. Nobody sees it until pipeline does.
Budget and cycle time set your pace together. Budget decides how fast data arrives. Cycle time decides how fast you act on it. A bigger budget only pays off as fast as your loop can use the data, so throwing spend at a slow loop mostly buys noise.
Turnover is part of the backdrop. Average CMO tenure across S&P 500 companies is 4.1 years, according to Spencer Stuart's 2026 study, and each handoff at that level eventually resets who owns the paid program below it. Inherited accounts are the norm, not the exception.
The audit costs you time before it saves you any. That upfront time is the price of keeping the account's learning intact. A week spent confirming what the numbers mean is cheaper than a month spent unwinding a change you made too early.
When to finally change something
Reading first earns you the right to move fast later.
Some fixes are safe almost immediately. Adding negative keywords, or repairing a landing page that contradicts its own ad, can move quickly and rarely disturbs the algorithm. The high-impact rebuilds come later, after you understand buyer intent and where each offer sits in the funnel.
This is the discipline Melissa Mackey of Compound Growth Marketing describes in Optmyzr's Google Ads audit guide: when you start managing a new account, run a comprehensive review of historical performance data before you make changes.
When you do change, protect the learning. Adjust targets before you restructure, since a target change keeps what the algorithm knows. Consolidate where budget and unit economics are shared rather than splintering campaigns and starving each one of data.
A 90-day read used to mean 90 slow days. It no longer does. The research and landing-page work that once made the loop slow are now fast. AI does them in a fraction of the time. What stays human is judgment: deciding what a result means and what to do about it. Smaller and faster beats bigger and slower.
FAQ
Should I pause the worst-performing campaigns on day one?
Rarely. "Worst-performing" usually means measured against the wrong conversion signal. Removing a campaign can also trigger a new learning period across what remains. Read what the account optimizes toward first, then pause only what is clearly wasting spend against real buyer intent.
How long before I can trust what I'm seeing in an inherited account?
Trust the tracking before you trust the numbers. Once you confirm the Conversions column reflects real buyer intent, give any change one to two conversion cycles to clear its learning period. Emily Wood puts it plainly: she does not trust performance until she knows the tracking works.
What is the single most common problem you find in an account you take over?
The wrong optimization signal. Across the accounts we run on Google and LinkedIn, the counted conversion is often a low-intent action like a form-start or a download. Smart Bidding then buys the cheapest version of that action, so reported numbers look healthy while pipeline stalls.
Will restructuring the account reset Google's learning?
Structural changes will. Adding or removing campaigns, ad groups, or keywords, or switching your bid strategy type, triggers a learning status. Per Google, it can take up to three weeks to clear. Changing a target value does not reset what Smart Bidding has already learned.
How many conversions does the account need for automated bidding to work?
Google recommends at least 30 conversions per month, per ad group, for consistent target performance. Treat that as a recommendation, not an activation minimum. Below it, Google says learning can take up to four weeks and performance can swing, medium to high, up to 100%.
Do the same rules apply to an inherited LinkedIn Ads account?
Yes, on LinkedIn's own mechanics. The discipline holds: verify the tracking, confirm what counts as a conversion, and find the real constraint before restructuring. As Emily Wood notes, whatever is spending the most is where problems cost the most, so start your audit there.
The account is yours to read first
The account is not yours to overwrite yet. It is yours to read. Before your first big change, map two things: what the account counts as a conversion, and how its structure reflects real buyer intent. Do that, and your first 90 days compound instead of reset. Every good decision after that starts from evidence the account already holds.
This is the work Thunder does as an operating layer for paid media. We run Google and LinkedIn end to end. One accountable operator, AI-enabled and embedded, reads the inherited account and runs the improvement loop toward qualified pipeline and revenue. Our team helped build the LinkedIn Ads platform and has run paid for Reddit, Gusto, Warp, and Linear. Our fee is decoupled from media spend, so more of your budget goes to growth.
See how Thunder runs paid media end to end.