Before you touch a bid strategy in an inherited Google Ads account, open the conversion action list. It is usually the reason the bidding looks broken.
A Google Ads conversion action audit is the first job in any B2B account you inherit, and it comes before bidding, budget or structure changes. Most inherited accounts carry years of conversion actions that nobody chose on purpose, and several of them are quietly telling Smart Bidding what a good click looks like.
This guide is for the marketing lead or new agency taking over an account with a messy tracking history. You will get a way to sort every action into one of four decisions, a firm opinion on how many should steer bidding, and the order to make the changes in so you only reset learning once.
Why the conversion action list comes before any bidding change
Smart Bidding optimizes toward whatever you tell it counts. It has no idea that one of your leads is a newsletter signup and another is a demo request from a buying committee. If both are set to count, it treats them as the same win and goes looking for more of the cheaper one.
That is why so many takeovers start in the wrong place. The new team sees a high cost per qualified lead, blames the bid strategy, switches it, and waits two weeks. Nothing improves, because the strategy was never the problem. The instructions it was given were.
I spent nearly a decade at Google looking at ad accounts, and the pattern that repeats most often is this one. The list gets longer every time someone changes a form tool, adds a chat widget or connects a new platform. Nobody ever takes anything off it.
Ready for paid ads that pay off?
Book your free auditWhat an inherited conversion action list usually contains
Open the conversions summary in almost any account that has changed hands two or three times and you find the same archaeology. It is common to see twenty or more actions where three would do.
→ A form submission from a form tool the website stopped using two redesigns ago.
→ A second lead action for a platform-hosted lead form, sitting next to the website form with a nearly identical name.
→ Events imported from Google Analytics that duplicate a tag which already fires on the same submission, so one lead counts twice.
→ A thank-you page trigger that also fires when someone lands on that page from a bookmark, an email link or a site search.
→ Lifecycle stages imported from the CRM under an old stage model the sales team replaced last year.
→ Chat starts, pricing page views and product analytics events that were added for visibility and never fired reliably.
Google does archive some of this for you. Its Remove inactive conversions page says "conversion actions that haven't recorded data in the past 13 months will be removed", and that removed actions "will be archived and you can re-enable them anytime."
The dangerous actions are the ones that still fire. A dead action is clutter. A live action that counts the wrong thing is steering your spend every day, and the 13-month sweep never touches it.
The two conditions that let an action steer bidding
A conversion action only influences bidding when two things are true at once. Google's About conversion goals page spells it out: "The conversion action must be set as Primary" and "The campaign must use the goal that contains the conversion action for bidding."
That second condition is where most audits go wrong. Conversion actions are grouped into goals by category, so a newsletter signup and a demo request filed under the same lead category sit in the same goal. If that goal is an account-default goal, every campaign bids toward both.
The distinction between primary and secondary is the lever you actually control. Per Google's About primary and secondary conversion actions, secondary actions "are for observation only" and report in the All conversions column, "but not for bidding, even if the goal they are included in is used for bidding." The same page adds one exception worth knowing: a secondary action inside a custom goal is used for bidding.
So the audit is really one question asked of every row. Should this action be allowed to tell Google what a good click looks like?
Sort every conversion action into one of four decisions
Go down the list one row at a time and give each action exactly one of these outcomes. Write the decision next to the name before you change anything in the account.
Target. The single action that best predicts revenue and fires reliably. In most sales-led B2B accounts this is a CRM milestone such as a sales-accepted or sales-qualified lead, sent back through an offline conversion import. Set it to Primary.
Observe. Real signals you want to watch but never bid on, such as raw form fills, demo page visits or free trial starts in a sales-led motion. Set them to Secondary so they stay in All conversions for reporting.
Remove. Anything tied to a retired tool, a stage model nobody uses or an event you cannot explain. Change the status to Removed. Google archives it rather than deleting it, so the history is still recoverable if someone objects later.
Fix. An action that measures the right thing but fires wrongly, like a thank-you page trigger with no form behind it. Leave it as Secondary until the tag is repaired and verified, then decide whether it earns Target.
Most lists end up with one Target, a handful of Observe rows and a surprising number of Removes. If the Remove pile is small, you probably are not being honest about what nobody uses.
How many primary conversion actions should a B2B account have?
One, in most B2B lead generation accounts. Two at the very most, and only when both mark the same stage of the same buyer journey, for example a website demo request and the equivalent call from a call extension.
Google hints at the same rule. The conversion goals page recommends that if you have "multiple standard conversion goals tracking different stages of the same conversion funnel", you set only one of them as the account-default goal. I would go further and apply that thinking to actions, not just goals.
The math is why. Say an account records 40 demo requests and 300 newsletter signups a month, and both are Primary. Smart Bidding sees 340 conversions, and roughly 88% of them are people who will never take a sales call. You have told it, very clearly, to find more newsletter readers.
Some teams push back that one primary action means too little volume for Smart Bidding. That is a real constraint, and the answer is to pick the earliest stage that still predicts revenue, not to pad the column with actions that do not. The reasoning behind that trade is covered in our post on Smart Bidding for B2B.
Set lead actions to count one conversion per click
The counting setting is the quietest inflation in most inherited accounts. Google's About conversion counting options page describes One conversion as "a good choice if you're not interested in the number of sales, but instead whether or not a certain kind of lead was generated."
Lead actions set to Every conversion count a prospect who submits the same form three times as three leads. On a B2B site where people resubmit to change a meeting time or correct an email address, that adds up fast. Set every lead action to One, and keep Every for genuine repeat purchases only.
Rename every action so reports stop disagreeing
Half the reporting arguments we see trace back to two actions with nearly identical names counting different things. One report pulls the old one, a dashboard pulls the new one, and a meeting gets spent on which number is real.
The fix is boring and it works. Rename every surviving action to say where it comes from, what it counts and what role it plays, for example "CRM, sales-qualified lead, TARGET" and "Website, demo form, OBSERVE". Anyone opening the account a year from now can tell what steers bidding without asking.
It also makes the monthly readout cleaner. If you report paid performance on a fixed set of reporting pillars, the pipeline number should come from the action labelled TARGET, every month, with no debate.
Make all the changes in one sitting, then leave them alone
Batch the whole cleanup into a single change, dated and announced. Google's primary and secondary page warns that "Smart Bidding strategies require a standard 7-14 day learning phase" and says to avoid frequent changes to "conversion goals during this period, as these adjustments will reset the learning window".
That warning is the argument for doing it once. Switching one action this week, another next week and a counting setting the week after resets learning three times and makes every weekly number unreadable. One batch means one reset.
Tell stakeholders before you do it. Reported conversions will drop, sometimes sharply, because you just stopped counting things that were never leads. That drop is the cleanup working, and it needs to be written down as a deliberate change before someone reads it as performance falling off a cliff.
Then protect the result. A list this clean degrades the moment someone connects a new tool and a fresh import lands as Primary by default. Pair the audit with ongoing conversion tracking monitoring so a new action gets reviewed before it gets a vote.
Where this leaves you
If you are taking over a Google Ads account this quarter, do this before any bid strategy change.
1.) Export the full conversion action list with status, primary or secondary setting, goal, counting option and 30 days of conversions.
2.) Write Target, Observe, Remove or Fix next to every row.
3.) Confirm exactly one Target, set to Primary, inside a goal your campaigns actually bid toward.
4.) Set lead actions to One conversion and rename every survivor.
5.) Make all the changes on one day, log the date and warn the people who read the reports.
6.) Wait two full weeks before judging the bid strategy at all.
If you want a second set of eyes on an inherited account, this audit is the first thing we do in our B2B Google Ads management work, before we change a single bid.
