Free resource
The Google Ads Account Audit
Twenty-four checks that find where your B2B software account is losing money.
Before you start
Open your Google Ads account in another tab. This takes about twenty minutes and you can finish it in one sitting.
Work in order. The tracking section comes first, and there is a reason. If your conversion data is wrong, every other number in the account is a lie, and fixing anything downstream is guessing dressed up as work.
Do not fix anything until you have finished. You are diagnosing, not operating. Half the things that look broken further down turn out to be symptoms of something in the first section.
Mark a check N/A only when it genuinely cannot apply to you. N/A is removed from your score rather than counted against you, which means marking things N/A to feel better works exactly as badly as you would expect.
This list is written for companies selling software or a technical product, where the money shows up as trials, demos, and pipeline rather than walk-ins.
Tracking
0 / 6Start here and finish here before you look at anything else. If the conversion data is wrong, every other number in the account is a lie, and optimizing against a lie is worse than doing nothing, because it feels like progress.
Auto-tagging is switched on
Where: Admin > Account settings > Auto-tagging.
If this fails: Without the click ID, Google cannot connect a click to a conversion. Everything downstream becomes guesswork.
The Google Ads Tag and the Conversion Linker both exist in Tag Manager and fire on every page
Where: Tag Manager > Tags. You are looking for a Google Tag carrying an AW- ID on the Initialization trigger, and a Conversion Linker on All Pages.
If this fails: Without the base tag, no conversion fires at all. Without the linker, the click ID is lost on the second page and your conversions get credited to nobody. Both fail silently and look fine from inside Google Ads. If you serve EU traffic, also confirm Consent Mode v2 is configured. A banner that blocks the tag before consent will quietly erase a large share of your conversions, and the account will look like it stopped working for no reason.
Conversion actions fire on a real action, not on a page view
Where: Goals > Conversions. Open each action and check what actually triggers it. If your site or app is a single page application, a thank you URL may never load as a real page. Confirm the event fires on the actual submit or signup, not on a route change you assumed happens.
If this fails: A thank you page view counts anyone who bookmarks or refreshes that page. Your count inflates and the bidding learns from noise.
Count is set to One on every lead conversion, not Every
Where: Goals > Conversions, open each action, the Count setting.
If this fails: Set to Every, one person submitting a form three times reads as three leads, and the bidding starts hunting for people who submit forms repeatedly.
Trial signups, demo requests and low-intent actions are separate conversion actions, and only the ones that produce revenue are Primary
Where: Goals > Conversions > Summary. Read the Goal and Action optimization columns, and check whether each action carries a value.
If this fails: A free trial signup and a booked demo are not the same event and should never share a bidding target. If an ebook download sits in Primary next to demo requests, your budget is buying ebook downloads. And if every action carries the same value, or no value at all, the bidding has no way to tell a $200 self-serve trial from a $40,000 enterprise opportunity.
Google's conversion count matches what your CRM actually recorded
Where: Take one complete month. Pull the real signups and demo requests from your CRM by hand. Compare to what Google reports for the same period.
If this fails: If the two numbers disagree by more than a little, every decision made from the Google number is wrong. This is the check that validates the whole section, and in most accounts it is the one that fails.
The settings Google chose for you
0 / 6Six settings that arrive switched the wrong way, by default or by recommendation. None of them announce themselves. Together they are the most reliable way to spend money on nothing.
Search partners is switched off
Where: Campaign settings > Networks, on every Search campaign.
If this fails: Search partners sends your budget to sites you have never heard of, and gives you no way to see which ones.
The Display Network is switched off on Search campaigns
Where: Campaign settings > Networks, same panel.
If this fails: Display traffic inside a Search campaign converts far worse and drags the entire campaign's numbers down with it.
There are no broad match keywords nobody deliberately chose
Where: Keywords, add the Match type column and sort by it.
If this fails: Broad match arrives by default and through Google's recommendations. It spends fastest on the least relevant terms. In software categories, broad match reliably drifts toward job seekers, students, and people looking for a free alternative to what you sell. Those clicks cost the same as the ones from buyers.
Location options are set to Presence, on both the include and the exclude setting
Where: Campaign settings > Locations > Location options. Check both settings; people fix the first and miss the second.
If this fails: Presence or interest shows your ads to anyone who merely searched about a country you target. If you sell in a handful of markets, you end up paying for clicks from places you cannot sell to, cannot support, and cannot invoice.
All ten account-level automated assets have been reviewed
Where: Assets > Associations > Account-level automated assets.
If this fails: Ten switches, all on by default, on a page most advertisers never open. Google writes its own text underneath your ads.
Auto-apply recommendations are all switched off
Where: Recommendations > Auto-apply, and check both tabs.
If this fails: Google applies changes to your account while you sleep, including budget increases and match type expansions.
The account
0 / 6Six checks that decide whether the money you spend reaches somebody who could actually buy your software, and whether Google ever finds out that they did.
Closed deals are sent back to Google, not just form fills
Where: Goals > Conversions > Uploads, or check whether Enhanced Conversions for Leads is enabled. Then confirm the GCLID is captured on your form and stored on the record in your CRM.
If this fails: This is the single biggest gap in most B2B software accounts. Google optimizes toward whatever you tell it a conversion is. If the only thing you tell it is "a form was submitted," it will get very good at finding people who submit forms and will never learn the difference between a buyer and a tire kicker. Until closed-won data goes back into the account, the bidding is optimizing for the wrong outcome no matter how well the rest is set up.
Brand and non-brand live in separate campaigns
Where: Campaigns. Check whether your own product name appears as a keyword inside campaigns that also target category terms.
If this fails: Brand searches convert several times better than anything else, because those people already decided. Mixed into the same campaign, they inflate the average and hide the fact that your acquisition campaigns are not working. You end up scaling a number that is mostly people who were coming anyway.
Existing customers and current trial users are excluded from prospecting campaigns
Where: Audiences > Segments. You need a Customer Match list of active accounts and trial users, applied as an exclusion on acquisition campaigns.
If this fails: Your own users search your brand name constantly, to log in, to find docs, to check pricing. Without the exclusion you pay for those clicks and then count them as conversions when they sign up for a second trial. It looks like acquisition. It is your existing base, billed back to you.
Each ad group points at its own relevant landing page, not the homepage
Where: Ads, read the Final URL column.
If this fails: Sending every click to the homepage makes the visitor do the work of finding what he searched for. Most of them will not. For software, the homepage is the worst possible destination. Someone who searched for a specific integration or a specific use case lands on a general product tour and has to go hunting. Most will not.
The search terms report has been reviewed in the last ninety days
Where: Campaigns > Insights and reports > Search terms. Last 90 days, sorted by Cost descending. If you run Performance Max, check its search terms separately under the campaign's Insights tab. It is a different report and it hides a different set of problems.
If this fails: This is the report that tells you what you are actually buying, as opposed to what you thought you were buying. Not reading it is the most common single failure in an account that is underperforming.
A negative keyword list exists and has been updated in the last ninety days
Where: Tools > Shared library > Negative keyword lists. Check the last modified date, not just that a list exists.
If this fails: Negatives are not a one-time job. A list untouched for a year is a list that stopped protecting you a year ago.
The page, and the business behind it
0 / 6Half of what looks like an account problem is not in the account at all. These six sit between your form and your revenue, and they cost more than anything above them.
Your conversion window matches how long your deals actually take
Where: Goals > Conversions, open each action, the Conversion window setting. Compare it to your median time from first touch to closed deal.
If this fails: The default window is thirty days. If your average deal takes seventy, Google never sees the conversions it produced, concludes those campaigns failed, and moves budget away from the ones that were working. Long sales cycles get punished by a setting nobody thinks to open.
The form works and the lead lands in your CRM, confirmed by submitting it yourself today
Where: Fill in your own form, then find that record in your CRM with the source correctly attributed.
If this fails: Broken forms and broken CRM routing are far more common than anyone expects, and nothing in the Google Ads interface will ever tell you. A form that submits but drops the lead into nowhere is worse than a form that visibly fails.
Your forms are not filling up with bot submissions
Where: Pull the last month of leads and read them. Look for junk domains, nonsense company names, and identical submission timestamps.
If this fails: Bot and spam submissions are counted as conversions, and the bidding learns from them like they were real. The account then spends more to find more of the same. Free email domains and missing company fields are the usual tell.
Demo requests get a response within five minutes during working hours
Where: Time it honestly for one week, across every timezone you advertise in.
If this fails: Speed to lead beats almost every optimization available inside the account. A five minute reply converts several times better than an hour, and the gap widens the longer you wait. If you advertise in North America from Europe, or the other way round, this is where most of your paid demos quietly die.
You know your cost per SQL, not just your cost per lead
Where: Your CRM, not Google Ads. Take one quarter, count the leads that sales actually accepted, and divide the spend by that number.
If this fails: Google reports form fills. Only your CRM knows which of them sales was willing to touch. The gap between cost per lead and cost per SQL is usually enormous, and every budget decision made from the first number is a decision made blind.
You know your CAC payback period, and it is a number you can live with
Where: Fully loaded acquisition cost divided by gross margin per customer per month.
If this fails: Cost per acquisition on its own tells you nothing about whether the channel is viable. A $4,000 CAC is excellent at a certain contract value and fatal at another. Without payback you cannot answer the only question that matters, which is whether to spend more next month or less.
Your diagnosis
Answer all twenty-four checks to see your diagnosis. 24 to go.