Google Ads Reported 444 Conversions. None of Them Were Signups.

A software client’s Google Ads account closed its flights on September 7 with a summary line most people would screenshot and send around.

$470.83 spent. 2,726 clicks. 444 conversions.

Every one of those 444 conversions was a YouTube channel subscription. The thing the business actually needed was people registering for a free trial on its website. Not one of the 444 was a registration, and Google Ads had no way to count one even if it happened.

The subscriptions were real. We checked. They just weren’t the goal. That gap, between a real conversion and the right conversion, is what this post is about.

The numbers, campaign by campaign

The client isn’t named. The figures were read off the Google Ads campaigns table on September 8, 2026, with the All time range. The account’s first impression was August 5, and the last flights ended September 7.

Launch, website clicks (Demand Gen, Maximize clicks), August 5 to 7: $113.38, 997 clicks, landing on the homepage. Zero conversions.

Launch, subscribers (Demand Gen, Maximize conversions), August 5 to 6, then August 31 to September 7: $214.25, 597 clicks, 444 conversions at $0.48 each. The short August run recorded none. All 444 came in the eight day run that ended September 7.

Free trial, website traffic (Demand Gen, Maximize clicks), August 31 to September 7: $143.20, 1,132 clicks at a 2.46% click rate and $0.13 a click, landing on the signup page with the free plan already selected. Zero conversions.

Total: $470.83, 157,487 impressions, 2,726 clicks, 444 conversions, and a blended cost per conversion of $1.06.

Look at that last number for a second. $1.06 a conversion is an average across three campaigns, and two of them weren’t bidding on conversions and recorded none. It’s arithmetic describing nothing.

What a “conversion” meant in this account

We opened the conversion actions table to find out what Google was actually counting. The account had three conversion actions:

• YouTube channel subscriptions. Source YouTube hosted, status Active, set to Primary, counting set to One, 30 day click window. This one carried every conversion on the account.
• YouTube follow on views. Set to Primary. Never recorded anything.
• Android installs. Set to Secondary. Never recorded anything.

There was no conversion action for a trial registration. None. The site had Google Analytics installed, but no key events were configured on the property, so there was nothing to import into Google Ads either.

So when the account said “444 conversions,” it meant exactly one thing: 444 times, Google attributed a YouTube subscribe to someone who’d interacted with a paid ad.

How Google decides what a campaign chases

This is the mechanism, and it’s worth getting precise because it’s where a lot of accounts quietly drift. Everything below is from Google’s own Ads Help pages.

Primary actions are the ones that bid. Google says primary conversion actions are reported in the “Conversions” column and used for bidding, as long as the goal they belong to is used for bidding. Secondary actions are “for observation only.” They show up in the “All conversions” column and don’t steer bids. The one exception is a custom goal, where a secondary action does get used for bidding.

Two conditions have to be true for an action to drive bids. The action has to be set to Primary, and the campaign has to use the goal that contains it. A primary action sitting in a goal no campaign selects isn’t used for bidding, though Google notes it may still be used to enhance predictions.

Account default goals spread. Google describes account default goals as the goals you’d like all campaigns to bid toward automatically, and when you create a new campaign, all of them are selected for optimization by default. Put a cheap engagement action into an account default goal and every new campaign starts life chasing it unless someone changes the setting.

Maximize conversions spends to get as many conversions as possible. That’s Google’s description, and it’s literal. It doesn’t know which conversions matter to your business. It knows which ones you told it count.

Put those together and you get the trap. The algorithm optimizes toward whatever sits in the Conversions column for that campaign. If the cheapest real thing in that column is a YouTube subscribe, the campaign gets very good at buying YouTube subscribes. It isn’t broken. It’s obedient.

What we got right, and what we’d do differently

The subscriber campaign did its job. It was built to grow the channel, pointed at the channel, and bid on subscriptions on purpose. Nobody tricked anyone. At $0.48 a subscriber across its life, it bought exactly what it was set up to buy.

We didn’t call them subscribers until we’d opened the table. Our internal rule for the first days of the September run was simple: don’t describe these conversions as subscribers yet. It lifted on September 5, when the conversion actions table showed a single YouTube hosted action carrying every conversion and the All conversions total matched the campaign table. Read the action before you name the number.

The mistake was letting the account total become the headline. “444 conversions” sat at the top of the account next to 2,726 clicks, and 2,129 of those clicks came from the two campaigns that weren’t bidding on conversions and recorded none. A summary row like that invites the wrong conclusion: that the whole program converted, and at about a dollar each.

The free trial campaign flew blind, and we knew it. It ran on Maximize clicks because there was no registration conversion to bid toward, and with no conversion history there was nothing for Smart Bidding to learn from anyway. But it means Google was never told what a signup looks like, so it had no way to find more people likely to complete one. If the business goal isn’t a conversion action, the platform can’t aim at it.

Clicks weren’t visits, either. Across August 10 to September 8, Analytics recorded 439 sessions from Google paid traffic averaging 2 seconds of engagement, while the free trial campaign alone had bought 1,132 clicks pointed at the site. Google’s own description of Demand Gen is campaigns that capture engagement across YouTube, Discover, Gmail, Maps and the Display Network. A click on that inventory is an interaction, not a person sitting on your signup page.

What we’d change: build the registration conversion first, even if it’ll only fire a few times a week. Mark it as a key event in Analytics, import it, set it to Primary, and put it in the goal the website campaigns use. Keep the YouTube subscription action where it belongs, on the one campaign meant to buy it. And report every campaign on its own row.

For how we structure and govern paid media, our paid advertising service page covers it. If you want to know what your own conversion column is actually counting, run a free audit.

Real but wrong versus fake

When a conversion number looks too good, people jump to “it’s fake.” Sometimes it is. Here it wasn’t, and that changes what you do next.

Fake means the count doesn’t correspond to anything that happened: a tag firing twice, a bot, a thank you page loading on refresh, a counter with no record behind it. We’ve written about that problem in The Dashboard Said One Lead. There Was No Lead.

Real but wrong means the thing happened, and it just isn’t the outcome that pays the business. You check it differently:

1. Open the conversion action itself. Name, source, Primary or Secondary, counting (One or Every), and click window. If the source is YouTube hosted, the conversion happened on YouTube, not on your site.
2. Segment the campaigns table by conversion action. Google’s help pages show this under Segment, then Conversions, then Conversion action. It tells you which action produced each number, per campaign.
3. Compare Conversions to All conversions. If All conversions is bigger, something is recording that isn’t steering bids, and you should know what it is.
4. Check the campaign’s goal setting. Account default or campaign specific, and which goals are included. Hover each goal to see its primary actions.
5. Find a witness outside the ad platform. For subscriptions, that’s the channel itself.

We did step five. The channel’s own subscriber count went from 170 on August 31 to 461 on September 8. Across July 25 to September 8, 96.3% of the channel’s views came through advertising, and the channel gained 538 subscribers and lost 135. That’s real growth from real people, on a channel where paid views were almost all of the traffic.

The channel’s net gain doesn’t equal Google’s 444, and you shouldn’t expect it to. Google counts a subscribe it attributes to a paid ad interaction inside its conversion window. The channel number is net of unsubscribes and includes every subscriber from every source. They’re two different measurements of overlapping activity. We report the channel’s number as growth and Google’s as the attributed action, and we don’t add them together.

What we could say to the client, and what we couldn’t

We can claim the subscribers. The channel grew from 170 to 461 subscribers between August 31 and September 8, at $0.48 per attributed subscribe across the campaign’s life, and the channel’s own count backs that up.

We can’t claim signups. Google Ads had no registration action, so it attributed zero registrations to any campaign. The site’s own registration tracking recorded no registrations from August 31 through September 6, and one on the evening of September 7 with no source attached. One unattributed registration can’t be credited to the ads, and we won’t present it as one.

We won’t quote the blended $1.06. It averages a campaign that converted with two that recorded nothing.

The honest summary reads like this: the ads bought channel growth cheaply and bought no measurable trial registrations. Both halves go in the report. Leaving out the second half would make the first half a lie by omission.

The flights ended on their own schedule on September 7. Nothing restarted automatically.

What this doesn’t mean

It doesn’t mean YouTube subscriber campaigns are a waste. A channel that grows from 170 to 461 in about a week has a bigger audience for every video that follows, and that has value. It just isn’t trial registration value, and it shouldn’t be sold as if it were.

It doesn’t mean Google Ads can’t drive signups. We don’t know yet what the same budget would do bidding on a real registration action, because that test hasn’t run.

And it doesn’t mean the client did anything wrong. A new account starts with whatever conversion actions exist. The fix is setup, not blame.

FAQ

What’s the difference between a primary and a secondary conversion action in Google Ads?

Google says primary actions appear in the “Conversions” column and are used for bidding when the campaign uses the goal they belong to. Secondary actions are for observation only. They appear in “All conversions” and don’t steer bids, unless they’re part of a custom goal.

Why would a campaign report more conversions than it has real sales or signups?

Because a conversion is whatever action you’ve set to count. If a YouTube subscription is a primary action in a goal the campaign uses, it lands in the Conversions column beside your real outcomes, and bidding treats it the same way.

How do I check which conversion actions my campaigns are optimizing toward?

Open the campaign’s settings and expand Conversion goals to see whether it uses account default or campaign specific goals, then hover each goal to see its primary actions. On the campaigns table, segment by conversion action to see which action produced each number.

Are YouTube subscription conversions fake?

Not necessarily. In this account they were YouTube hosted, counted as One per interaction, and the channel’s own count rose alongside them. The better question is whether they’re the outcome your business is paying for.

Can I import signups from Google Analytics into Google Ads?

Yes, if the event is marked as a key event in Analytics and the property is linked to your Google Ads account. Google notes imported conversion data can take up to 24 hours to show up in Google Ads, and conversions created through Analytics come in as secondary, so you’ll need to set them to primary in Google Ads before they drive bidding.

Count the thing that pays

The account did what it was told. It found the cheapest real action in the Conversions column and bought 444 of them. Before you judge a campaign by its conversions, find out what a conversion is. Then make sure the thing that actually pays your business is in that column, and the things that don’t are kept out of the headline.

Want a second opinion on a campaign that isn’t working? Book a 15-minute call. No deck, no fluff. Strategy first. Tactics second. The work works.

Reliable PR & Marketing is a strategy-first marketing agency in Bakersfield, California. We run integrated SEO, PR, web, and content for founder-led companies across Kern County and nationwide. Strategy first. Execution always.