The audience test that paid for itself: 30 days in Plyto's own ad accounts
Nov 10, 2026

The number I did not expect
Four of the five campaigns we ran in the last 30 days produced either zero or one conversion. The one labeled "Already running ads" spent $166 and brought back 4 conversions on its own. Our broader conversions campaign did 5 on $259. Everything else, combined, spent $323 and produced 1 conversion.
That is the story of the month. We spent $748 across five Google Search campaigns, got 34,586 impressions, 464 clicks at a $1.61 average CPC, and 10 conversions. And the audience split told us more than the totals did.
What each campaign actually did
Here is the full picture, ad-account only, aggregated across the last 30 days:
- Oct 30 test, Already running ads: $166 spent, 9,147 impressions, 105 clicks, $1.58 CPC, 1.15% CTR, 4 conversions
- Oct 30 test, Agencies: $67 spent, 1,904 impressions, 27 clicks, $2.48 CPC, 1.42% CTR, 0 conversions
- Oct 30 test, In-house marketers: $114 spent, 4,833 impressions, 59 clicks, $1.93 CPC, 1.22% CTR, 0 conversions
- Oct 30 test, Local service owners: $142 spent, 7,389 impressions, 74 clicks, $1.92 CPC, 1.00% CTR, 1 conversion
- Plyto campaign, Conversions: $259 spent, 11,313 impressions, 199 clicks, $1.30 CPC, 1.76% CTR, 5 conversions
The Conversions campaign is the cheapest clicks and the highest CTR. The Agencies audience is the most expensive clicks and zero conversions on 27 of them. Local service owners had the lowest CTR at 1.00% and got one conversion out of 74 clicks. In-house marketers clicked at a 1.22% rate and converted nothing on 59 clicks.
If I was only looking at CTR, I would have said Agencies was fine. 1.42% is not bad. But 27 clicks at $2.48 each with no conversion is the most expensive silence in the account.
Where the clicks landed
Landing page behavior tells the second half of the story. Visits, reads past 15 seconds, and users who started typing in the signup field:
- / : 293 visits, 156 read, 18 started
- /go/chatgpt-ads : 196 visits, 108 read, 27 started
- /go/local : 87 visits, 43 read, 2 started
- /go/marketers : 76 visits, 37 read, 2 started
- /go/agencies : 41 visits, 22 read, 1 started
- /lp/cu9oc8cefbem : 26 visits, 22 read, 0 started
Look at /go/chatgpt-ads. 196 visits, 108 of them read past 15 seconds, and 27 people started typing into a signup field. That is a 13.8% start-to-type rate off the visit, the highest of any page by a wide margin. The homepage got more traffic (293 visits) and more reads (156), but only 18 starts.
Meanwhile /go/local, /go/marketers, and /go/agencies all show the same shape: decent read rates, almost no one types. 43 of 87 read /go/local. Two started typing. That is not a traffic problem. That is a page that does not convince the reader to act.
The quiet one is /lp/cu9oc8cefbem: 26 visits, 22 reads, 0 starts. A 22-of-26 read rate is extraordinary. Zero type-starts means whatever that page is doing, it is not asking for the signup clearly enough.
Signups
39 signups in the window. 36 came from ad clicks. 3 did not. Ads are doing the work. The organic and referral floor is small.
What I would do tomorrow if this were your account
1. Kill or rebuild the audience-split test, keep the thematic one.
The "Already running ads" audience converted at 4 on $166. The "Conversions" campaign converted at 5 on $259. Together that is 9 of 10 conversions on $425 of the $748 spent. The Agencies, In-house marketers, and Local service owners splits ate $323 and produced 1 conversion. If I had pulled those three on day 10 and rolled the money into the two winners, I would be sitting on roughly 50% more conversions for the same spend. The lesson is not that those audiences are bad forever. It is that a 30-day test with 27 to 74 clicks per arm is not enough signal to keep paying $1.90-plus per click on silence.
2. Clone the /go/chatgpt-ads pattern onto the audience pages.
27 type-starts from 196 visits is the page in the account that is working hardest per visitor. The /go/local, /go/marketers, and /go/agencies pages get read and then abandoned. Whatever /go/chatgpt-ads does in its first screen, its signup placement, or its promise, copy it onto the audience pages before spending another dollar sending traffic there.
3. Figure out what /lp/cu9oc8cefbem is doing right and wrong.
22 reads out of 26 visits is the best engagement in the account. Zero signup starts is the worst conversion in the account. That is a page that earns attention and then wastes it. Either the CTA is missing, buried, or pointed somewhere else. Fix the ask and you will likely see the best-performing page in the next window.
The takeaway for a small business running their own ads
Do not average your account. Average hides the two campaigns carrying the load and the three that are quietly bleeding. Look at conversions per campaign, type-starts per landing page, and cost per click by audience. Three columns, read weekly. That is how you find the $323 you are wasting.
Plyto is your AI marketing assistant: it makes and runs your ads
It writes the ad, builds the landing page, captures the leads and shows you which dollar came back. Try it free, no credit card.
Start free