Linking live ad campaigns: To turn signups into customers
Linking live ad campaigns: To turn signups into customers
I spent four months trying to get more people to pay and got basically nowhere. The thing that finally worked was a step we'd been treating as optional.
I spent four months trying to get more people to pay and got basically nowhere. The thing that finally worked was a step we'd been treating as optional.
Problem
People who connect an ad account pay at 20x the rate of everyone else. But it was buried in three different places, none of them said why you'd bother, and only 3% of new users did it on day one.
People who connect an ad account pay at 20x the rate of everyone else. But it was buried in three different places, none of them said why you'd bother, and only 3% of new users did it on day one.
Solution
Rebuild onboarding around the four reports people actually wanted, and land them on home, where the reason to connect already lived.
Rebuild onboarding around the four reports people actually wanted, and land them on home, where the reason to connect already lived.
Impact
Day-one connections went from 3.2% to 14.4%, and the new connectors were better than the old ones.
The research changed who we build for. Our founders took it on as company strategy, not just an onboarding fix.
Day-one connections went from 3.2% to 14.4%, and the new connectors were better than the old ones.
The research changed who we build for. Our founders took it on as company strategy, not just an onboarding fix.
What is Playad and how does it work?
What is Playad and how does it work?
If you want to run ads, you have to learn Meta Ads Manager. Then Google Ads. Then whatever comes next. If you're a small business owner without a marketing team, that's usually where it ends. You either never start, or you run something bad and quietly give up.
Playad puts an AI agent in front of all of it, so you're making decisions instead of learning dashboards.
Connecting your ad account is what lets Playad see your live campaigns. Without it, the agent is just guessing from scanning your website. The website collection is done during signup with a URL submission.
Without connnection
The agent forms a new hypothesis from your website
The agent forms a new hypothesis from your website
In the dashboard
In the dashboard
In chat knowledge
In chat knowledge
With connnection
The agent reads your live results and builds on what's winning
The agent reads your live results and builds on what's winning
In the dashboard
In the dashboard
In chat knowledge
In chat knowledge
I tried two things first. Both were me pushing harder.
I tried two things first. Both were me pushing harder.
Since May I'd been trying to move paid conversion.
Since May I'd been trying to move paid conversion.
A step-by-step walkthrough
A step-by-step walkthrough
About 5% of people touched it. Turns out people show up for pretty different reasons, and plenty of them just want to generate creative. The step-by-step hinders them from reaching their goal quickly.
About 5% of people touched it. Turns out people show up for pretty different reasons, and plenty of them just want to generate creative. The step-by-step hinders them from reaching their goal quickly.
A hard paywall
A hard paywall
Between 49% and 65% of new subscriptions got cancelled within 30 days. User's comes to try it but left not enjoying the creative quality most of the time.
Between 49% and 65% of new subscriptions got cancelled within 30 days. User's comes to try it but left not enjoying the creative quality most of the time.
Both assumed people needed moving along. Neither asked what they'd come for.
Both assumed people needed moving along. Neither asked what they'd come for.
The diary study changed the question I was asking
The diary study changed the question I was asking
I'd been asking how to get people to pay. After a week of watching three advertisers use the product, the question turned into something simpler: what do people actually find useful here?
Vince ran a beauty app spending $50k a month. Xaviera was pre-launch on $250. Sky ran a family photography business in Seattle at about $120 a day.
All three rated analysis and competitor research the highest. Creative generation, the thing we'd built the whole onboarding around, was the weakest part of the product for every single one of them.
I'd been asking how to get people to pay. After a week of watching three advertisers use the product, the question turned into something simpler: what do people actually find useful here?
Vince ran a beauty app spending $50k a month. Xaviera was pre-launch on $250. Sky ran a family photography business in Seattle at about $120 a day.
All three rated analysis and competitor research the highest. Creative generation, the thing we'd built the whole onboarding around, was the weakest part of the product for every single one of them.
Sky, Snapmate Photos
"I don't use the video or image. I like to use analytics. That works pretty well for me."
Xaviera, Kijivu
"Analysis is a game changer for me, then having suggestions of what to do next is the cherry on top."
Sky, Snapmate Photos
"I don't use the video or image. I like to use analytics. That works pretty well for me."
Xaviera, Kijivu
"Analysis is a game changer for me, then having suggestions of what to do next is the cherry on top."
Then I noticed the thing that changed everything, and it was about when they connected their ad account.
Then I noticed the thing that changed everything, and it was about when they connected their ad account.
Xaviera connected at the end of day one and ended up the most habitual user in the group. Sky connected during onboarding, and analytics was the one thing he rated well all week. Vince didn't connect until day five. He spent four days messing with creative and rated his first two days a 1 out of 5 on how he'd feel if he lost us.
Xaviera connected at the end of day one and ended up the most habitual user in the group. Sky connected during onboarding, and analytics was the one thing he rated well all week. Vince didn't connect until day five. He spent four days messing with creative and rated his first two days a 1 out of 5 on how he'd feel if he lost us.
Whoever connected early got to the good stuff faster.
Whoever connected early got to the good stuff faster.
Nobody found the good stuff until day three anyway
Nobody found the good stuff until day three anyway
The reports all three of them rated highest weren't what anyone found first. Xaviera didn't get to competitor analysis until day three, and only because a quest list happened to put it in front of her.
The reports all three of them rated highest weren't what anyone found first. Xaviera didn't get to competitor analysis until day three, and only because a quest list happened to put it in front of her.
Competitor Report
Competitor Report
Performance Report
Performance Report
Xaviera, Kijivu
"I wouldn't know I can do competitor analysis if not for it."
These three had committed to using it for a week. Nobody else does that. Only 6.2% of our signups were still opening Playad on day three.
These three had committed to using it for a week. Nobody else does that. Only 6.2% of our signups were still opening Playad on day three.
It also broke the metric we'd been steering by
It also broke the metric we'd been steering by
We'd decided activation meant five messages to the agent in week one. Part of the study was checking whether that actually held up.
It didn't. Vince sent one message the entire week and finished as our warmest lead. Meanwhile our most activated user by the old definition was pre-launch and spending $250 a month.
We'd decided activation meant five messages to the agent in week one. Part of the study was checking whether that actually held up.
It didn't. Vince sent one message the entire week and finished as our warmest lead. Meanwhile our most activated user by the old definition was pre-launch and spending $250 a month.
We were measuring how much people typed, not whether they'd found anything worth paying for.
We were measuring how much people typed, not whether they'd found anything worth paying for.
There were really two kinds of people signing up
There were really two kinds of people signing up
We'd been building onboarding for both at once, which is why it wasn't landing for either.
We'd been building onboarding for both at once, which is why it wasn't landing for either.
So we stopped building for everyone, and choose to design for the operator
So we stopped building for everyone, and choose to design for the operator
The first group's experience lives or dies on model quality, and sitting around waiting for that isn't a plan. The second group's depends on whether we can show them their own data and tell them what to do with it. They're also the ones already spending money on ads.
This wasn't something I could just decide inside onboarding, so I took the research to our founders and made the case. They went for it, and it became our actual strategy: not just how onboarding works, but who we're building for.
The first group's experience lives or dies on model quality, and sitting around waiting for that isn't a plan. The second group's depends on whether we can show them their own data and tell them what to do with it. They're also the ones already spending money on ads.
This wasn't something I could just decide inside onboarding, so I took the research to our founders and made the case. They went for it, and it became our actual strategy: not just how onboarding works, but who we're building for.
So here's what I actually had to solve
So here's what I actually had to solve
1. People arrive wanting different things. One person wants an image ad, the next wants competitor analysis. Whatever I built had to let them start wherever they wanted instead of marching everyone down one path.
2. Operators don't have time, and some of them barely know how ads work. They came to make decisions, not to manage campaigns. Handing them a pile of tools and wishing them luck wasn't going to cut it.
3. Connecting had to be findable. Performance analysis, the report people rated highest, is gated behind it. And it was sitting in three places that each failed differently: an onboarding modal most people skipped, a chat component that only appears if you ask for performance analysis or try to launch an ad, and a connections page in the side menu you'd only visit if you already knew you wanted it.
The tracking made it obvious. Every single connection start in the product came from that chat component. Home, the page where I'd written the whole case for connecting, had zero, because nobody landed on home. You landed in chat.
1. People arrive wanting different things. One person wants an image ad, the next wants competitor analysis. Whatever I built had to let them start wherever they wanted instead of marching everyone down one path.
2. Operators don't have time, and some of them barely know how ads work. They came to make decisions, not to manage campaigns. Handing them a pile of tools and wishing them luck wasn't going to cut it.
3. Connecting had to be findable. Performance analysis, the report people rated highest, is gated behind it. And it was sitting in three places that each failed differently: an onboarding modal most people skipped, a chat component that only appears if you ask for performance analysis or try to launch an ad, and a connections page in the side menu you'd only visit if you already knew you wanted it.
The tracking made it obvious. Every single connection start in the product came from that chat component. Home, the page where I'd written the whole case for connecting, had zero, because nobody landed on home. You landed in chat.
Connections page
Connections page
Onboarding modal
Onboarding modal
In chat
In chat
First, a task list so people could see what was here
First, a task list so people could see what was here
Not a walkthrough this time. Just things you could do, in any order, with credits for doing them. If you came for competitor research you start there. If you've got live campaigns you start somewhere else.
Not a walkthrough this time. Just things you could do, in any order, with credits for doing them. If you came for competitor research you start there. If you've got live campaigns you start somewhere else.
People who finished a task came back in week one at 13.1%, versus 1.3% for people who didn't.
People who finished a task came back in week one at 13.1%, versus 1.3% for people who didn't.
Then we turned it into five steps to automating your ads
Then we turned it into five steps to automating your ads
A list of stuff to try is still just a list. It shows you the product without telling you where you're going. So we framed the reports as a path to something: automate your paid ads, in four steps.
Connecting ad account. The most important first step is to grab live data and to launch and manage ads.
Competitor analysis. What everyone else in your category is running.
Performance analysis. How your live ads are doing.
Setting your marketing goal. The agent uses this as a benchmark, and the user can see if progress has been made.
A list of stuff to try is still just a list. It shows you the product without telling you where you're going. So we framed the reports as a path to something: automate your paid ads, in four steps.
Connecting ad account. The most important first step is to grab live data and to launch and manage ads.
Competitor analysis. What everyone else in your category is running.
Performance analysis. How your live ads are doing.
Setting your marketing goal. The agent uses this as a benchmark, and the user can see if progress has been made.
Same four tasks, users valued during the diary study.
Same four tasks, users valued during the diary study.
And put connecting where people actually land
And put connecting where people actually land
Two changes. Home became the page you open into instead of chat, which finally put the case for connecting in front of everyone rather than just the people who went looking for it.
Then we added a connection panel down the right-hand side of home, so noticing it doesn't depend on reading anything first. It sits there whether or not you arrived meaning to set something up.
Deciding who we were building for is what let the panel be this blunt. If we were still trying to please everyone, it would have had to work for people who don't even have an ad account yet, and that usually means softening it until it stops asking for anything. We'd already decided those aren't our users, so it just asks.
They can still use the product. Nothing is locked. But without an account connected, Playad is a generation tool, not something that manages your ads, and the panel is honest about that rather than pretending the two are the same.
Two changes. Home became the page you open into instead of chat, which finally put the case for connecting in front of everyone rather than just the people who went looking for it.
Then we added a connection panel down the right-hand side of home, so noticing it doesn't depend on reading anything first. It sits there whether or not you arrived meaning to set something up.
Deciding who we were building for is what let the panel be this blunt. If we were still trying to please everyone, it would have had to work for people who don't even have an ad account yet, and that usually means softening it until it stops asking for anything. We'd already decided those aren't our users, so it just asks.
They can still use the product. Nothing is locked. But without an account connected, Playad is a generation tool, not something that manages your ads, and the panel is honest about that rather than pretending the two are the same.
Day-one connections went from 3.2% to 14.4%, and the new connectors were better than the old ones.
Day-one connections went from 3.2% to 14.4%, and the new connectors were better than the old ones.
You can fake this. Make a prompt pushy enough and people will click it whether they want it or not. If that's what I'd done, the people connecting after the change would connect and then vanish.
Before: 106 connected on day one, 26% came back in week one. After: 213 connected, 33% came back.
You can fake this. Make a prompt pushy enough and people will click it whether they want it or not. If that's what I'd done, the people connecting after the change would connect and then vanish.
Before: 106 connected on day one, 26% came back in week one. After: 213 connected, 33% came back.
