GPT-4 for building SKAN 4 conversion schema

Can AI help you with your SKAN 4 conversion schema? Absolutely, thanks to the custom GPT that Lucas Moscon trained for building out SKAN 4 conversion models. Check it out in our recent Growth Masterminds episode. Hit play and keep reading.

We know SKAN 4 will be the default attribution methodology on iOS at some point in the future. 

That’s likely to be fairly soon: this quarter or the next. SKAN 4 postbacks jumped as a percentage of all SKAdNetwork postbacks in November as Reddit made the switch from SKAN 3, and the trend is continuing in the new year. The Trade Desk, Unity, Persona.ly, Smadex, AppLovin, and Twitter are others who are driving SKAN 4 adoption.

Singular’s SKAN 4 adoption dashboard currently shows a definite if slow upward trend over the past few months:

The challenge for busy mobile marketers is how to get started. 

Yeah, this isn’t always easy

For many, building out your SKAN 4 conversion schema from scratch is super-challenging. 

Part of the problem is that SKAN 4 is new and complex, and part of the problem is that many just haven’t been able to dedicate the significant amount of time needed to dig into SKAdNetwork and understand it in detail yet.

“We are more than 2 and a half years since SKAN was introduced into the industry and maybe more than 90% of people working here don’t get the fundamentals yet,” Moscon says.

Lucas Moscon is a senior user acquisition manager for Monkey Taps. He led performance at Goin, worked at Winclap as a performance specialist, and was a professor of paid media at Coderhouse. He’s also recently been building tools for mobile marketers, including an ad monetization hub that offers templates and scripts for remote controlling your ad monetization setup.

(Moscon credits that with 2 to 3Xing his ad revenue in specific apps. Check out the whole episode above to get all the details.)

Prepping for SKAN 4 can be hard when you’re also managing partners, building campaigns, commissioning creative, crunching numbers, and worrying about where your next pocket of growth will come from. It is definitely not easy managing everything.

But AI can help.

A custom GPT for building your SKAN 4 conversion schema

Everyone knows about GPT-4. 

Not everyone knows that if you’re an OpenAI subscriber, you can customize GPT-4 for your own needs and even release that version to the public. These specially-trained versions are called GPTs, and Moscon trained one for building out SKAN 4 conversion schema.

Check it out here.

The SKAN 4.0 Builder is super simple:

1. Tell it to build your SKAN 4 conversion schema
2. Give it some key information:
   1. App category
   2. Free trial (yes or no)
   3. Daily ad spend
   4. Business objective (revenue, sign-ups, etc)
3. And that’s it …

Pretty much instantly, the SKAN 4.0 Builder comes back with a starting point for your SKAN 4 conversion schema. Here’s what it gave me for a health app with a free 7-day trial that I’m “spending” $4,500/day on with a primary goal of driving sign-ups.

Is it perfect? Absolutely not.

I might have some different events going on in my app because it’s structured a bit differently. And I might want to spend my fine postback bits on more events because I’m pretty sure that I’ll achieve 64 possible values most times in postback 1 due to my reasonably high volume of ad spend and dedication to not spreading it amongst too many separate campaigns.

But …

It’s a good starting point. And that’s what most people need: a starting point to kick off their SKAN 4 journey and help them build out a SKAN 4 conversion schema that makes sense.

“I realized that people go to different webinars or podcasts … but they don’t visualize the schema in some way,” Moscon says. “They need it in their face to understand the starting point from where they can build different models.”

It also shows how to align the initial fine postback and subsequent coarse postbacks so that your measurement stays focused on your core goal and you get valuable information from which you can extrapolate cohort value for evaluating ad partners and individual campaigns.

Next levels would involve working with your ad partners to ensure you hit tier 3 levels of data as often as possible for all your SKAN 4 campaigns:

• 4 digits of Source Identifier
• Source app info
• Maximum data in your fine conversion value for postback 1
• And data for postbacks 2 and 3

No more excuses to not get started

Since you can now outsource at least the starting point, there’s no more excuse to not get started. Check out the SKAN 4 Builder, see what it returns for you, and ping Singular if you have questions or concerns.

The very good news right now is that Singular’s built SKAN 4 support so that it’s completely backward compatible with SKAN 3, because you’re going to have both versions of SKAN for at least 2024, and possibly into 2025. That’s partially due to partner adoption, but the SKAN version also depends on the user’s iPhone model and software update level. 

You want to maximize signal where possible with SKAN 4, while not missing out on SKAN 3 when that’s all that is available. And now with a GPT to help you build out your SKAN 4 conversion schema, it’s easy to start.

Much more in the full podcast

We also chat about ad monetization for subscription apps and how to double or triple ad revenue with smart ad management. We discuss what admon teaches you about your users, and why hybrid monetization is so critical in subscription apps.

And, of course, much more.

Subscribe in all the usual places including YouTube: find them all at our Growth Masterminds home page.