John Koetsier is a journalist and analyst. He's a senior contributor at Forbes and hosts our Growth Masterminds podcast as well as the TechFirst podcast. At Singular, he serves as VP, Insights.
It has been personally depressing to me that the transition to SKAN 4 has been taking so long. One little bug that totally borks conversion values and the whole ecosystem got cold feet, and stayed chilly for months. However, I’m happy to finally be able to report some positive news: as a percentage of all postbacks, SKAN 4 postbacks are definitely increasing in a months-long positive trend.
One big player that’s joined the SKAN 4 party?
Reddit.
SKAN 4 postbacks trending up
First off, here’s the trendline. Don’t pop the champagne yet, and hold off on the celebratory SKAN 4 IS HERE parties, but there’s a definite trend from August through to November of gradually increasing SKAN 4 postback share. (The big bulge in late July is Meta going SKAN 4 just before the SKAN 4 bug bit.)
So what changed?
Well hi there, Reddit
In the last 30 days and especially the last week, Reddit has ramped up its delivery of SKAN 4 postbacks to almost the 90% level. Reddit announced the update in — of course — a subreddit, saying that they “launched support for SKAN 4.0, and is excited to uplevel our product solutions and provide our partners with a slew of improvements and new features.”
A few notes the Reddit for Business team added:
SKAN reporting for Reddit will be at the ad group level AND the ad level, which Reddit says will help achieve crowd anonymity quicker
Each app ID in Reddit can have up to 200 active ads (up from 100 under SKAN 3)
You can have up to 20 ad groups active
Each ad group can have up to 10 ads active
Reddit has streamlined SKAN ID management, improving visibility into how many IDs are available for new ad groups.
The result of Reddit’s moves is that Reddit is now the leading platform/ad network for SKAN 4 adoption on Singular’s SKAN 4 adoption dashboard.
Who else is joining the SKAN 4 trend?
Other leading ad partners include:
Jammp
Unity
Smadex
Dataseat
Unicorn
Mintegral
AppLovin
Moloco
Appier
Liftoff
Google Ads
Kaden
Remerge
What we haven’t seen yet is Meta reverting back to issuing SKAN 4 postbacks en masse. Google is still in a testing/holding phase as well, as are platforms like TikTok and Snap.
In a recent webinar, most experts on the panel predicted SKAN 4 would only hit majority postback share in Q1 2024.
My guess is that the ecosystem is basically ready right now, most iOS devices have been updated with a SKAN 4 CV reset bug fix, and the sticking point is that platforms are investing in and supporting their own modeled results preferentially over SKAN 4. They’ll get there, and they’ll eventually flip the switch on SKAN 4, but only when they’re good and ready … and hopefully their customers are primarily relying on internally modeled numbers rather than SKAdNetwork.
Maybe I’m just cynical like that.
In any case: we are now seeing an uptick, and as additional small, medium, and large ad networks begin tipping over into SKAN 4, we’ll likely see that continue.
BTW … have you signed up for SKANATHON yet?
We’re doing a thing. And it’s a pretty awesome thing: SKANATHON. As SKAN 4 postbacks trend up, it’s a smart thing for you to considering joining.
SKANATHON is a live webinar series spread over 2 days and 4 sessions with 17 speakers to get you ready for SKAN 4 in 2023.
Session 1: SKAN review SKAN 3, how it works, what to do
Session 2: ASA and ASO Running Apple Search Ads and boosting app store optimization to mitigate impacts of SKAN.
Session 3: Singular’s SKAN solution There’s a reason the industry sees Singular’s SKAN solution as the leading measurement product for iOS. Get a deep-dive into how it works and why it gives you better results than the competition.
Session 4: SKAN 4 deep dive SKAN 4 is (almost) here. Get ahead of the curve and find out how to take advantage of all its goodies, while not losing backward compatibility with SKAN 3.
How do you make SKAN work in the real world? Well … you can start by watching this video. Hit the play button and keep scrolling …
Apart from SKAN on iOS, the measurement options are challenging. You might try to use modeled ad network data for campaign measurement. You might use media mix modeling to attribute advertising results. You might still be trying to use fingerprinting. You might be using first-party data and logic. In fact, you probably should be using many methods.
But on iOS, SKAdNetwork is the only option that is deterministic, ensuring that (under certain conditions) you will get postbacks for app install attribution, and ad networks will get the data they need for campaign optimization. So it makes sense to use it also, and to learn to use it well.
The problem is that to make SKAN work in the real world, there are plenty of challenges:
Choosing the right model
Selecting the right key measurement events and thresholds
Encoding enough information into the limited space in your postbacks
Designing the right app experience to get quick feedback via SKAN on results
So I asked 2 of Singular’s smartest SKAN whisperers to spend some time with me on making SKAN work in the real world. Both of them have worked with literally hundreds of clients to help them tweak their settings, models, and apps and extract the absolute maximum amount of information from SKAdNetwork.
Victor Savath, VP Solutions Consulting
Nabiha Jiwani, Customer Success Team Lead
Here’s what they told me …
1. Realize it’s not 1 and done
Under IDFA, you could collect everything and figure out what you need after. Under SKAN, you need to be much smarter in selecting the right measurements.
But you’re not smart enough. None of us is. Realize you’ll be iterating to get it right, and realize you’ll be iterating more to get it righter. And furthermore, realize that as the world changes — Apple switches something, a partner changes something — you’ll be iterating yet again.
“There needs to be a mentality and understanding that it’s not a one-and-done exercise,” says Savath. “We see iteration as part of the core philosophy of approaching SKAdNetwork. Not just because your KPIs change or your product changes … the ecosystem changes.”
2. Understand the goal
Do you mostly want measurement for internal teams? Do you specifically want optimization for ad partners? Is ROAS the key metric you want, or is something like the customer journey more important? What does the product team need, specifically, versus the growth team?
“With my clients specifically, I’ve seen a deeper focus on how partners can understand and read the SKAN data and make sure that those events connect,” says Jiwani. “The second layer would be, okay, let’s focus on revenue.”
SKAN is technology, but you make SKAN work first and foremost by having a clear strategy.
3. A big benefit of ad monetization measurement models under SKAN
Ad monetization has a massive advantage over monetization models like subscription when using SKAN, because feedback is far quicker.
“When I think about conversion models more holistically, I’m always thinking about what gives you good early signals and then what can also give you strong predictors of quality user, strong LTV over time,” Savath says. “So it’s often a blend of discrete variables, continuous variables. But in the world of admon, you can actually have both because you’re going to have early signals (ad impressions, generally speaking) so you have these signals with a high level or decent amount of variance.”
That means you can segment users easily just based off volume of ad impressions alone, and you can use those segmentation to make and test predictions about which cohorts might be more likely to buy IAPs or even subscribe later on, if you have those options.
There’s a lot of modeling under SKAN 3 thanks to the very quick measurement period. That modeling can be very good, but it’s still modeling.
So take some of the early signals you’re measuring with SKAN and compare that with your Android data.
“Taking those smaller signals like a session start or some sort of indication there and coupling it with what you know of how your Android users have been performing and comparing those 2 subsets have helped a lot of my clients understand where their iOS data has been going,” says Jiwani.
Plus of course, with SKAN 4 you’re going to get longer measurement periods, which will help considerably. That may not arrive at scale, however, until early next year. SKAN 4 postbacks are currently only around 15% of all postbacks that Singular is seeing, though the trend is rising.
5. Make SKAN work: understand your users better than they do themselves
Clearly you want to use early indicators to predict future behavior. The first step is understanding what is actually happening in your app versus what you just happen to be currently measuring.
Once you do that, you can build your SKAN models off real behavior.
“Instead of just choosing revenue buckets based on, let’s say my average product price tags in an in-app purchase, let’s just look at my IDFV data set and say: what is the average amount generated on day one amongst users that complete the tutorial?” says Savath. “Perhaps you use that as your revenue bucket thresholds when defining a conversion model, because now it gives you a good segmentation group or cohort to observe.”
Watching that group over time — much more time than you can currently measure with SKAdNetwork — gives you good insight into monetization potential for that segment. You can then experiment with early predictive signals that indicate a new SKAN install should be assigned to that segment.
6. Iterate SKAN conversion models monthly when testing
Iterating in the real world with SKAN is a challenge. You can’t do it daily, because you need some amount of time to let SKAN campaigns flush through your ad network partners’ ecosystems. Iterating too quickly will be messy, and provide insufficient data to make smart decisions on.
But how often should you iterate when in testing?
“I’ve had clients change it month over month initially,” says Jiwani. “You might optimize towards tutorial complete, registration, some sort of account creation, right? Changing those initial metrics that indicate whether someone will or will not purchase, or will or will not deposit and then taking that data and then modeling it … changing that first initial event indicator has been really strong with some of our customers.”
Of course, you’re probably not going to continue that cadence forever, but you’re also not likely to keep the same model for a year.
A quick note: initially when switching SKAN campaigns you had to pause everything, wait 48 to 72 hours, then restart.
Singular now offers technology that makes iteration much quicker and simpler: just switch and go. Singular also offers a product feature where you can simulate SKAN changes without actually making them, and see what your update measurement data would look like.
7. Check your log-level data to optimize revenue buckets
Don’t just set up conversion value revenue buckets and forget about them. If your revenue buckets don’t match what users are actually doing, you’re essentially wasting bits and failing to maximize information return.
Make SKAN work by checking log data:
“Anyone who has a mixed model or IAP revenue model, you’ll notice in any sort of log-level data … there’s dips or there’s segments in your conversion model that are just unused,” says Jiwani. “Making sure you tweak those buckets … to actually capture that subset of users, or making sure that you either expand or minimize those buckets to ensure that there are no dips, that each actual conversion value is used in that model … will ensure that you’re getting the most amount of data you can.”
Jiwani says she’s done that exercise time and time again with customers, and it almost always results in better data capture.
8. SKAN 4 helps you break cohorts down into sub-groups for higher fidelity
In SKAN 3 you might be able to assign a user to a cohort like did_tutorial, observe later behavior, and set a revenue estimate based on early indicators for that kind of cohort.
In SKAN 4, you’re going to be able to go a lot deeper.
“You have a cohort of users that did ‘tutorial complete’ on day 1 and performed an in-app purchase …that’s operating as one of those 64 buckets,” says Savath. “That’s a user cohort. The beauty of P2 and P3 is you observe that user as they enter the P2 timeframe or the P3 timeframe and you say, oh … that single cohort now breaks into 3 additional groups … users that did ‘tutorial complete’ day 1 and purchased … and then in P2, they did a repeat session. They came back. Or they made another purchase, or they made a huge purchase, right? Now you have 3 groups: that one cohort split to three.”
All of it adds to modeling rigor and your ability to build better ROAS and LTV models based on early indicators.
A quick note about SKAN 4: yes, we still don’t have a majority of SKAN 4 postbacks yet. In fact, far from it.
But with Singular, you can set up a SKAN 4 conversion model, benefit from any SKAN 4 partners you’re working with, and not lose any data from SKAN 3 partners … because it’s backwards compatible.
9. Streaming verticals and subscriptions: yes, you do have early data
Subscriptions are hard under SKAN, especially if you used to have a 7-day trial period.
But there are always early events you can use for proxies, and in streaming media verticals like music, entertainment, and video, there might be more than most. Make SKAN work by using them.
“If you think about any of your streaming services … you do have continuous variables that you could read signals into,” Savath says. “They’re good early indicators and they’re plentiful.”
Examples:
How many times did they listen?
How many movies did they watch?
Did they stream to a bigger screen, or to an external speaker?
Also, there are segmentation indicators, like what kind of subscription they tested: a family plan, an individual plan, student plan, and so on. All of that gives you more information to build segmentation and look for predictive indicators.
Such as: in the family plan, did anyone else sign up and get added?
10. Multiple monetization methods helps drive better data
If you only monetize via subscriptions, you have one thing to measure, and it’s hard, and often takes longer than SKAN’s available measurement period.
If you add ad monetization, then you’ve got additional signals that will come quickly and give you more information. And if you add in-app purchases, you’ve got even more information that will help you build smarter predictions and more accurate assessments of cohort value.
Even better, you’re automatically allowing your users/customers/players to segment themselves, and you’re allowing people who don’t want to make an immediate long-term commitment the ability to try before they buy.
11. Retail: first purchase is easy, subsequent value is hard
Retail apps often work fairly well under SKAN for initial purchases at least, because people often download a retailer’s app for a specific purpose, and they pull the trigger immediately.
So the first purchase often happens quite quickly.
The problem is getting adequate measurement for subsequent purchases.
That’s where you have to look at engagement variables and usage variables: sessions, views, searches, add to carts, and more to get a sense of how likely a specific newly acquired customer is to buy more.
A personal note of caution here: I have installed retail apps, purchased nothing, and then made hundreds of dollars of purchases literally months later. You simply cannot assume that if nothing happens right away, nothing will continue to happen forever.
SKAN 4 will help, but not be a panacea for that:
“The unlock with P2 and P3 postbacks is just having those additional signals that indicate someone’s coming back or someone’s making a different purchase, or they’re viewing another item in a specific catalog,” says Jiwani. “That will be helpful with SKAN 4.”
12. Retail: make sure you differentiate between the reporting layer and the partner optimization layer
It’s important to be able to measure the quality of profitability of new user cohorts. It’s also important to communicate the value of new users to ad partners. In retail apps, that often means a mixed SKAN conversion model with measurement for engagement events as well as revenue.
A key tool here: IDFV.
“You might have an engagement and events funnel model,” Savath says, referring to sign-ups, cart adds, etc. “If you have that type of model, Singular’s in a place where you still have revenue reporting because you could say: I’m using that funnel as a way of segmenting my IDFV data set, and I’m going to observe these cohorts and see the actual revenue that they generate over time, and report on those revenue inferences within the Singular reporting interface such that the networks can be optimizing off events based off how the model is configured, but from an analytics or LTV reporting perspective, I can also see the revenue.”
13. Fintech: mix engagement and revenue metrics
Fintech can be tough to measure under SKAN. There are great events to look for, such as account creation, connecting a bank, depositing money, but these are big steps for people to take, and they don’t always happen quickly.
Mixed conversion models, therefore, are the way to go:
“I’ve seen most customers do a mixture of both engagement metrics and revenue metrics to capture users that who initially have engaged with the app, have inputted a certain level of information into the app, have connected various accounts within that app as an indicator of how active that user is,” Jiwani says. “And then at the same time on the revenue front, we’re capturing potentially the amount of money deposited or the amount that’s been used in a transaction.”
Funnels models are more rare in fintech, but there is potential here, she adds.
14. On-demand: you lucky SOBs!
On-demand apps are super-lucky under SKAN: most people who download an on-demand app do it as part of a purchase or engagement process.
Example: you want a ride, you download Uber or Lyft, enter your payment information, and take a ride.
But there’s more to look at to make SKAN work when you want to estimate LTV, Savath says.
“Then you go into the world of continuous variables … revenue amounts are definitely not as common given the high level of variance between trip length duration. So it’s really around the engagement and frequency of utilization.”
More, generally, is good. (Of course.) But it can be misleading too: vacation or business travel users might be very sporadic.
15. Games: hyper casual vs mid-core
Hypercasual games might have been just made for SKAN. Admon and speed are both common factors here:
“It’s almost like it was designed for this use case because you’re talking about users that are engaging what they do within the first day, and then oftentimes the life cycle of hyper was much more truncated,” Savath says.
Also, if we can help in any way, experts like Victor Savath and Nabiha Jiwani work with customers and prospects every single day to ensure they maximize their marketing campaign values with Singular tools. Sometimes you even get Singular CTO Eran Friedman, perhaps the most knowledgeable privacy-and-measurement person on the planet.
Get in touch, book some time, and find out how we can help.
On April 26, 2021, half the mobile marketing world changed as Apple released iOS 14.5 and SKAdNetwork. Something similar will happen for the other half of the mobile universe, likely at some point next year in 2024, when Google flips the switch on Privacy Sandbox on Android and — perhaps simultaneously — Privacy Sandbox on the web. We recently hosted a webinar with Google, Gameloft, and Tinuiti to help the ecosystem prepare. One of the things we learned: what marketers consider to be the hardest part of Privacy Sandbox.
We invited panelists from Google, Gameloft, and Tinuiti to join Singular co-founder Eran Friedman:
Kelly Gieschen, strategic partner manager, Privacy Sandbox, at Google
Vasil Georgiev, UA director at Gameloft
Mollie Sheridan, senior manager mobile app paid search, Tinuiti
As it turns out, the hardest part of Privacy Sandbox could be the same as the most important part: understanding measurement results.
When we asked participants during the webinar what they thought would be most challenging, here’s what they said:
Understanding measurement results: 47%
Setting up conversion models: 20%
Keeping track of cohorts: 12%
Targeting: 12%
Retargeting: 8%
That and other challenges just in learning what will become the new official system for attribution, targeting, retargeting, and SDK management on Android is stressing marketers out.
68% are concerned. 17% are terrified, according to the sample of marketers attending the webinar. Only 15% are either “happy” or “fine.”
The good news is that Singular, Gameloft, and Google are already beta-testing the hardest part of Privacy Sandbox as well as every other part. Check out the webinar for an update on how that test is going.
“Privacy Sandbox believes first that user privacy and a healthy mobile ecosystem are not at odds. And, two, that a blunt approach without providing working alternatives does not work and will make users worse off. So with those two principles in mind, we envision technology where privacy takes precedence while businesses can still thrive and be successful.”
– Kelly Gieschen, strategic partner manager, Privacy Sandbox, at Google
Essentially, Gieschen says, it’s about allowing businesses to continue their growth and marketing initiatives without having to use granular user-level data, or capturing device identifiers that could be used for cross-app tracking.
Advertising APIs and context APIs
Most mobile marketers know the core APIs in Privacy Sandbox by now. It is interesting, however, to see how Google insiders approach them: as building blocks for the industry to innovate on top of.
There’s 3 advertising APIs, 2 of which are “relevance APIs:”
Topics and Protected Audiences are relevance APIs
The third advertising API is Attribution Reporting
“Topics provides high-level user signal interests and may be combined with contextual signals and first-party data so that SSPs and publishers can select relevant ads. Then we have Protected Audiences, which supports more granular remarketing use cases, enabling ad tech marketers, developers, advertisers to reach audiences who’ve shown interest in a specific brand or product in a privacy-preserving way.”
– Kelly Gieschen, strategic partner manager, Privacy Sandbox, at Google
Those matter in the MMP space, and there’s some work happening on them, says Singular cofounder Eran Freidman, but the Attribution Reporting API is naturally where marketing measurement companies are going to focus.
“Naturally, it’s the biggest focus for us as an MMP and we’re putting a lot of resources on that, from testing the framework, integrating with the different media partners, or designing the product to… provide the essential performance reporting.”
Differences between SKAN and Privacy Sandbox
“There are similar principles between the frameworks … but if we talk about the differences, there are many,” says Eran Friedman.
Some of the differences:
Privacy Sandbox aggregation keys for campaign, creative, placement and optimization data provide far more range than even SKAN 4’s campaign IDs
Wanting more data points comes with a cost in Privacy Sandbox: the more values you encode and the more granular you try to go, the more random noise gets injected into the data. In SKAN, there are fewer data points, but once you pass privacy thresholds (SKAN 3) or crowd anonymity (SKAN 4) you get essentially all your data.
In Privacy Sandbox you always get some data, even at very low scale campaigns, while under SKAN you need to pass certain minimum install number thresholds. That number is less with SKAN 4 than with SKAN 3, but it remains. The tradeoff for Privacy Sandbox for Android is that at low thresholds, more noise or junk data is inserted.
There’s another key difference between SKAN and Privacy Sandbox that Gameloft UA director Vasil Georgiev highlighted, and that’s testing.
You’ll be able to do far more testing far easier under Privacy Sandbox than under SKAN, simply because you have more ability to encode variables.
“It’s very clear that one of the first differences is that the opportunities for testing will be enormous,” Georgiev says. “We are not going to be limited to the things that we can test.”
From tracking everything to trade-offs
You don’t achieve privacy without cost. There will be loss of signal, similar to what we’ve seen on iOS with ATT and SKAdNetwork. Probably less loss of signal, but loss nevertheless.
“Today marketers can track everything they want, everything they can,” says Gieschen. “And that practice would have to fundamentally change. And depending on what marketers want to look at specifically, there may be trade-offs on granularity and richness of information versus noise and delays.”
Some of the trade-offs refer to how much detailed information you, as highlighted above. Some refer to how quickly you want information, says Tinuiti’s Mollie Sheridan.
“If you’re pulling reports more often, there’s going to be less accurate data and Google is going to inject that noise data just to protect privacy,” she says. “You’re going to have to decide if you want to wait longer periods for more accurate data or if you want shorter periods.”
That said, most things marketers want to do today will still work under Privacy Sandbox for Android. The art and the science will be balancing granularity versus aggregation as well as speed versus accuracy to achieve the best possible — not perfect — results for marketers’ attribution needs and ad networks’ optimization needs.
“You’re going to be able to optimize towards the events that you want to optimize towards, whether that be CPI, target ROAS, specific events, they’re still going to be available to you,” says Sheridan. “It’s just going to be the frequency of your reporting and the segmentation of your reporting … balancing that out to make sure that you’re getting enough volume to get the most accurate data available within this privacy-centric framework.”
The good news: actual hard-core high-scale marketers think this is going to work.
“It’s very clear that Google recognizes the minimum viable state of data and they’re trying not to block marketers from continuing to do optimizations,” Georgiev says. “And I also believe that they are trying to avoid making it more complex than it should be.”
The timetable for Privacy Sandbox full roll-out
Short answer: there isn’t one yet.
But Google promises plenty of notice before the full Privacy Sandbox roll-out does actually happen.
“We don’t have any updates that we can share publicly at the moment in terms of when 100% migration would happen,” says Gieschen. “But just as we’ve done in the past, we’ll be giving the ecosystem and partners ample notice prior to any changes regarding the beta and general availability.”
Much, much more in the full webinar, including progress on our Privacy Sandbox beta with Google and Gameloft
Meta is introducing a privacy-safe way to get back user-level click-through and user-level view-through attribution on Android for ads on Facebook or Instagram, the Meta Install Referrer.
This significantly expands on the measurement possibilities already available via Google Play Install Referrer and is a big deal. For the first time in years, Android app advertisers will get view-through attribution back on Meta, providing additional details and data to build up a more accurate picture of the results of their campaigns.
Meta Install Referrer (MIR) is supported by Singular as of November 2023.
Here’s a quick comparison of the Google referrer versus the new Meta Install Referrer:
Google Play Install Referrer
Meta Install Referrer
Purpose
Attribute Android installs via ads on Meta
Attribute Android installs via ads on Meta
Use cases
Click-through
Click-through
View-through (most scenarios)
Different session click-through
App stores
Google Play
Google Play
Other Android app stores
Singular supports both the original Google install referrer and MIR, but because MIR includes all Google referrer use cases and adds more, Singular will prioritize using the Meta Install Referrer for user-level attribution decisions.
It’s important to note that there are some caveats. View-through attribution is available via MIR when using:
Advantage+ App Campaigns
Broad Targeting Manual App Promotion Campaigns
The supported campaign configuration for Manual App Promotion Campaigns include:
Default age setting (18-65+)
All genders
Country or country group
Interest segments, behaviors, and demographics set to broad targeting
Custom audiences also set to broad
How the Meta Install Referrer works
Most people understand how the Google Play Install Referrer works, because it’s based on the concept of a click referrer on the web:
User clicks on app ad on a Meta property
Meta encrypts campaign metadata
Meta appends it to the referrer parameter in the Play Store URL
The Play Store URL brings the user to the app listing
The Play Store saves the referrer string
Singular’s SDK reads the referrer from the Play Install Referrer API
Singular decrypts the data for install attribution
The Meta Install Referrer operates differently, but essentially achieves a similar purpose. At a very high level it works something like this:
User views or clicks on an app ad on a Meta property (and then installs the app)
Meta encrypts campaign metadata
Meta saves the campaign metadata to local on-device storage in either the Facebook or Instagram app, wherever the ad was shown
Singular’s SDK (for the installed app) reads the campaign metadata from local storage
Singular decrypts the campaign metadata for install attribution
Privacy and marketing attribution
The Meta Install Referrer is on-device attribution.
When we looked at the design of MIR we were excited to see it does not need to rely on leveraging a Google Ad ID or other device identifier, nor on the need for transferring any device identifiers to a Meta or MMP server to make it work. As such, we believe it’s a much more privacy-friendly solution than existing GAID or IDFA-based solutions. Another way we think of it, the solution could be said to function pretty much like UTM parameters on a standard link on the web, even in view-through scenarios.
As such, it’s possible other major platforms could adopt similar mechanisms.
The result of this new methodology is that marketers will get a clearer picture of what ads, creative, and campaigns on Meta resulted in conversions, and they’ll be able to gauge the value of their investments better.
Enabling MIR on your campaigns
Talk to your customer success contact about enabling MIR on your campaigns. If you’re not currently a Singular customer, here’s a good place to start.
It seems like just yesterday that one of the core challenges of high-volume growth teams was scaling ad creative. Making the thousands of pieces of art needed for testing was hard, even with some rudimentary automation. But yesterday, Google announced that P-Max is adding generative AI for ads in beta, and it has some super-cool capabilities.
Over the next few months the floodgates will be opening for adtech platforms to make generative AI for ads a common core feature for all their advertisers. And while right now many of the platforms are sandboxing generative AI created images in creative suites that help marketers make the images they want, soon that will transition to on-the-fly creation of images and content personalized to individual people.
In that, the big platforms will have huge advantages.
Google’s Performance Max and generative AI for ads
P-Max’s generative AI is both impressive and limited.
It’ll take your brand ambassador and drop her into a corporate, home, beach, or country setting. It’ll generate headlines and descriptions, and let you create core assets to use in multiple ad types and backgrounds. If you already have product assets, P-Max will import them and allow you to try as many variations as you like. Google says it will never generate the exact same image twice, so your competitor won’t have an ad that looks shockingly similar to yours.
(Cue the let-me-try-prove-that-wrong crowd.)
It’s U.S.-only at the moment, and it is rolling out gradually, so not every advertiser will get access this week or even this month. If you’re in a sensitive vertical, such as politics or pharmaceutical, you won’t get access either.
You won’t be able to create images of specific people or celebrities, for obvious reasons, or branded items. (My recent attempt to get OpenAI with Dall-E to make a picture of Oprah failed recently, but Creative Diffusion allowed it.)
And all images will be watermarked with SynthID so there’s a track record and accountability to surface the fact that they are artificially created.
What P-Max’s generative AI for ads solution isn’t is a real-time per-person per-product generative AI solution that combines what it knows about Google users and what it knows about advertisers’ products and offers, and crafts a completely personalized one-off ad in real-time or near real-time to maximize relevance.
Join the party: everyone’s doing it
Everyone is joining the generative AI for ads party. Some are further ahead than others in their ability to integrate generative AI with ad tools.
Googlelaunched Performance Max generative AI ads; availability starts now
Metalaunched generative AI in Ads Manager: started last month, rolling out globally “by next year” for background generation, image expansion, and text variations
Amazonlaunched image generation in beta last month, primarily focused on lifestyle backgrounds for product images
Microsoft is adding Copilot to the Microsoft Advertising Platform which will generate new images using Microsoft’s image library, as well as offer conversational chat … debuting in closed beta in “early 2024”
TikTok just launched a generative AI “creative assistant” to “to spark creativity and be a launchpad for curiosity” as you make ads for the platform
Snap is running text ads in My AI, but not generative AI in ads yet
Pinterest hasn’t announced anything yet
Reddit hasn’t announced anything yet
Big brands are using Dall-E and other tools to make their own generative AIs
Agencies like WPP, Publicis, Omnicom are doing the same
Ultimately, this will be a standard feature and a checkbox item on all major platforms, and from most significant ad agencies and ad networks.
The big platform advantage for on-the-fly generative AI, and what’s next
While all advertising platforms will likely add generative AI for ads tools, add-ons, or plugins eventually, the biggest platforms have huge advantages. Not only can they throw more engineers and more money in solving the challenges of bringing generative AI to their advertising tools quicker, they have an on-platform data advantage.
That means that when it comes to on-the-fly generative AI creation, they can do it with much greater knowledge of their users’ interests, habits, and behaviors, giving them a much greater chance of crafting a message that resonates.
There’s still a lot of work to be done, of course, as the platforms themselves acknowledge. One big area of improvement: allowing marketers to define brand colors, imagery, styles, even conversational tones so that the ads they generate fit the brand and build the brand.
“There is still work to do on delivering outputs customized to every brands’ unique voice and visual style,” Meta says. “We’ll need to define new ways of partnering with brands and agencies to help train these models on brands’ unique perspective.”
The other is doing this all in a way that is cost-effective on GPU time. Especially for on-the-fly generative AI, the GPU load is going to be intense.
Snap is already thinking about that, and has plans to do the work on-device. I’m not sure that will work in all cases, but with the kinds of chips that Apple is putting in its devices these days — and the new Snapdragon Gen 3 chip on Android — that’s like to be a possibility at some point.
Mobile web for user acquisition is growing, iOS and Android spend patterns are normalizing, and CPIs are dropping. That’s just a fraction of the data-driven insight in the new 2023 State of UA Report, which is based on a $10 billion subset of Singular cost data, trillions of ad impressions, hundreds of billions of clicks, and billions of app installs.
It’s also based on insight into billions of creative optimization decisions from our partner on this report, Apptopia.
So what is the state of UA in late 2023?
It’s a weird, weird time, and not just for user acquisition pros.
Spend is down: the global economy is feeling the aftershocks of Covid, military conflict, and increasing political polarization. There’s also still post-ATT and SKAdNetwork disruptions, and visions of more of the same in the coming Privacy Sandbox changes.
What’s inside the report?
So in this state of UA report we focus on:
Massive ecosystem shifts
Verticals most impacted
iOS/Android spend changes
CPI trends
Spend distribution in key global geos
Ad format shifts, including
Video
Banners
Interstitials
Growth insights based on brands like
Calm
Sephora
Temu
The state of UA is that it’s late in the year
We’re nearing the end of 2023, and the all-important holiday season is just around the corner. The data in this report will help you navigate end-of-year as well as the first quarter of 2024, sharing how user acquisition specialists are adopting mobile web, testing new ads. We also dive into spend distribution between Android and iOS globally as well as in key countries in Africa, Asia, Europe, North America, and Africa.
Thanks to our partner Apptopia we also have dozens of insights into ad format changes, what’s working, what’s changing, and what’s hot.
In addition, we dive into how Calm boosted downloads 83% month over month, and more than doubled them quarter over quarter. We examine how Temu tested new partners and boosted Android installs more than 3,000%. And we look at how Sephora transformed from underdog to top dog in a mobile app install battle with a key upstart competitor. All of this is gold not just for user acquisition managers but also for growth strategists looking to deploy capital, creative, and campaigns at scale to win.
The first-ever banner ad is a long way from generative AI in advertising. But one of the same people who worked on placing that tiny paid piece of internet history is still engaged in the adtech industry. Now, of course, he’s focused on generative AI and other emerging technologies.
The first banner ad
It was 29 years ago in 1994 when AT&T paid actual money for the very first banner ad on HotWired.com, now just Wired. Part of a crystal-ball ad campaign that foretold people working remotely from the beach on laptops, or having video conferencing meetings, the text-heavy banner ad said simply in a rainbow-colored font: “Have you ever clicked your mouse right here? You will.”
It wasn’t even called a banner ad, initially.
The first name was “tile.”
“The original concept was drawn on a whiteboard and we were drawing outlines of the websites that were out there … and we called them tiles,” says Tom Zawacki, now president of enterprise solutions at Data Axle, but formerly employed at Modem Media. “Originally, we said, if we could just take a tile and put the tile on one website and have them click there and go to another website, that would be really cool. And that was it. So that was the original original concept.”
A box, with text, linked to another website.
A fairly humble beginning for internet advertising, you might think. Pretty far from today’s ideas of generative AI in advertising.
Generative AI in advertising personalization
Now of course generative AI is the hottest tech, not the humble banner ad, and the opportunity for creativity and variety is increasing exponentially. Ad creative personalization is one big opportunity Zawacki sees in generative AI.
“One of the nice things about using generative AI is it allows us to increase the volume of production when creating copy and or visual design,” he says. “Forever we’ve been promising the delivery of personalization … what’s gotten our way is the volume and velocity of variables that create these combinations of creative message and visual design that humans just can’t create fast enough.”
I see that too, and my mind is blown by the opportunities that Amazon, for instance, has in generative AI ads. Most Amazon ads are shown on the Amazon platform, of course, which means that Amazon knows a lot about the people seeing them: purchase history, search history, maybe some of what you watch on Amazon Prime Video, maybe some of what you listen to in Prime Music, maybe what you read on Kindle.
Imagine the personalization Amazon can create — especially on-platform — in generative AI ads for the 10s of millions of products it sells. Text and image ads should be relatively easy. Video ads are also fairly doable, with a heavier compute lift.
Personalized playables are probably coming as well.
Off-platform, of course, that’s harder. With ATT, SKAN and Privacy Sandbox taking away access to device identifiers, off-platform personalization will be harder and less targeted.
But other platforms, walled gardens, and retail media platforms ought to be able to do similar things in their own worlds: Facebook, TikTok, Snap, Pinterest, DoorDash, Uber, and others. The golden age of marketing personalization is likely going to be found in on-platform generative AI in advertising … with the possible addition, for brands, in owned spaces like apps and websites, and permissioned communications.
Generative AI to build creative campaigns
Another question: how will marketers use generative AI to build their core creative?
It’s easy, as I stated in the conversation, to want the machine to do the work. To sort of poke MidJourney with a stick and ask it to do something cool. It’s harder to make excellent, world-class, brand-compliant creative that completely fits your needs for a specific campaign or ad opportunity.
For Zawacki, the combination of biological and artificial intelligence is always going to win.
“IBM Watson did some great research in 2017, taking human intelligence, taking an artificial intelligence … and having them do a series of events,” he says. “And in every case, the combination of humans and AI working together — augmented intelligence … they called it cognitive computing at the time — won out in all those series of tasks.”
His goal: the Tony Stark model, where you have a human intelligence directing an AI that multiplies and accelerates innovation. We’ll all probably have our own JARVIS — Just A Rather Very Intelligent System — at some point in the near future.
And that will be a game-changer.
“We are using augmented intelligence to turn our clients and our employees into superheroes,” Zawacki says.
Meta will charge the equivalent of $127.40 for an annual subscription to its services in the EU starting in November 2023. That’s essentially double what Meta makes in ad revenue per user in Europe, which is a total of not quite $64 over the past 4 quarters.
Meta announced today that due to changes in EU regulations, it will be offering an ad-free subscription to Facebook and Instagram. The core reason: EU legislators have rejected Meta’s use of “contractual necessity” as a legal basis in Europe for processing user data for personalized advertising and have pushed Meta towards a subscription option which would then offer EU citizens a data-processing-free means of accessing Meta’s global-scale social platforms.
The subscription will be optional, Meta said today in an announcement post. The option is simple:
Either pay for the service
Or, consent to targeted, relevant ads
How many subscribers would it take to replace Meta revenue in the EU?
Thanks to Meta’s detailed quarterly and annual reports, it’s easy to understand both how much Meta makes from advertising right now, and how many Europeans would have to subscribe to Meta’s services to replace that ad revenue.
Meta has averaged 408 million monthly average users over the past 4 quarters
Each user returned an average of $15.99 per quarter
Meta made $63.97 per MAU from advertising
Note that total revenue per user is slightly higher than ad revenue per user, simply due to the fact that Meta offers some products for purchase.
But Meta’s subscription plan would theoretically bring in far more revenue, per average user, than ads.
Subscribers on the web will be charged $10.60/month for an annual total of almost $130 (so far there is no mention of an annual discount, though that could come). In-app subscribers will pay more, but I’m using the web numbers as Meta is following Twitter’s lead in charging more for in-app purchases to cover Apple’s and Google’s cuts.
That $130 is almost twice what Meta makes from targeted advertising per average user.
At these rates, Meta would need 208,572,327 European users to buy a subscription to replace all ad revenue.
Of course, as we’ve seen from Twitter (OK, X), very few people will subscribe. On X, about 640,000 people pay for premium, formerly Twitter Blue. If we take Twitter Ads Manager’s estimate of 372.9 million addressable users, that’s far less than 1% of users. To be precise, it’s under .2% of users. And there’s very little to indicate that Meta’s users would be substantially different enough to impact the economics on Facebook and Instagram.
This is not about a shift in Meta’s business model
208 million Europeans are not going to start paying the Euro equivalent of $130/year to access Facebook, when you can get it simply by consenting to ads.
Rather, this is simply about dotting I’s and crossing T’s so that Meta can point to the subscription model and tell European regulators that citizens can totally and completely opt out of data processing for personalized advertising if they choose to pay for the service.
That’s good news for advertisers, who don’t want to lose a valuable way of connecting with consumers, players, customers, and users.
It’s also good news for Europeans, who will continue to have a valuable service that connects them with friends, loved ones, communities, and celebrities, and who can continue to do it for free now that Meta has (almost certainly) cleared a legally plausible rationale for continuing to process user data for personalized advertising.
Google is starting to quietly signal an upcoming Chrome feature called IP Protection that will act much like Apple’s Private Relay feature, which hides IP addresses to make tracking — and marketing measurement — more challenging. Add IP Protection to Google’s soon-to-come Privacy Sandbox technology, and you’ve got interesting parallels between Apple and Google privacy technology, plus some parallels — and gaps — between the two tech giants’ technologies for marketing and attribution.
Comparing Google’s and Apple’s privacy, marketing, attribution tech
From Google, this set of software and standards includes:
Privacy Sandbox on Web and Privacy Sandbox on Android
Clearly, we’re seeing the emergence of separate but often related suites of software, standards, frameworks, and requirements from the tech giants. These tech giant initiatives are in 2 distinct but very related areas:
Privacy enhancement
Marketing measurement
The key reason for the connection: marketing measurement has typically required tracking, and that tracking has significantly impacted privacy. These tech giant initiatives are intended to rip out granular tracking as a vector for measurement and replace it with something very different: cloaked deterministic evaluation of advertising impact, with some noise sifted into the data, to provide analytics and preserve privacy.
Here’s what I’m seeing so far. (Let me know if I’m missing anything!)
* See the Privacy Sandbox website: “Privacy Sandbox also helps to limit other forms of tracking, like fingerprinting, by restricting the amount of information sites can access so that your information stays private, safe, and secure.”
These are complex beasts on both sides, with some parts baked in as OS-level components in iOS and Android, some grafted into the app submission and review process, and some that act more as platform-level directives than actual hard-coded realities. They are not monolithic projects or programs that are neatly subdivided, which makes them harder to fully grasp, and to fully understand the overall impacts on privacy as well as marketing measurement.
And, of course, they both deal with the world of mobile apps and the world of the open web, further complicating the overall landscape.
Intelligent Tracking Prevention vs the new Google IP Protection
Apple Intelligent Tracking Prevention, first introduced in iOS 11 and MacOS High Sierra in 2017, fights cross-site tracking by blocking third-party cookies, quickly deleting many first-party cookies, and blurring device characteristics to make fingerprinting harder. In conjunction with Private Relay hiding your IP address and App Tracking Transparency for requiring permission for the IDFA on mobile, it’s a powerful tool for privacy, plus a challenge for marketing measurement.
Now there’s a similar technology coming from Google for the Chrome browser, increasing an interesting degree of similarity — and divergence — between the Apple and Google stacks for privacy, marketing, and measurement.
The new technology from Google has been signaled in a Google Groups post by a member of the Chromium team. Chromium is an open source browser engine that forms the foundation of Chrome itself, as well as any other Chrome-based browsers, like Microsoft Edge, the Brave browser, and Opera.
“IP Protection is a feature that sends third-party traffic for a set of domains through proxies for the purpose of protecting the user by masking their IP address from those domains,” writes Brianna Goldstein, a senior software engineer at Google.
It’ll be an opt-in feature that will roll out in phases, she says, and will be “just focused on the scripts and domains that are considered to be tracking users.”
Functionally, this will work very similar to Apple’s Intelligent Tracking Protection, Goldstein says. The experiment does not currently impact Android WebView, the technology that allows an Android app to display web content, and will be limited in the beginning to Google’s own domains. It could cause some security concerns, Bleeping Computer notes, because proxied traffic “may make it difficult for security and fraud protection services to block DDoS attacks or detect invalid traffic.”
Private Click Measurement and SKAN vs Privacy Sandbox everywhere
Despite the fact that Apple absolutely needs privacy to be its crucial calling card as it expands its mobile universe to an ever-more personal PC that you wear on your face with no fewer than 12 cameras on it and in it looking both at your world and your face — plus 6 microphones — the company understands that that advertising drives free apps and the free web.
And that requires measurement, because advertisers need validation that they are getting ROI.
Google, of course, as an ad network primarily — at least in terms of revenue — never needed to learn that lesson.
Private Click Measurement measures both web-to-web and mobile app-to-web ad clicks, providing an 8-bit identifier on the source for up to 256 simultaneous ad campaigns per website or app, and a 4-bit identifier on the conversion, enabling measurement of 16 different conversion events. There’s a built-in time delay of between 24 to 48 hours, similar to SKAdNetwork, and measurement postbacks for both advertiser and ad network are handled in-browser and on-device.
Along with SKAdNetwork for mobile apps — which I won’t talk about here since we’ve covered it prettyexhaustively on the Singular blog — Apple is iterating through an increasingly richer advertising measurement framework. Yes, PCM pales in comparison to cookies (first or third-party) and SKAN pales in comparison to unfettered IDFA access, but that’s the point: they’re privacy-safe, and Apple will continue to add features over time.
On the other side of the fence, Privacy Sandbox on Web and Privacy Sandbox on Android are full-fledged initiatives to redefine the basics of how advertising works. Apple’s initiatives are more about mitigating adtech’s problematic capabilities; Google’s are about reinventing the world within which adtech exists.
Again, I won’t go into huge detail on Privacy Sandbox in this post: we’ve done it extensively already (focused, of course, on Android and not so much web, because mobile is where Singular primarily lives):
The one big obvious difference between the two suites in the marketing measurement area is that Google has provided capability for needed functionality in advertising and marketing: targeting and retargeting. It’s privacy-safe, which means it’s limited and restricted, but it’s there. Apple, on the other hand, while it will offer the ability for a retargeting signal in SKAN 5 so you know you’ve marketed to an existing user, player, or customer, does not offer any capability for targeting at scale in a privacy-safe way, or retargeting existing or former users.
That, perhaps, will wait until SKAN 6 or SKAN 7?
Google & Apple’s privacy/marketing/measurement suites: parallels and divergences
Ultimately when you boil down the privacy requirements of our evolving digital marketing ecosystem, you need a combination of items to limit tracking via cookies, identifiers, or device characteristics.
As far as all of that goes, there are clear parallels between the Google and Apple technology platforms. Despite the fact that Apple banned third-party cookies much earlier (2020) and that ITP has been in-market for years, as has Private Relay, Google’s coming Privacy Sandbox along with IP Protection will achieve roughly the same results. (Note: likely those technologies will be applied with varying degrees of vigor: Google after all makes almost all of its revenue from advertising, whereas Apple makes almost all of its money from devices, but the broad strokes are similar.)
But there are clear divergences as well, like for the same reason just mentioned.
Google’s Privacy Sandbox is in essence a reinvention of the entire advertising model, as we’ve already said. It’s something that has been called a 360-degree advertising suite by InMobi’s Sergio Serra:
“Privacy Sandbox for Android is a complete advertising suite … it goes 360 degrees from targeting, retargeting, fingerprinting crackdown, and attribution.”
That’s clearly beyond the scope of Apple’s ATT and SKAdNetwork, which focus entirely on privacy and privacy-compliant marketing measurement, disregarding targeting or retargeting.
The emerging privacy-safe marketing infrastructure means we need hybrid measurement
Put it all together, and you have the emerging privacy-safe advertising infrastructure.
It’s defined by:
Increasing respect for the individual and therefore, respect for the individual’s privacy
Decreasing data-gathering capabilities for the adtech ecosystem
Decreasing ability to track people from site to site and app to app
Increasing marketing measurement complexity
Growing reliance on semi-independent attribution frameworks and technologies (Privacy Sandbox, SKAN)
All of this is happening while we’re seeing increasing complexity in marketing mix, moving from just web or just mobile to web AND mobile AND CTV and outdoor AND custom SMS AND retail media AND influencer AND desktop AND console AND more and more channels and platforms … all of which is pumping the tires of the growing need for media mix modeling (MMM).
It’s also increasingly requiring what Singular calls hybrid measurement: marketing and advertising attribution based on a multiplicity of platform, cost, campaign, delivery, attribution, and first-party signals. Some of those are derived from deterministic sources such as SKAN or Privacy Sandbox, even if they are aggregated and noise has been added. Some are based on probabilistic technologies, like MMM itself. And others are based on deterministic signals that are the most accurate and detailed and precise of any that a marketer could hope for: your own first-party data.
All of this is a tremendous shift that is literally pulling the rug out from under the feet of marketers. But it’s both an industry and global legislative shift that won’t stop.
The one thing Singular can guarantee in all the change is that we will be providing everything you need for marketing measurement, optimization, and growth.
Singular CTO Eran Friedman spent some time with Redbox CTO Samual Chorlton on the AdBites podcast. The topic: everything SKAdNetwork, especially SKAN 4 help for those working on a transition from SKAN 3.
Hit play to watch it now, and keep scrolling for some of the highlights …
Data return to advertisers: from SKAN 3 to SKAN 4 help
In SKAN 3, as we know, Apple provided privacy thresholds to anonymize users. Low volumes of conversions from a campaign results in few or no conversion values. That works, but it punishes smaller advertisers, reducing the feedback they receive from their ad campaigns and lowering their trust in ROI and ROAS numbers.
“The idea is for anyone to be able to use SKAN. If you’re just beginning, you have barely a budget, you’re just testing things, you’re going to get some limited information, but not too much,” Friedman says. “But as you scale and you need to become more advanced, you’ll get more and more granular information for optimizations.”
Under SKAN 4, just 15 installs per campaign will start to give advertisers at least some data: at least a coarse conversion value: low, medium, or high. It’s not much, but at least it’s some signal to start calibrating and optimizing.
If that provides confidence to boost your ad spend, you’ll get more conversion values and will not just get coarse but fine values: 64 potential values. Increase scale even more, and you’ll get source identifiers, providing more detailed data you can use to tag campaigns, geos, or ad placements. And that tagging informs campaign optimization and improvement: getting more of what you want.
SKAN 4 help: defining terms in SKAN
One of the more challenging parts of SKAN in general is learning the language. That’s especially true for people who are new to mobile marketing, but it’s also the case for veterans of the industry, because many of the terms are new, or used in different ways.
So Friedman defined the terms for the AdBites audience:
Conversion values
A number that you choose that represents the value of a user. When SKAdNetwork encodes that number into a postback, and your MMP decodes it for you, you get clues about the effectiveness of an ad campaign.
Coarse conversion values
Low-volume campaigns like those we just talked about can only have coarse conversion values: 3 potential values like low, medium, or high to represent user value, and therefore campaign effectiveness.
Fine conversion values
When campaign volume is high, SKAN 4 permits more data to be encoded into conversion values: not just the 3 possible values of coarse conversion values, but the same 64 possible values that were available in SKAN 3.
(Note: in SKAN 4, you can only get a fine conversion value for the first postback. The second 2 postbacks are always going to be coarse conversion values.)
Source identifier
In SKAN 4, the source identifier is additional data you can get from your campaigns. Like conversion values, it is connected to crowd anonymity: high volume supplies more potential data than low volume.
If you achieve high crowd anonymity, your source identifiers will be 4-digit numbers that you can encode with data about your campaigns, geos targeted, ad sets used, ad placements, and more.
What an MMP does for you under SKAN
When SKAN first came out, some thought it meant there would be no need for MMPs anymore. After all, SKAdNetwork can send postbacks right back to advertisers themselves, potentially short-circuiting the need for independent results measurement.
Complexity turned out to be one of the core challenges. Plus the ability to be able to interpret advertiser models for ad networks so they could optimize based on known good results.
That’s one of the core reasons SKAN 4 help is so desperately needed.
“This is where we believe it’s the perfect kind of world for MMPs to provide the technology and management of all the SKAdNetwork framework: basically using the APIs, managing those conversion values, getting back those postbacks, and essentially trying to abstract all those technical terms and details so the advertiser doesn’t even need to think in terms of those encoded numbers and all the details, and they just get kind of the bottom line,” Friedman says.
That means campaigns, installs, dollars, registrations: human terms.
Plus, given the privacy-centric obfuscation of SKAdNetwork, including randomness Apple adds to the numbers, being able to use Singular’s AI-driven modeling in SKAN Advanced Analytics restores missing data in marketing measurement while not impacting user privacy.
SKAN 4 adoption: yet to scale
One other topic the two hit on the podcast: SKAN 4 adoption, which is lagging right now for many ad networks and especially the big platforms.
“I think all of them for sure are working to upgrade to SKAN 4,” says Friedman. “Some of them have, for example, started beta testing SKAN 4 and have selected advertisers that are already working with and running SKAN 4 campaigns. Others have done full launches and we already see most of their traffic has arrived to SKAN 4 … it’s on a network by network level.”
The timelines I’ve heard most industry experts mention are in the Q1 2024 range. More on that, likely, in a future Singular blog post, but the key point is that if you’re needing SKAN 4 help, you still have some time.
Looking for guidance on your SKAN 4 transition?
Watch the video above, but also go check out our SKAN 4 transition guide here. It will give you all the details you need to get started.
Once you’ve kicked that off, book a session with a Singular expert to go through your planned implementation, and how Singular can help make it all much, much easier.
