It feels like I wrote the predictions for 2024 just a minute ago. But here I am again, shocked about how quickly time goes by, as well as the fact that the only constant we can rely on is change.
If this is the first time you ever read a prediction post I make – know that I usually get most of these right. I published a 2024 retrospective recently, and I’m happy to say that I got 7 out of 8 right! Not too shabby!

With that in mind, and given that I do believe I’m some sort of an adtech messiah (in my own self-centered world) – I decided to title this years’ article as Prophecies.

Predictions vs. Prophecies

A prediction is a statement about an uncertain event. 

A prophecy is a prediction by someone who believes they have seen the future. 
While I do not own a time machine (yet…), I do wholeheartedly am willing to phrase my predictions as prophecies. 
My predictions are based on what I see.

I “predicted” that Attribution Fraud would become prevalent already in 2015.
In 2019, I saw that Apple will eliminate access to user level data. I did not predict it, but I actually saw it, as Apple specifically said they will do this already in 2019. 
I also “predicted” consolidation in the attribution space, and now we have one player controlling +75% of the market.
Coming up with these predictions, prophecies, or whatever we call them, relies on my ability to understand the market I’ve been a part of for almost 25 years.
It’s through networking with other industry members (new or veterans), and constructing a mental 4 dimensional puzzle to forecast the future. Coincidentally, this is very similar to how INCRMNTAL works.

Now for the fun stuff – My take on 2025.

Modeled Conversions: Moloco, Liftoff, IronSource, Smadex and others

Last year I wrote a two part article about the incredible success shown by Google, Meta, and AppLovin. These three companies who were all initially affected by the privacy changes, managed to flip things around and become dominant players in the advertising landscape. 
In the articles, I claimed that Modeling Conversions allowed these companies to generate superior results for Advertisers, all while not relying at all on attribution postbacks.

Attribution postbacks were previously considered as the “currency of adtech”. Attribution mechanics gave 100% credit to one source (the last touch), leading the ad platforms to often compete over who wins credit, rather than how to generate value to advertisers.
Attribution postbacks fueled an industry of wasted ad spend, and ad platforms cannibalizing organic conversions, or other ad platforms.

Now that it’s clear that modeling conversions lead to better results, and more budgets allocated to Meta, AppLovin, Google, and TikTok – my “prediction” is that the next top ad platforms will likely follow suit and deploy their own modeled conversion products, able to drive superior results to advertisers as well. 
If I had to take a bet on which ad platforms would those be, my bets would be: Moloco, Liftoff, Smadex, IronSource.

(More) Platforms Modeling conversions will trigger two related changes. See the next two predictions.

Optimization based on Incremental Value. Not Postbacks

Ad platforms WANT to generate great results for their Advertisers. At least the good ones do. 
And the definition of what good results are has dramatically changed over the past year.

Good results used to be defined by what an MMP reported:

• This click lead to this purchase
• This click did not lead to a purchase

But last click attribution did not report on incrementality, leading many advertisers to waste millions (if not Billions) on redundant ad spend over results that were already there.  

Modeled conversions bring a ton of value, but they still lack the capability of measuring true incrementality. If each platform models their own conversions, Advertisers will still need a way to report the incremental contribution of each platform, to make decisions such as budget allocation, or budget optimization. 
Such as task, previously easy to do using an MMP, will no longer be possible to do 
without the tools “speaking to one another”.

Meta has already said that it will be starting to consider incrementality signals in its’ optimizations. We predict that Google, and TikTok, may likely follow or announce a similar feature in the next months.

The MMPs will Disappear. Long Live the MPs

While writing these predictions, It was not yet known that AppLovin will have announced a massive layoffs in Adjust. I’m sorry and sad for the people who lost their jobs, many of which I know personally well. 
Adjust is/was one of the first and leading MMPs, but the future of an MMP’s CORE offering is questionable.

MMPs are a pillar in our industry, and I have respect towards the MMP companies. They are a source of knowledge, they bring the community together, and they are really awesome at making sense of data from many sources.

But in a world where MMPs are no longer doing the attribution, but only reporting on it, the importance and value of an MMP as a Measurement platform drops significantly.

Fingerprinting attribution is not a sustainable business, and definitely not one worth the $550M / year* Advertisers spend on MMP services 
(*my estimation of the total MMPs revenue pie. DM me for details)

I predict that the directions MMPs will go will be to any or all of the following:

• Creative generation, and A/B testing tools (organically or via acquisition)
• Monetization tools for Publishers (likely via acquisition)
• App Subscription management features (organically or via acquisition)
• Omni-channel ad spend consolidation (including TV, OOH, Influencers)
• Incrementality measurement product offering (NOT a service such as GeoLift Experiments) (organically or via acquisition)

While it’s a stretch – maybe some MMPs will venture into out of app store billing (acquiring AppCharge, XSOLLA, or similar)

MMPs NEED to change their core offering ASAP, as the attribution part lost its value.
Removing the “Measurement” part of an MMP makes it into an MP. Mobile Platform, or Marketing Platform.

MMM will go back to the attic

Media Mix Modeling platforms has been the pain in our asses during 2022 and 2023. We witnessed a dozen or so companies offering MMM platforms as a solution to measurement in a privacy first world. Many companies who tested and failed with MMM resisted in testing with INCRMNTAL, since they thought INCRMNTAL also used MMM technology. (for the millionth time – we’re NOT!)

Media Mix Modeling is an awesome marketing scenario planner when used correctly. MMM requires strong analytics, major assumptions, frequent calibration, and more than anything: Patience. 

MMM is not a bad methodology for advertisers to utilize, but there was a big gap between what MMM was pitched as vs what MMM was capable to do.

During the past years, MMM platforms pitched MMM as “ready to go”, or “replacement for attribution”. None of the MMM companies wanted to pitch MMM for what it is: 

• A heavy analytical service
• May require 9-12 months to calibrate
• Strict prerequisites to which Advertisers are a good fit (and which are not)
• Limited granularity (Medium, or channel level insights at best. Globally)
• High costs (compute + service charges)

Based on our conversation with companies who tested MMM, we assume that over 90% of those who tested it stopped using MMM.

I think that MMM will continue as an AWESOME tool for the companies who will use it for the purpose it was designed for.

Google Sandbox – Not in 2025

This prediction will be short.

I don’t think that Google will go ahead with its’ privacy sandbox in 2025, nor in 2026. 
Google has been challenged by several fronts with often conflicting allegations: 

• Privacy Lawsuits vs. Abusing its dominance in search
• Antitrust vs. Controlling demand & Supply

There are even articles saying that Google will be recommended to sell off Chrome…

Given Google’s unique positioning in our space as the #1 advertising platform – Google cannot afford making one sided drastic changes in its’ advertising measurement policies without getting extreme backlash.  

Therefore, my prediction is simple: Sandbox isn’t happening anytime soon. But it doesn’t matter. Privacy is moving forward.

Bonus Prediction: iOS App Store Redesign

While working on our upcoming white paper (App Featuring related) I noticed something about the history of the app store.
When launched (2008) the initial design lasted for 4 years. 
The second redesign lasted for 5 years.
It has now been over 8 years since the last major redesign for the app store.
I have NO IDEA what may a major redesign look like, and I have absolutely no evidence, rumors, to back my prediction other than the fact that Apple likes redesigning stuff, and the amount of recent technological improvements (anyone said: Apple Intelligence?) tells me that a major redesign is coming… 

Closing words: 

I do sometimes get it wrong though. As I started this article: The only constant in adtech is change. While my predictions are mostly based on evidence– the frequent and unexpected changes in our industry may often cause predictions to be off or even wrong. 
So, please don’t go buying stock options, shares, or shorting any companies based on my predictions alone.

I hope you enjoyed reading these, and do hope that my 2025 retrospective would prove most of these as correct.