When a measurement tool provider doesn’t appreciate the value of accurate data

Recently, an article was published on VentureBeat summarizing various marketshare stats provided by AppsFlyer. While I have great respect for VentureBeat and Dean Takahashi, who authored this piece, I take issue with the wildly inaccurate data sets presented by AppsFlyer in the originating press kit. Of note, AppsFlyer makes statements of marketshare and churn based on data sets attributed to MightySignal.

A few things to highlight:

  • MightySignal only detects SDK integration with apps. If an app using a measurement tool and instrumenting via a server-to-server approach (which many top apps do), any SDK detection tool wouldn’t have any method to identify it. We built our platform for flexibility in integration which means we can support clients without a rigid requirement for SDK integration.
  • The SDK detection space (of which MightySignal, Datanyze and others are categorized) is consistently inconsistent with one another. This is not a measurement vertical that has a track record of deterministic and consistent results across vendor. As an example, MightySignal will detect an SDK used in an app and Datanyze will not — and vice versa.

Based on our own assessments, Kochava represents over 10% market share from the top 200 apps — and over 50% for the top 10.

Further, I know we have some of the lowest churn rates in the industry. My data science team ran the numbers again this week; we show 2.5% (or more descriptively— a 97.5% retention rate)! We know who churns and we have data to support it.

I think I’ve made my point about the accuracy of the stats presented by AppsFlyer. If these metrics were both inaccurate and presented with bold bias, it begs the question if there was inaccuracy across the other data points presented as well (including the other vendors mentioned). Further, what’s the reason to push such an inaccurate narrative that relies on spotty metrics?

Aside from cringing at the inaccuracies, reading the article prompted me to ask an almost amusing question: Is this a cry for help?


The following is a bit of a longer read, but I’m sharing because 1.) I fundamentally believe it and 2.) I believe there are many in the industry that are interested in a deeper look into the macro-dynamics of this industry, and I’m happy to provide my perspective.

A bit of background. Very clearly, Kochava is the leader in providing the most advanced tools for marketers across planning, targeting, activation, measurement, and optimization. I have written several posts about what makes our leadership unique — I will spare you the re-summarization. Two of the recent, most notable ways we have innovated beyond being a traditional MMP is with the availability of the Kochava Collective (the largest independent mobile data marketplace with over 3.5b mobile devices available for targeting and audience enrichment) and the addition of Kochava Engagement as a native capability for push and in-app messaging in the platform.

We have historically (and intentionally) focused on the head of the market (where a smaller number of large advertisers spend big budgets on ad spend) vs. the torso or the tail of the market (where a large number of individual apps reside, all with smaller ad budgets). We have done this for a few reasons:

  • We wanted to establish our brand as being the white-glove provider in the market — where crazy customer requirements are possible because of the flexibility of our platform and premium service of our team. The head of the market values these attributes.
  • The head of the market values ‘signal’ (the quality of the full stream of click and install data), not simply conversions. This is something we are uniquely passionate about as a company…working to educate the industry about how they can fight fraud and junk traffic, and providing solutions.
  • The head of the market appreciates that the value of real ROI on ad spend massively eclipses price premium on a more advanced measurement product over a commodity MMP.

Torso and tail customers don’t need the things mentioned above. They need a basic tool that generally works and has the same common graphs/charts presentable. This is what I call black-box attribution and it satisfies many, many customers. It’s also a commodity.

In general, I’ve reduced the macro market to date as depicted in the following graphic:

Because of the focus on the torso and tail, Adjust and AppsFlyer have aggressively competed with one another, driving each other deeper into price-focused competition (vs. unique value-add capabilities like configurable attribution, attribution back testing, fingerprint equalization, real-time alerting, traffic verification or add-on solutions like the Collective or Engagement, for example). In the meantime, some larger customers (no doubt) selected AppsFlyer or Adjust along the way, but I believe they did so not knowing what is possible via a premium measurement provider like Kochava (especially given their spend amounts in media). It’s a relatively new space — it’s understandable! Singular is the flyer — because they bought their way into attribution and have a solution set that historically focused on cost aggregation and later augmented with measurement via the Apsalar acquisition.

Adjust has a better UI than AppsFlyer and a lower cost to operate (because data isn’t persisted long term for their customers and record level insights are purged) resulting in Adjust being profitable due to lower cost of infra/delivery. Customers that use Adjust typically have their own in-house data systems so the absence of persistence hasn’t historically scared them off. AppsFlyer, while providing record level data for 90 days, is losing money at scale with a price-focused strategy.

Not to be outdone, I’ve been told by prospects that AppsFlyer sellers introduce themselves and start their pitch with a primary selling point of being ‘the cheapest’. While this approach gains market share in the torso, it does not serve head customers and, over the long term, it doesn’t serve the torso either. We are seeing that reality come to fruition now.

Subsidizing measurement (which is arguably the smallest percent amount of a media budget and has the opportunity to move the needle the most in media spend efficiency) is not something we’ve subscribed to.

Eventually, rates have to go up or the competing commodity vendors have to keep raising capital to subsidize pricing. There is an alternative — vendors taking this strategy could selectively choose to make larger customers’ pricing lower (due to volume), and charge more per transaction to the small guys, making up the difference in aggregate… more on this later.

Macro Dynamics of the Measurement Market

Two years ago, Kochava recognized that early-stage tail/torso marketers wanted to use ‘something’ as they first started their initiatives, and a cheap solution was more attractive. To these prospects, we would espouse the merits of configurable attribution, fraud abatement, and correlation of signal via an R2 graph and they would look at it like we were speaking greek. We would highlight that we were a solution to grow with — the whole value prop. At the end of the day they had simple requirements to start and they were super price sensitive.

AppsFlyer and Adjust leveraged this and characterized Kochava as complicated or expensive. They said: “you don’t need to care about clicks, just conversions” (a recipe now better understood by the rest of the market as a framework that enables fraud at scale).

The trouble is, after selecting a simple tool, when those marketers grew, they would grow out of AppsFlyer or Adjust, but would then find themselves stuck, because to unwind the integration with their existing SDK required time and dev resources. This issue alone is enough to lead a marketer to limp along with an inferior tool for longer than they’d like.

To serve early stage marketers with a more simple solution, we decided to provide a basic version of Kochava (which perfectly suits the requirements of the market’s tail) for free. We call this FreeAppAnalytics.com and we came up with a more feature-rich attribution analytics package for the torso (but shy many features from enterprise). Those early stage marketers could get started with FreeAppAnalytics and upgrade as they grew.

The data from FreeAppAnalytics is first-party licensed to Kochava, and monetized to cover the costs of the free service. The data from FreeAppAnalytics feeds the Kochava Collective, which has become the world’s largest independent mobile data marketplace. In our experience, FreeAppAnalytics customers understand fair value and are comfortable with licensing their data in exchange for the tools. To be clear, none of our enterprise customers data has anything to do with the Kochava Collective. Further, as soon as a FreeAppAnalytics customer upgrades, that data feed is turned off. Interestingly, customer demand for the Collective data is largely from our enterprise customers who buy the data for targeting and audience insights against their own first party audiences. Many Adjust and AppsFlyer customers currently spend more on data from the Collective for targeting and enrichment than they do on measurement from their legacy tool. This is an example of value.

Instead of having an expensive premium brand to compete with on price, last year AppsFlyer had to start contending with Kochava providing a free version where upgrade vendor lock-in wouldn’t apply.

Then, Branch got included in the Facebook MMP program — frustrating AppsFlyer’s attempt to be ‘the cheapest in the market’. Branch has remarkably raised even more money than AppsFlyer and is unashamed in their approach to deliver a free product. One could argue that Branch is attempting to out-AppsFlyer AppsFlyer.

Mobile App Economy Second Act

Eric Seufert wrote a great piece recently about the Second Act of the mobile app economy.

If you haven’t read it I’d encourage you to do so. If you think about the implications of the second act in the context of measurement and targeting tools — things are going to need to get more advanced in measurement (not more commoditized), data and activatable access to data is key to success, and finally, the pie is getting larger. Where formerly just the head of the market was the leading buyer segment, it’s trending towards more torso marketers. This means the torso needs head-like mature tools.

AppsFlyer has so focused on being the low-cost leader that its difficult for them to invest in these more mature tools — or differentiate the basic tools from the advanced tools by raising their prices (especially with the existence of branch) — but they need to raise them eventually as burning cash means the money will run out. Further, they can’t just keep the top end customers pricing cheap but charge more for the small guys on a per-transaction basis — because branch and FreeAppAnalytics are applying pressure with free solutions. All the while, the torso AppsFlyer customers are demanding more intricate tools as they mature and those tools are not historically a strength of AppsFlyer.

A Perfect Storm

So, while my question about whether the article was a cry for help seems amusing — I think there is some merit.

AppsFlyer needs to either get another round closed (which would make the already encumbered liquidation preferences on their cap table feel easy), or, it needs to have an exit (get bought). Despite the improving economy, my experience tells me that no large public company wants to carry a burn — they want to buy a net-neutral asset or one that’s profitable and able to grow. Further, no company that understands the macro dynamics of this sector will miss the reality of this moving market.

This article appears to be a public announcement that something needs to happen for them as their burn is catching up to them AND the stated metrics (inaccurate as they may be as they reference us) are likely going to be changing for the worse over the next year. I believe that the next year will show:

  • There is only one premium measurement brand — focused on ROI — not commodity — and deep as an ocean in features for advanced marketers.
  • The premium measurement vendor also has a mature targeting and audience enrichment framework (the Collective) — critical in the next wave of mobile. Because non-Kochava customers can use the Collective, further parasitic drag will be applied to maturing torso media buyers that use AppsFlyer for measurement and the Collective for targeting.
  • Several vendors will continue to try to out-subsidize each other on price.
  • FreeAppAnalytics is going to start to significantly scale as a result of several channel partnerships we’ve established — putting more bottom-up pressure from the tail on those above mentioned vendors — vendor lock in won’t be as strong of a consequence for early stage marketers who mature with break-out apps and scale their media spend
  • 1 year + veteran marketers are going to want a more advanced tool than they have — resulting in churn metrics in AppsFlyers business — going to the premium provider or resulting in strong product pressure.
  • Particular verticals — like the hyper casual gaming market will realize that the commodity vendor is actually more expensive than the premium vendor — making up the pricing gap by charging overage charges on conversions and resulting in poor overall ROI.

My perspectives are just that — my perspectives. Having said that, the data indicates a different reality from that which AppsFlyer is desperately trying to convey to the market. Regardless of your position — I think it’s going to be a pretty exciting year.

If you are one of the torso marketers I’ve described — I’d welcome the opportunity to have our team serve you and help you make the move to a premium vendor. If nothing else — we’d love to show you what’s possible so you can make your own assessment.