Best Practices, Heroes

Last Click Attribution Is Dead. Now What?

Stara Gipson and Brad Puder by Stara Gipson and Brad Puder | July 9, 2018

Stara is a Performance Marketing Manager at Pocket Gems. She focuses on Episode, an interactive storytelling platform. She started her career in scientific research, but eventually made the switch to tech. Her first tech job was working for a very early age startup, doing everything from marketing to web development. She ended up at Oracle working in the Marketing Cloud and finally found her way to Pocket Gems.

Learn more from her Mobile Hero profile.


In a gaming marketer’s perfect world, every person who interacts with an ad or our brand will fall in love with the game and become a VIP, high-value player at that instant. But in reality we know that is not the case. It can take 5, 15, even 30 interactions with an ad or a brand for a person to convert to a player, let alone a purchaser. So as marketers, if we can accept the fact that it takes multiple touch points to convert a user, why do we accept only one touch point as our sole source in regards to attribution?

The vast majority of the industry currently works off of a last click attribution model, but as technical barriers are broken down and buzzwords like multi-touch attribution become part of our daily jargon, the days of a last click model will soon be behind us.

Why a last click attribution model is insufficient

We know the obvious, that a last click model isn’t a holistic representation of a user’s journey, but how does that affect the bigger picture: growth and revenue? When attributing all installs to a single source we might see that a given channel has a negative return on ad spend (ROAS) and in turn allocate spend away from that channel to a more positive or neutral channel in terms of ROAS. This leaves room for two holes, one for false negatives and one for false positives. The channel with a negative ROAS may be contributing to installs down the line and be undervalued. Meanwhile, the channel with a positive ROAS is getting more credit and being overestimated in terms of return when it’s success is due to a chain of events that got overlooked.

This becomes even clearer when thinking about the types of inventory available in any product’s marketing playbook. For example, passive banner ads may be less likely to drive clicks than native or video ads, but the content is still being consumed by a user. A user may see 5 banner ads, make a search engine query for the product they saw, click, and then download, sign up, or install. With the last click attribution model the conversion credit would go to the SEM channel source. But what if we were able to follow that same user from the first impression all the way through conversion and attribute the credit accordingly, what would that mean for the greater business?

Benefits of multi-touch attribution

1. Increasing internal optimization accuracy

Some of the largest strides resulting from multi-touch attribution can be made internally and lead to more accurate revenue metrics. If we take the example above of channels with passive banner ads driving less clicks but still having an impact on conversion, we could attribute a portion of internal revenue to these channels in turn increasing their ROAS. The same goes for over-attributing to SEM channels, with that revenue dispersed across all touch points leading to conversion we have a more accurate measure of ROAS and can then allocate spend accordingly.

2. Frequency Capping

By having the technical capacity to track a person through all ad interactions we can minimize the chances of oversaturation. Current tracking models allow for ad limitation by channel, but with a universal multi-touch solution and the ability to track a single user across channels we can set caps at the user level rather than at the channel level.

3. A deeper look into organic installs

With single source attribution when users get attributed to organic installs we have no further data. By moving beyond this it provides a deeper look into how organic users came to conversion and if there is room for optimizing for these types of users.

Challenges of multi-touch attribution

Although moving beyond a last click attribution model feels closer than ever, there are still some major challenges standing in the way.

1. Universal adoption of the same model

For a multi-touch model to be used in it’s full capacity all partners must buy in to the idea. We see the issue today even with last click and self-attributing channels like Facebook. If only some channels buy into a new model it can lead to incomplete data which is just as bad as incorrect data and even overpaying for a single conversion.

2. Big channels will be affected the most

As mentioned above, big players like SEM and social channels, which claim the majority of conversions and spend, may be impacted at scale if their conversion attribution is cut and divided amongst other channels.

3. Data limitations

With changes to privacy regulations like GDPR and increasing numbers of users with LAT turned on, the data may simply not be available for a multi-touch model that can be used across devices and channels. Although the industry is breaking down barriers in technical limitations, new obstacles continue to arise.

It’s clear that a last click attribution model is not sufficient for modern day, direct-response mobile marketing. But what does a universal multi-touch solution look like? How do we overcome the increasing limitations in terms of tracking and privacy? And how is it implemented at scale? These are all questions that the acquisition space is on the horizon of resolving. As marketers we can educate ourselves to adapt and prepare to innovate on the many possibilities that multi-touch attribution will open up.