If you’re curious to know what is multi-touch attribution marketing and why the multi-touch attribution model/multi-channel attribution is more beneficial than just giving credit to the last marketing touchpoint for all marketing teams, then our new video is for you.
In this video, Prajwal Barthur, InMobi’s Director of Product Management discusses why assigning credit after the first touch or closer to the conversion first came about and why it’s problematic to not distribute equal amounts of credit to multiple channels for conversion events. The video also highlights how any type of multi-touch attribution is preferable over allowing one channel to receive 100 percent of the credit, especially for real-time and data-driven marketing efforts.
Hi, this is Prajwal and I work with InMobi as the Director of Product Management. Today’s Whiteboard Wednesday, we’ll be talking about multi-touch attribution, its use cases, how marketers can utilize the same in their campaigns.
Before we get into multi-touch attribution, I just want to cover a little bit about performance marketing and how its evolution has been. We look at performance marketing in different ways, and what they want to drive is key performance indicators. One of them, we all know, is to drive installs.
The system has evolved to not just drive installs but drive an event post-install, which is basically maximizing lifetime value, or LTV, event for the user. There are a lot of campaigns which drive transactions, for example the retail use cases. And then a brand first customer base marketers who look at driving pageviews and also thereby increasing the LTV of users they acquire.
So in this performance marketing, attribution plays an important role in understanding which of these channels that we talk about drive the best ROI for them. And a marketer in general uses different channels for running their campaigns. They use search, which is one of the most important channels for driving volumes. Of course they use display, affiliate marketing and retargeting tactics to drive the performance marketing KPIs that we looked at.
In all of this, attribution plays an important role. And we’ll just talk about how attribution has evolved over time. Primarily in performance marketing, people use something called a single touch attribution model, which basically gives credit to any of these KPIs to just one particular touchpoint. And that can be the first touch or the last touch. So this is actually what a lot of the marketers use currently.
Here’s an example where this user here was displayed a bunch of ads on his phone - we’re talking about a single use case of app install. This is his customer journey through a period of, let’s say, 28 days. He has seen a bunch of ads that have been shown to him by various companies, be it Facebook’s view or InMobi’s view, and then clicks were by various ad networks. And finally he does an install at this point in time.
Now when you go by single-touch attribution, and go with the last touch attribution, the credit is probably given to the last click before this particular install happened, and that goes to one of the networks that has been mentioned here.
However, in the app install space, the credit is actually given to an install only after the install’s been open by the user, which is basically meaning the KPI’s not this but it’s moved to something called an app open, which is when the user opens the app. Now the credit has shifted from the network here, which drove the click, to a network which actually drove the click after the app was installed. And that’s what we see here, right? Now in a last-touch attribution model, this network driving the click before an app open is who gets credit for this app open if you go with the last-touch model. If you go the first-touch model, the network which drove the first view or the first click, in this case let’s say Google search (paid search), search click, will get attribution for this entire app open and LTV that is driven by this app open.
Now if you’re looking at transaction, which is an app event that happens let’s say on Day 28, none of these touchpoints matter at all. The marketer might have spent a bunch of dollars actually getting eyeballs from the user and actually making the user go through all of this, but this app event will be actually attributed to the last click that the retargeting network drives. And everything done so far has no value but all the credit goes to this.
Now there’s some flaws with this model. Of course it’s recently become more nuanced and a lot of people are making use of this. One is, one major flaw of this model is ad fraud.
If you look at the same diagram here, there’s incentives to drive multiple clicks, assuming that this particular click will somehow become the last touch or first touch, and that gets attributed to the KPI that is in mind. Here what I’m showing you is what’s called as ad stacking, when multiple clicks from the same device are triggered off.
There’s another type of fraud which is induced in such environment, which is basically called click sniping, where just before an event happens which is the KPI that is measured for, a click is driven, which is this click. It can be a valid click, but a lot of fraudsters figured out that because the attribution is last touch, they kind of inject these clicks so that they get credit for any of the opens or any of the KPIs that the marketer looks at. Now, so this is what single touch is all about.
Now if you have to move this away and then give credit, due credit, to the networks that are playing a part in achieving the KPIs, the marketer has to work with vendors to do multi-touch attribution. What does this mean? If you look at this example itself, there are at least 10 different touchpoints before a KPI happens. If you have to give credit equally and distribute the credit to different networks that actually engage with the user, you need to employ something called a multi-touch attribution model.
And why you need to do that is to make sure that fraudsters don’t get into this model and then take credit because they drove the last click. It’s the right thing to do as well because you’re actually giving credit to networks that actually influence the user.
Now how you want to give credit is based on the modeling that you will use. It can be a model where you give equal credit to all the networks. It can be model that will be driven by data, which says that these are certain networks which have influenced the user and so I give higher credit to those networks.
Now in moving to multi-touch attribution, these are the few things that any marketer should be aware of. One is you need to pick the right vendor who’s able to actually collect all of these data points, which I’m basically calling out as data collection. It’s important to understand different touch points of user.
Especially when performance marketing is moving to a cross-device environment, it’s no more about collecting clicks and impressions from the same device. It’s about collection of data points across different devices.
Now the next part is using the right channels, because there are some channels which are incentivized to do fraud. You should weed them off, and the only way to weed them off is figuring out channels that drive incrementality, and that goes back to finding the right vendor who can actually work with multi-touch models.
Now the modeling itself is another important aspect to multi-touch attribution, to figure out which is the right model that you need to use to make sure that you’re incentivizing the channels which drive your incremental KPIs - that can be installs, that can be revenue, that can be transactions, whatever you are actually looking at.
Note: Common weighted modeling options include the U-shaped model, the W-shaped model, the time delay model, full path model and the linear model. Some companies also elect to go with custom models depending on need. Check out this post from Bizible or this post from Adjust for more information on these different options.
So I would just want to end this session by calling out these four key important things that a marketer needs to look at, 2019 and beyond, because performance advertising is moving multi-channel, and working on single-touch attribution does not solve for all use cases.
Be sure to reach out today to schedule a chat with us on how multi-touch attribution can boost your campaigns and what you need to get started.