How to use Funnel and Adtriba to create a powerful solution.
The attribution models many currently use, like first or last click, are often only considering a few pieces of the puzzle. Successful and failed conversions, mobile and web activity, and much more is usually missing. That may sound like a lot of data to collect and analyse, but that’s where Funnel and Adtriba come in.
In our webinar with Adtriba’s co-founder and CEO, János Moldvay, we take you through why multi-channel attribution that is fuelled by the right kind of data can help you make the best decisions for your business and marketing budget. We also do a quick dive into how the two solutions work together to provide the most holistic view of your marketing data.
You can learn more about our partner, Adtriba, here.
- 4:56 What exactly is data-driven multi-channel attribution?
- 18:40 How Adtriba Works
- 24:53 What is Funnel?
- 28:52 Why Funnel & Adtriba have partnered
- 33:47 Benefits of Funnel + Adtriba
- 36:05 How to get started
- 37:03 Live Q&A
[7:58] Why data driven attribution is better than the static models (first/last-click) mostly used today.
[Using last-click attribution] Is similar to only paying the goal scorer on a football team. It doesn’t really make sense for the goal-keeper because it’s like only paying a salary to the players who have scored a goal and obviously leads to inefficient games and plays. And in football you would really never do this, but in Marketing it seems to be very widespread.
And even if there are other static attribution models such as first-click attribution, or linear-attribution - linear meaning that you would give €25 of credit, or attribution value, to each of these 4 different touchpoints - you don’t really do that in professional sports either, have it so that every player gets the same salary. But what’s more common is that every player is getting paid according to their efficiency or their importance for the team, and we should probably do this in marketing as well.
[9:16] Why you should look at the non-converting customer journeys as well as the converting ones.
In football you would not only look at the successful passes that a mid-field player has played, but also at all of the unsuccessful passes to really evaluate the efficiency of that particular play. And that’s the same that we should be doing in Marketing, we should be looking at, how many times has email happened as a touchpoint in the converting journeys versus the non-converting journeys? That, in essence, is what data-driven attribution is about, and the multi-touch aspect just means that you do that for all of the touchpoints. So not only within a particular ad platform, but for all touchpoints…
In essence, how our data-driven multi-touch attribution works, it’s a daily update of the machine learning based model...we apply automated machine learning, it’s based on a particular type of deep learning framework. It’s called Long Short Term Memory Networks... LSTMs are a specific type of deep learning and they’re really good at understanding sequential data. Google used to apply them to the Google Translate product quite successfully, because, obviously, language is just a sequence of sentences, and sentences being sequences of words, and words being sequences of letters. So that worked quite well for Google, and works very well on customer journey and user journey analytics.
We also consider the time difference between the user journey touchpoints - meaning that for us and the algorithm it makes a huge difference if a display view has happened an hour before the conversion, or a minute before the conversion.
We also include onsite events, for example ‘add to cart’ or whether the product has been viewed in detail. [We do this] because you really want to make sure that at each stage and each touchpoint you know exactly what the inherent intent is for the client to purchase, and what the probability to actually convert is.
So for example you want to know after an ‘add to cart’ event when the retargeting pixel has fired, how long does it take for a retargeting click to actually be effective and not just piggy-backing on the already high-intent of the user (which you can tell by the user having added the product to the basket)?
[37:00] How do you (Adtriba) track users which do not allow cookies and ad blockers?
The short answer is our tracking is more or less cookie based, so the standard tracking that’s implemented would not be allowed to track these types of users. I think a kind of underlying question here is also, what do you do with Safari users or ITP tracking prevention? And there are different types of ways to circumvent that; one heavily used right now is called DNS Cloaking...but we wouldn’t really recommend it, there are other processes that work much better. For example, overriding the cookie duration via the backend implementation. And, ultimately, what one should aim for is basically back-end tracking which we are also working on a standard solution for that. Google Analytics has a standard SDK for that, I think Segment as well. So I think that’s a kind of related question to that.