11-07-2017

This blog post was originally posted on marketingfacts.nl by our colleague Sylvia Vroklage

Last-click attribution has its time, I wrote in the article “Why e-commerce and traditional retail look alike”. I recommend that you review that article before you read further here. In this article, I’ll give you five tips to get started with a attribution model that’s future-proof.

[In February, I completed my Executive MBA at Nyenrode University with a thesis titled “The conversion Attribution Riddle”. Part of my research was a series of interviews with CEOs and managers, among others, from the top 10 e-commerce companies in the Netherlands. My research is the basis for this article.]

Marketing Management: A game with incomplete information

If you followed the link above, you know that – while last-click conversion attribution  leads to suboptimal marketing investments – managers base their decisions, to some extent or even completely, on such a model. Why? Because marketing management is a game with incomplete information. In such a game, there are things that you know, and things that you know you don’t know.

For CMOs, last-click belongs to the first category; Things of which they know the value. So they put it in. But what about the second category? How to learn more about channels and touchpoints in the customer journey whose value is less well-known? Through my research, I distinguished three gaps when it comes to conversion attribution:

  • Multichannel: Mapping both online and offline channels, and the ways in which those channels influence each other is not possible in the current models;
  • Multidevice: Tracking customers across multiple-screens and devices is complex;
  • Relative Value: Common models make it difficult to determine the relative (weighted) contribution of online channels to a conversion.

What can you do to seal these holes? An important question, after all: managers with the most valuable information make the best decisions, and in the end win the game. The managers I spoke to for my research come up with five recommendations:

  1. Begin small. Test, test, test and learn!

Especially when you start with attribution, it’s important that you keep it simple. Start with one or two product categories, or just with your online channels. This way you can easily gain valuable insights.

To name a few: Branded search has relatively little value to the managers I spoke with. A customer has often decided to purchase at the moment they search for your brand name. The value of cashback sites is low as well. On the other hand, opening an email has a relatively high value, even if such a touchpoint does not directly contribute to a conversion.

Remember: Conversion attribution is a tool, not a goal in itself.

  1. Define objectives and KPIs

“If you do not know where you’re going, any road will take you there.” Remember: Conversion attribution is a tool, not a goal in itself. Have you determined which moments are decisive in a purchase? Then your next step is to connect the right KPIs to those moments. Channels like Facebook or display ads are often the first moments consumers have with your brand. Therefor you want to send those channels to other metrics than, for example, affiliates or comparison sites, which only play a role later in the buying process.

  1. Choose the tools that suit your situation, budget, and level.

For your first experiments, certainly choose simple tools with a steep learning curve. I’ll reference a few.

Double Click from google was mentioned by almost all the managers I spoke with. With Double Click, marketers have a data-driven attribution model that make investments shift towards the beginning of the customer journey.

Device matching, which happens when a consumer users the same email address to sign in to different devices on your site, app or service is a great way to seal the multidevice hole.

I recently saw a very creative solution from Mediamarkt, which, with their StoreMode app, tries to link mobile traffic to physical shopping. And thanks to Beacon technology, smart TVs and streaming services like Spotify, segmentation on traditional mass media comes closer and closer. Through this the gap between online and offline is continually growing smaller.

“Useless data can always be discarded at a later date; Valuable data, once gone, you will never get back.”

More complex tools are Attribution 360ConvertroAnalytics from Adobe, and Visual. Also, in the academic world, attention has recently been on conversion attributes. Academic models based on, for example, Baysian logic or game theory, are aimed primarily at solving the third gap – thus the problem of the weighted contribution of channels and touchpoints. Currently, however, the academic models are still mainly that: academic.

From this category, special attention deserves be paid to a recent release: Google Attribution. Attribution (now in beta) gives organizations access to a data driven model that easily integrates into your existing Analytics and AdWords applications. The best part is: it’s completely free. There are a few people that have had a good look at it so far, and for the time being, Attribution only integrates online channels, but this is definitely a tool to keep an eye on.

  1. Collect clean data in one place

More than half of the managers I interviewed pointed out the importance of collecting clean data in one central spot. Marketing departments often consist of various teams, all of which work with their own tools and data. This could result in different teams claiming the same conversion. With an integrated solution, you make sure all stakeholders look at the same data.

Save all of your data, even if you don’t get started right away. Useless data can always be discarded at a later date; Valuable data, once gone, you will never get back.

  1. Work constantly on your marketing team’s knowledge

‘Statistical analysis and data mining’ is the number two ‘top skill’ in 2016 according to LinkedIn. With good reason, because there is a huge gap between the demand for knowledge in the market and its supply. Eight out of the 10 manages I spoke with recommended to hire at least one data specialist or to train someone within your company for that role.

The good news: As a smaller e-commerce player you often have a head start on the bigger  guys.

At the same time, it’s important that your entire organization has at least some idea of exactly what you’re trying to achieve and what essential role data and analysis play in that. Eventually it becomes about really understanding your customer and for that you will need all your eyes and ears.

Little is nice Little is good

The good news: As a smaller e-commerce player you often have a head start on the bigger guys. With large players, not only data, but also the teams are often in silos; With fragmented customer experience across the different devices and channels as a result. Keep in mind,  fewer customers and fewer products often mean less data, easier data management, and less diverse and complex customer journeys. Less legacy means easier integration of the various tools with your CRM or DMP.

Last but not least: As a smaller and younger company, you are more likely to be more open and move faster towards an organizational change that focuses on your client.