Attribution Part 3: From obstacle into opportunity

Summary

• By now, you should know why attribution is so important and how to build an attribution model.

• Often, however, the output of all this hard work is a rarely-used dashboard by a single team, due to organisational barriers in rolling out the attribution model. Some of the biggest barriers include misaligned incentives legacy processes and siloed platforms.

• In this article we discuss different ways to prepare for and overcome these challenges by getting buy-in at leadership level, running scaled training programs and integrating output data into current platforms.

Don’t let the Attribution Monster escape

You try and try again, but every time you go in for the catch, the Attribution Monster finds a different way to escape. Dodging this way and that, it seems to know a way out of every situation. However, you are better prepared and start learning its tricks, so that every day you get closer and closer to catching it.

Finally… after months spent understanding why and how to catch the Attribution Monster, you succeed.. You finally catch The Attribution Monster!

Fail to prepare, prepare to fail

Change management is a daunting task, and many people start along the journey and don’t follow through. Much like trying to catch the Attribution Monster, you need to prepare by understanding the ways you might fail and take action so that any obstacles become opportunities. We believe those actions fall nicely into the People, Process, Platform framework which we commonly use here in SingleView:

• People: Aligned incentives

• Process: Frictionless processes

• Platforms: Automated platforms

People

One of the biggest blockers is a lack of organisational support based on differing incentives. Imagine your display marketing manager finds out you are thinking about shifting to a non last click model. He is furious and fumes: “If my responsibility is to manage the retargeting spend, and my bonus is based on performance of retargeting then I would have no incentive to move to anything but a last click model. What do I have to gain?”. So how can we create win/win situations and drive this organisational support? The key is changing the incentives. But to reach that end goal requires the right preparation.

Let’s take a step back and agree on why we want to move to a different attribution model. Remember, the attribution model is not the outcome. It is a tool to achieve your actual goal. So what do you want to achieve? Do you want new customers? Increased spend? Greater profit? We need to get buy in at the highest level, so understand what the CEO or CFO cares about so that they appreciate the value of the model. Once we have an idea of the KPI, we are ready to implement an experiment.

Based on the KPI and the insights from the model, frame a hypothesis e.g. ‘Shifting $100k from Paid Search to these specific Display campaigns will drive an incremental $500k in revenue over the next 4 weeks’. Write down the hypothesis and a potential outcome. Most importantly have an idea of what you might do if the experiment is a success. Will you go to the CFO? Or would you scale up the experiment to cover more channels? The goal is to show how using attribution insights drives results.

Think about that one small experiment that, if successful, would deliver results capable of convincing anyone it is worth their attention/time. Find a big reason that the CEO should care about. Ideally, this experiment is small enough to not affect anyone’s day to day processes yet big enough to cause a visible difference. Find a team that might benefit from the experiment and work with them to run the experiment.

Now let’s say we have defined a core KPI and the experiment has shown that our model results in an increase in the KPI. We may need a few more experiments to build a more solid case. Eventually we present to a more senior stakeholder and they note the impact that this change can achieve.

Now comes time to align the incentive program: we need to restructure incentives to encourage optimisation across channels as a pose to within channels. We might start by measuring team performance based on overall sales vs channel sales to encourage collaboration. This will help us encourage the movement of a portion of one team’s budget to a different team to increase overall sales.

Process

For the successful implementation of an attribution model, the data and the recommendations should be used in daily operations and spending decisions, which may require new processes and new platforms (we will come onto the second point in a bit). Not only that, there needs to be training so that the team understands what they do with the data, and how it will help them. Ideally, we minimise the process changes necessary, keeping systems the same so that friction does not become a factor preventing mass adoption. As an example, if the team uses a specific Analytics report to view the data, we add in the new metrics to the same report rather than creating a new report.

Train the team on how they can use the data, what it means, how it will help them and what different process they need to go through to use the data. Stick to existing processes if possible, at least at the start, whilst the team becomes more confident in the new measurement metric.

The team may need to hire new insights strategists to navigate the new landscape, and translate attribution reporting into recommendations. However, start with your existing team, invest in an automated process, provide the training, and then trust the team to have the intuition to execute on the new recommendations.

Platforms

Often teams are wary about using the platform from a specific provider because of perceived biases, but having one source of truth is essential. Most importantly, the new platform should allow easy integration of attribution data and connect directly with marketing and bid management platforms.

Find a platform that connects directly with bid management platforms and can act as the single source of truth. Make sure you understand the technicalities of the platform: what is the look back window? How is the data stored and is it easily accessible?

Within Google, we used to have ‘platform experts’ whose role was to know everything about the tool (this was even included in their job responsibility). With so many different platforms, all changing on a regular basis, it becomes difficult for one person to manage them all thus having a platform point of contact really eases the transition to the new software. Not only that, this person can be responsible for training and also finding new ways to leverage the functionality of the platform.

Change necessitates broad thinking

As you can see, you can’t implement a data driven attribution model simply by sharing a report… instead you need to encourage organisational change. If you have not been able to drive change in the past, now might be the time to give it another go!

At SingleView we believe that data is only a tool and it is the People, Platforms and Processes that make the difference. We follow a consultative approach with all our clients, guiding clients through the entire end-to-end process and ensuring that they aren’t left with just a report, but instead are left with a way to drive change in their organisation. Often you don’t need a paid Analytics platform to build a model: you simply need ‘log-level’ data to understand the interactions that each user makes. But remember, the ‘perfect’ model often isn’t the only thing that you need. Change necessitates broader thinking around organisational barriers and how to overcome them.

Contact us for a free consultation and find out how we can help solve your attribution challenges.