Attribution Part 2: Taking the plunge into the world of attribution

Summary

• By now you will have seen where attribution fits in your overall measurement strategy. As you start on your attribution journey, the first stage is to build an attribution model to measure the true value of your media. It is often easiest as a first stage to use your current reporting platforms (such as Google Analytics) whilst also noting their limitation, such as low actionability. A next (albeit more time consuming) step would be to build a custom attribution model that can support through providing both flexible reporting and increased actionability.

• After (or even before) you start attributing revenue in a more data driven way, think about the types of actions that you can take with the attribution insights. Most commonly, they can be used to influence budgets, optimise campaigns and to build cross functional teams that are focused on overall performance as a pose to media specific KPIs.

Catching the Attribution Monster

Then it happens. You catch a glimpse of the fabled Monster in the distance, no more than a flash of the tail. “This is it…”, you think. It’s time you try to catch up with old ‘Lochie. You’ve been studying its patterns day and night, your mind poring over the best way to catch it. Finally now, you take the plunge and low and behold — with the knowledge you’ve built up — you manage to get ahold of the beast.

What are we talking about? That breakthrough moment you first make sense of the attributed impact of your media spend.

This Attribution Monster is sneaky; it’ll easily slip between your fingers at the first chance, so once you catch it you must take action (with the attribution insights)… for once it has gone away it will be hard to find again.

Finding your glory-hunting strikers

Have your top performing channels become glory-hunting strikers shining in the limelight whilst other team players sulk silently in the background? To find out, you need to start distributing credit where credit is due by measuring the the assisted conversions with an attribution model.

An ideal measurement solution would fully integrate with your current reporting capabilities: attribution should become an after-thought, so that every change that you make to your marketing, whether to budgets, bids or creative, is based on a more data driven reporting as default. However, as Google realised with the failed launch of Attribution 360, this panacea is far from being a reality. So as you take the plunge, you find yourself limited by your your current reporting capabilities. Let’s explore a few popular reporting solutions and understand how they handle attribution, from the perspective of reporting and actionability.

Google Analytics has its own attribution modeling solution, which performs strongly in reporting assisted conversions. It is simple to use, providing one-click access to attribution reporting, and allowing easy comparison with a default model (and remember to compare to Last Non-Direct Click, the default GA attribution model). However, what Google Analytics has in reporting simplicity it loses in actionability, with siloed reports allowing once-in-a-while accessibility of attribution insights, and little way to integrate with other data sources and no easy way to export out.v

Figure 1: Model Comparison Tool in Google Analytics

Marketing platforms such as Facebook and Google Ads also have their inbuilt attribution functionality, however, since they are based on media bought through their own platforms, their reporting capability is limited. However, where these solutions score low on reporting, they often score higher in actionability, as they allow the switching of the default attribution model to a data-driven attribution model. As the famous philosopher Ockham once said, the simplest solution is often the best, and in this case the simplest solution would involve changing the default attribution model so that you can optimise to more useful attribution model, and not think further about it.

Figure 2: Edit Conversion settings in Google Ads to switch the default attribution model

inally, for all those fearful of the death of the all-precious cookie, Google and Facebook clean rooms will provide a consistent ID for your custom attribution modelling. As an example, Ads Data Hub allows you to export user journeys across Campaign Manager, Google Ads and YouTube, as well as measure impressions across these journey. However, be aware of privacy thresholds, where any paths included in the attribution modeling tool has to have been taken by at least 50 user IDs. In addition ADH provides impression level data for when (and only when) you have mastered the basics of click-based attribution.

A colleague who has often been frustrated by the complexity of the attribution world, shared this useful framework to help in thinking about the different providers.

Figure 3: Finding the optimal measurement solution involves tradeoffs

However, where these platforms really provide value, is when you move beyond the standard reporting and start to build your own custom equipment (in this case, custom modelling) to catch the Attribution Monster. This provides you with the benefits of the reporting capability of Analytics platform, but with the additional actionability of the Analytics platforms, given the ease of integration with the output data with other data sources (and into your current reporting). This also allows you to track longer paths (GA attribution is limited to five paths), gives you the option to include impressions (if working with Ads Data Hub) and also allows you to include offline conversions. As an example, we recently built an attribution for a leading retailer in Australia and found that Facebook played an upper funnel role in driving in-store conversions, but interestingly, not on online conversions: this convinced the retailer to refocus their investment on social channels.

If you’re interested with building a custom model, the first stage is to start capturing user level data (which GA4 provides at no extra charge with its data stream). Once this data has been collected for a few months, ask your friendly data analyst (or even ask a data partner) to sprinkle a bit of coding magic onto the data, and build a custom attribution model. Finally export into a reporting platform, to allow teams to start taking action; catching the Attribution Monster without taking any action will provide you with the bragging rights which will soon become an afterthought when the next shiny thing comees along.

Measurement without action

I can imagine some of you may be thinking: “So why should I implement data driven attribution? I have so many priorities and I can’t see the benefits of prioritising this. It seems so hard and the impact seems so small.”

You may be raising a fair point. If you have short customer journeys with few channel interactions (such as with the food and grocery sector) then maybe you should not be worrying too much about attribution. If you don’t track conversion (or any type of event), then you have other priorities to address first. A quick check: Look at the percentage of conversions with three or more touchpoints in your analytics platform. If a significant proportion of paths only have one touchpoint, then attribution will probably not be the biggest priority for your business at this stage.

If not, read on. There are three areas where a better attribution model will make a big impact:

Allocating budgets: Do you really need to allocate all that money to affiliates? Should you cut video spend? During weekly revenue reporting, use attributed revenue to understand the true performance of a channel. This might raise your heart rate a little, particularly if the changes are negative for the channel that you manage. Later on I will talk through ways that we can change incentives to ensure that any changes in reporting result in a win/win outcome for both the business and yourself. But initially, think about reallocating across your channels to drive greater outcomes e.g. shifting budgets from the Brand Search campaign to the Generic Paid Search campaign. The result will be more money allocated to the campaigns and channels that influence consumers to convert and overall revenue will increase.

Optimising creative and targeting: What are the goals of your upper funnel media? Branding might be one common metric for upper funnel media, measured by reach or frequency. However, there are a lot of ‘performance display’ and ‘performance video’ with the ultimate goal of driving sales. Yet ‘success’ is measured by VTR, or CTR or a once a year Media Mix Model. What if you could understand the exact impact of a specific creative with a specific celebrity? Then ‘success’ would be based on factors beyond the number of likes the promotion gets on Instagram (unless that is your only goal). Use an attribution model at a campaign and creative level to move from a world of Decision Driven Data, where you use data to back up a decision you have already made, to Data Driven Decisions, using the data on actual sales to drive creative and targeting decisions..

Greater cross functional collaboration: If each football player on a pitch focused on scoring a goal, then even the best team would lose every game. Instead, working together to find strategies that take the best of every player to score the greatest number of goals will drive the best outcome for all. Within your organisation, when you are not working in silos and are collaborating, we can share learnings and achieve greater outcomes. As an example, if you were to coordinate the video production team to show a video of Chris Hemsworth at the same time as you were sending emails about the launch of your new range of Avengers t-shirts then you will be able to drive greater conversion overall.

—

Now we know why attribution is important, a few considerations to make in building your attribution model, and some simple actions you can start taking with the insights.

Whilst this may sounds relatively straightforward, often there are People, Process and Platform barriers in implementation. Find out more in the final part of our three piece series.

Having got to the bottom of this article, I’m assuming you are interested in attribution, so if you wanted to find out how we can help, do reach out for a free consultation on your attribution challenges.