Sabrina Garufi is an agency development manager at Google, talking about how we can measure complex customer journeys.
Sabrina started by using the analogy of how we got to the conference. We all walked through the door, so should urban planners put all their resources into making pavements? No. Because people drove or came on public transport first. Last click attribution is like focusing all your resources on footpaths.
Measurement used to be a lot more simple. Customer journeys used to be simple and it was great. People would go straight to a website before converting there and then or visiting a store. Now the journey is much more fragmented than it used to be. For the marketer it’s a lot harder to identify where a customer is coming from and how they’re getting to you. It’s a huge web that you have to untangle.
Let’s say you sell beauty products. In the past you would have seen the last move a customer made before coming to your website. This is partly Google’s fault; for a while, last click attribution was the only model available. Last click, however, blinds you to a significant portion of the journey. Were you investing marketing spend in the right place? Who knows.
In 2017, we should take everything into account and work out which stages of the customer journey are the most important to us. We have to move beyond last click attribution to stay with the times.
The current free solutions to measure attribution are AdWords and Analytics.
Adwords attribution falls into rule-based and data-driven attribution models. Data-driven is new and AI-powered, why look beyond? But data-driven requires a certain volume of interactions before it can be of value to you (or even available).
Last click, first click, linear, position-based and time decay are the 5 rule-based models at Google. Linear attribution gives equal weighting to all stages of the funnel. Position based gives credit across the board, but more to the first and last. Time decay gives more value to the last click and gives decreasing value to earlier interactions.
Position-based is a good model for when you know the conversion is important, but you also want to know how they first found out about your products (e.g. if you launch a new product line). Time decay drives efficiency while bearing all touchpoints in mind. It gives most value to the final click, but it takes into account everything leading up to that.
But what is data-driven attribution? It calculates the incremental impact of each click. It’s a dynamic model that continues to take new clicks into account at all point. It gives credit to different stages based on the impact it has on final conversion rates. This model works out the impact that a stage has on conversion rate when it’s included as opposed to when it’s not, which leads to working out the value of that step (and that channel as a whole, once you gather enough data).
Data-driven models constantly change, whereas rule-based models have to be manually revisited and changed. But both types are based 100% on your data.
All of this applies to cross-device. Mobile appears to have a lot more value when you move away from last click. Users did the research on mobile that led to the final conversion on desktop.
Why are we talking about attribution now? Because data-driven attribution is already driving better performance for thousands of search advertisers. Moving beyond last-click alone delivers 5% better returns. Current studies show that cross-channel attribution was the number 1 concern for 2017. The key message is that if you’re not moving from last-click, you’re missing out.
Last-click insights show big spikes for Black Friday and Christmas. However, data from Google’s retails clients who were using data-driven attribution shows that more than two thirds of conversions on Black Friday came from queries that occurred in the days leading up to Black Friday. Google’s data shows that data-driven attribution drives 5% more conversions at the same cost per conversion.
Case studies show widespread increase in conversions and decrease in cost. These case studies used smart auto-bidding to increase ROI and efficiency of their marketing processes.
However, you can’t just move on from last-click and assume that things will magically improve. You have to be prepared for decimal points and you have to allow for 30 days for the data to come in and start making sense. Ultimately, the real impact comes from acting on the new data that you’re receiving through the new attribution models. You should be re-evaluating your budgets and the keywords that you have over-prioritsied before.
Auto-bidding is the key element to bring in to make sure you see an improvement. You’ll see an increase in the value given to generic terms. You need to be adjusting the budgets between campaigns. You should also be prepared to test new keywords.
Sabrina reiterates: move away from last click. It takes two minutes in Adwords and then you’re all set up.
Moving on to Google Attribution. The new tool will come out at the beginning of 2018 after a small Beta of about 400 advertisers. The clients that were on the beta were selected so that they gave as much data as possible. When Google Attribution was created they took feedback from these advertisers into account. It’s meant to be easy to use, actionable, cross-device and cross-channel.
Cross-channel is now available in Attribution and it will send the report to Adwords. Attribution will interact with all different touchpoints and channels. It’s made to be easy to set up: simply log in and connect your properties, then you’re good to go.
It’s cross-channel because it pulls in Analytics (which was the missing piece before). It will roll out in stages, so more features will become available over the course of the next year.
And it’s cross-device. The number of trackable logged-in users is a big strength of Google.
It’s actionable. You’ll be able to pull reports straight from Attribution into Adwords, make changes and set up auto-bidding.
