As the way people buy online (and offline) becomes more complex, it’s increasingly difficult for marketers to measure the true impact of every interaction. People see multiple ads from the same brand and various competitors before making a buying choice, so how do you know which ads really make a difference?
To help solve this problem, Google is urging advertisers to move over to data-driven attribution – something we’ve been doing for a long time now – which promises to show you the interactions that contribute most to the final conversion.
Here, we take a closer look at data-driven attribution and why you should care about it.
The quickest explanation is to say that the consumer journey is more complex than ever with people moving across different devices, channels and types of content as they interact with multiple brands – both online and offline – before making a purchase.
Data-driven attribution identifies each of these user actions and attributes value to them. This allows you to do two things:
When a user clicks your ad, scrolls through your landing page and buys from you on the spot, it’s easy to see the impact of each interaction and the effectiveness of your targeting and messaging.
Most consumer journeys are far more complex though.
Users are far more likely to click your ad, scroll down your landing page and leave before converting – but this doesn’t necessarily mean your ad campaign was ineffective. The same user’s next move could be to read some reviews about your product/service, check out a few of your competitors and call your support team with some questions before buying or getting a quote over the phone.
In this case, your ad still started the conversion process, your landing page copy convinced them that you’ve got what they’re looking for and your support team sealed the deal over the phone. If you fail to recognise the contribution each of these interactions make to the ultimate sale, you’ve got yourself a problem. Sooner or later, you’re going to pull the plug on that ad campaign, thinking it’s not getting results, and this is going to cut the source of leads your phone team has been turning into customers.
With data-driven attribution, you know how valuable each of your marketing messages are and how they contribute to the final conversion – even if they don’t convert users right away.
In May 2016, Google unveiled a new attribution model for Google Ads (then AdWords), which is suitably called data-driven attribution (DDA). Then, in May 2017, it announced a new product called Google Attribution – a platform that aims to provide a holistic overview of your advertising efforts by pulling in data from Google Ads and Google Analytics.
Google Attribution is still in beta but Google tells us it’s essentially a simplified version of Attribution 360 and the idea is to make data-driven attribution manageable for advertisers at every level.
Once again, it all comes down to assigning value to interactions at earlier stages of the consumer journey.
For maximum efficiency, a lot of advertisers use the last-click and time decay attribution models in Google Ads. Last-click assigns all value to the last interaction before the conversion and time decay gives more credit to interactions that happen closer to the conversion. Both of these attribution models will assign little or no value to the initial ad that starts the conversion process.
This makes it difficult to understand the true performance of your ads and you risk pausing campaigns that generate valuable leads. Likewise, it’s almost impossible to optimise earlier interactions for better performance when you can’t assign true value to them and measure results. While you’re also going to assign too much value to the final interactions at the end of the conversion process by using attribution models like time decay or last-click.
Google wants to fix this problem with its data-driven attribution model.
In Google Ads, you’ve got multiple attribution models to work with – each one coming with its own pros and cons. For example, linear modelling assigns value to every interaction but it doesn’t help you understand which ones are making the most impact. While position-based attribution places 40% of value on the first and last interactions with 20% shared out across each touch point in between.
With data-driven attribution, Google uses machine learning to attribute value to each interaction so you can see which ones are contributing most to the final conversion. So, while linear and position-based attribution might assign value to an interaction that doesn’t actually contribute anything, data-driven attribution will help you single out touch points that aren’t adding value.
Now you can pause campaigns with confidence, knowing that you’re not going to cut off an important source of leads.
In the example above, Campaign A might not convert any users at all and slip under the radar of last-click and other attribution models. However, data-driven attribution can determine that the presence of this campaign before Campaign B and Campaign C has a significant impact on conversions.
Essentially, Google crunches data from converting and non-converting traffic to determine how the presence of certain touch points affects the likelihood of conversions. It uses this data to create probability models and then assigns value to each individual paid click along the path to purchase.
Beyond measuring the performance of each interaction, data-driven attribution allows you to determine how your ad budget should be distributed across each touch point. For example, if you pinpoint which search and remarketing campaigns are making the most impact, you can up your bids and optimise your ad groups to boost performance where it matters most.
Google also has a number of automated bidding options to handle this kind of bid optimisation for you but this is one area where you probably want to have full control. We’ve built our own bidding automation models that gives us full control and transparency over how bids are managed and we recommend you do the same.
As for Google’s data-driven attribution model, the main downside is that you need at least 15,000 clicks on Google Search and a minimum of 600 conversions over a 30-day period. Unfortunately, this is going to be too much for many businesses who will have to stick with other models such as linear or first interaction.
Duncan was a PPC specialist at Vertical Leap.
Categories: Data & Analytics, Data Science
Let us review your search marketing campaigns and provide a proposal free of charge.