Case Study
How Tapestry helped BAM untangle marketing channel attribution
Recently, we shared how Tapestry helped clothing company BAM transform their marketing strategy from catalogue-centric to multi-channel, enabling them to drive 30% growth in sales. The eco-friendly brand – who sell clothing made from supersoft, sustainable bamboo – wanted to better understand consumer buying habits to optimise channel impact.
Since then, with digital channels becoming more prominent in the marketing mix, Tapestry and BAM have worked together to untangle the complex web of multi-channel attribution. As part of this strategy, Tapestry has enabled channel incrementality testing across different channels and geographical regions, as well as validating the accuracy of external performance metrics across social media advertising
Pinning down ROI across channels
Exploring regional engagement
While offline channel attribution is relatively straightforward to track (for instance, customers taking advantage of a promotional code in a catalogue), combining this with digital channels makes tracking ROI much more complex, since it is unclear how much each channel contributes to conversions. Tapestry helped BAM to isolate the effect of each channel so they could gain a true insight into how channels interact with each other to generate sales and ensure marketing spend is allocated accordingly.
The key to this was using ‘holdout groups” to test the incrementality of channels. Essentially, a campaign audience is divided into a test and control group: one that is exposed to the campaign over a period of time, and one that isn’t (the holdout). So long as other variables are kept the same, the difference in sales between groups will show the incremental lift provided by that channel.
Naturally, holdout testing takes time because only one variable can be changed per test, and it also sacrifices a certain number of sales per test. But the data this testing provides is invaluable for attribution modelling going forward, and more than makes up for the initial investment. BAM can now confidently make decisions on where to place marketing investment to maximise the return on investment in recruitment activity.
Using this strategy to gain more specific insights, BAM has taken holdout testing a step further to understand channel impact across geographical regions in the UK. By dividing the country into a handful of different regions and switching off Google Ads in each region for a period of time, the marketing team has been able to measure the decline in sales in each area (and therefore accurately measure the value that Google Ads contribute).
Recognising that this principle can be applied to isolate any variable, the marketing team can now use holdout testing to gain more granular insights across a range of metrics and channels. For instance, it is now being used to explore PPC (comparing branded and non-branded ads), and more recently, incremental testing on social media channels.
Fact-checking Facebook
By using holdout testing, BAM was able to post some content on Facebook only, with a call to action to sign up to their newsletter. They used a unique landing page so they knew that sign-ups from that page could only have come from Facebook users. However, when the team compared their results with metrics from Facebook Ads, they found a huge difference as the platform was dramatically over-reporting the conversions. Concluding that they couldn’t rely solely on these external metrics, BAM is now conducting further in-house incremental testing to accurately measure the impact of the social network.
This tactic could potentially be applied to any organic social traffic, so BAM is looking forward to using their attribution model to track the value of a range of audience segments over time.
With BAM’s spend shifting from 70% catalogue to 70% digital over the past few years, building a reliable model for multi-channel performance testing has become a top priority for the marketing team. With it, they can now truly untangle attribution.
