Background  

Primark is a surprising business. It’s hugely successful – and never advertised. Their business model is 100% bricks and mortar. Competitors with DTC offerings excelled (Shein, Matalan, Next). With ecommerce not on the roadmap for the foreseeable future Primark recognised recovery would be slow, if at all (net annual store closure 8.9x higher in 2021 vs 2015. Source: PwC).  

For the first time in their 50-year history Primark turned to media to get shoppers back into stores. Senior stakeholders were sceptical as to whether media could drive growth for a business as large as Primark. We had to prove that it would. And we had to do it in a way that would allow other markets to scale fast if our experiment was successful. It was daunting, we will not lie.  

Challenge  

Our biggest challenge was to prove that media could drive incremental growth and hit the revenue targets. Possessing limited customer data, and starting from scratch, we put learning first and treated everything as an experiment. We built a bold, intelligent year-long test and learn the programme and the most comprehensive measurement framework you’ve ever seen.  

Solution 

With 12 months to prove media works and hit growth targets, traditional Econometrics models couldn’t deliver actionable learnings, so we built a new model for our test and learn programme. 

Smart testing framework:

Based on their current rate of sale tests were 4 weeks on, 2 weeks off for analysis. Learnings from each test informed the next; Next Gen Econometrics at its best. 

Incremental revenue:

We used 3 years of daily sales data from 200 stores to power a regression model and create a robust sales baseline. The UK was split into 8 statistically comparable regions of 4 comparable cell pairs- one test, one learn. We used regression in our attribution methodology as it is comparable to MMM.  

Faster learning:

We set up a sales data API (daily volume, value data per store) and reported on ROI within 10 days. Without this collaboration, our learning would have been massively reduced.  

Modelling:

Data was run through statistical and predictive modelling to determine which factors were drivers of success. Insights fuelled further hypotheses to test. Together, a squad formed of Primark clients and the Starcom Data Sciences, Digital, Tech, Planning and Activation experts, ran tests on… pretty much everything. From the most effective format and frequency, most visited collections (Denim, by the way) and whether dynamic creative drove ROI (it did), collecting learnings across 150 data points along the way. These data points represented the relationship between media investment and incremental sales. 

Transformational Results: Global roll out 

Since July 2022 we: 

• Proved media drives incremental revenue in store. Our great experiment achieved more incremental revenue than the target. 
• Created a robust, scalable testing methodology: We collected activation insights on 150 data points and our optimisations increased average ROI by 39%, and our hypotheses were robust, proven by the fact that 83% of our tests were positive and rolled out. 
• Our great advertising experiment surpassed all expectations and unlocked further UK investment and investment in 5 other markets.