# Case Study: The Account Where 'At Best a Guess' Was the Verdict on Attribution | Crank Source: https://wearecrank.com/case-studies/mens-grooming-attribution-2017 Missing event data, inconsistent tagging, and a trading decision that had been mistaken for a tracking fault. --- ![Men's grooming case study: The Account Where 'At Best a Guess' Was the Verdict on Attribution](/images/case-studies/mens-grooming.png?dpl=dpl_2uJjggmfov9ibhArygujfRBb3pax) Men's Grooming, 2017 Audit # The Account Where 'At Best a Guess' Was the Verdict on Attribution Missing event data, inconsistent tagging, and a trading decision that had been mistaken for a tracking fault. Challenge The audit's own verdict on attribution was blunt: at best a guess. Key event data didn't exist before August, tagging was inconsistent, and server-side measurement was inflating direct traffic while suppressing its bounce rate. Spend allocation was, in the audit's own words, at best a guess, and rising order value paired with falling conversion was heading for diagnosis as a broken tag and an expensive 'fix', when it was in fact a trading decision working exactly as designed. Approach Confine the analysis to the periods the data could actually support, product category analysis, for instance, only existed for three months of the year, and check whether apparent performance shifts were tracking artefacts or real trading decisions before recommending any fix, including corrected tagging, corrected server-side measurement, and a route for trading changes to reach marketing before they surface as a performance problem. Outcome Correctly identifying the delivery-cost pattern as a trading decision rather than a broken tag was itself the immediate value of the audit, it stopped a working system being expensively 'fixed' for a fault that didn't exist. The audit also set out the specific route back to reliable attribution, for once its measurement recommendations are followed through. 01 ## An audit that called its own attribution a guess Key event data did not exist before August, tagging was inconsistent across the rest of the year, and server-side measurement hits were inflating direct traffic while suppressing its bounce rate. Spend allocation decisions made on this data were, in the audit's own words, at best a guess. 02 ## Three months of data standing in for a year Product category information was present in the data for April to June only. Any analysis of what actually sold, by season, by device, had to be confined to that window, or it wasn't analysis at all. 03 ## A trading decision mistaken for a tracking change Average delivery cost rose in step with average order value from mid-year, suggesting a minimum order threshold had been introduced without anyone telling marketing. Order value rose; conversion fell, a genuine business decision, not a broken tag, and one that would have been diagnosed as a measurement fault without checking. 04 ## What the audit changed, and what it recommended Correctly identifying the delivery-cost pattern as a trading decision rather than a broken tag was itself the immediate value of the audit, it stopped a working system being expensively 'fixed' for a fault that didn't exist. The audit also set out the route back to reliable attribution: corrected tagging, corrected server-side measurement, and a channel for trading decisions to reach marketing before they surface as a performance problem. 05 ## What this means for you Before you fix a performance drop, check whether anything in the business changed. A confident, well-evidenced and entirely wrong diagnosis is the usual result of not asking. ## What a year of unreliable attribution looked like, and the fix it avoided 0 Months of reliable event data before August 3 of 12 Months with usable product category data “At best a guess” The audit's own verdict on attribution 1 Undocumented trading decision, correctly identified before an expensive 'fix' was made Drawn from Crank's account audit conducted in 2017 for this men's grooming brand, before the account was taken on. Client not named. ## Want to know what this looks like in your own account? These findings came from asking straightforward questions before any budget moved. We're happy to do the same for yours. [ Talk to Wearecrank](https://wa.me/447457416922?text=Help%20my%20business)[Get in touch](/contact) [← Back to all case studies](/case-studies)