My job at Supermetrics is to help our users build the solutions they want to get when they buy our product. I’ve met hundreds of people with data analytics problems and I regret that I never started a count of how many projects I’ve built either myself or with my team. 

Those projects often have an interesting story in them. Here’s one that stuck in my memory:

A few years ago, a popular international brand got Supermetrics in order to build a centralised data platform for their marketing reporting. As a big international business, they had siloed regional marketing departments that operated under guidance from the HQ but otherwise were making their own decisions when it came to the details of running their marketing. 

They were setting their campaigns up themselves and were also running their own marketing reporting, sending results back to HQ in dashboards or spreadsheets.

You can imagine that this situation was causing friction as HQ had difficulties comparing the performance marketing data globally and data wasn’t flowing freely nor on time in some cases. 

So we scoped out a Data Warehouse project, chose Google BigQuery as the target data platform and Looker Studio (it was still called Data Studio back then) as the dashboarding tool.

The wonderful folks at the HQ were excited to finally get a single source of data for their global marketing operations and we were progressing nicely through the setup of connecting their global ad accounts through Supermetrics, scheduling data transfers into BigQuery and modelling their reporting data views and tables. 

But pretty soon we realised that things were going to be a bit difficult when it came to their campaign naming conventions. See, more often than not, my clients swear to high heavens during project scoping that their business does have a campaign naming convention and that everything can be easily extracted or mapped thanks to that. 

Equally often it turns out that even though they have defined a naming convention this doesn’t mean that everyone is really following it correctly.

It turned out that we had a similar problem in this project. The brand HQ had defined a global standard for marketing campaign names but regional teams were not reliably following it. 

My team and I worked with the client to patch as many outliers with custom rules as we could in order to extract information from the campaign names and create the custom dimensions that they wanted to see on their final dashboards but if you’ve ever done one of these exercises yourself, you’ll probably know that each exception to a rule that you have to write SQL code or create a custom rule for becomes a long term maintenance problem and most of the time catching 100% of the outliers is impossible anyway.

We completed the project anyway and handed over the marketing reporting dashboards to the client at brand HQ. In the final handover meeting they told me something unexpected:
They were extremely grateful for our work with their messed up naming conventions because through that they were able to finally identify which regions, marketing channels and individual campaigns were not following the guidelines and that made it possible to start fixing the problem.
This was more valuable for their business than the dashboards they initially saw as the main outcomes from this project.

Over the next months they reached out to all marketing teams that had quality issues with the campaign names. The match rate of our standardised mapping rules began increasing. The performance dashboards became better and better.

They’re still a happy Supermetrics customer today.

Did you ever experience something like this scenario? A data project bearing unexpected fruit? Share your story, I’m genuinely curious because these are some of my favourite anecdotes in this space