Garbage in means garbage out – it’s an old but relevant adagePublished: Thursday, 30 November 2017 by Matthew Tod, Head of Analytics
The study we recently commissioned from Forrester Consulting into how companies are collecting the data they need to power differentiated customer experiences contained a table that really caught my eye – it was about the technology used to manage customer interaction data.
Why did it catch my eye? Well the top pick for the technology most used to manage customer interaction data and track customers was web analytics….but web analytics tools don’t really understand customers.
Web analytics tools can provide aggregated information about the way visitors to your digital property have interacted, but identifiable individuals are not their forte. Can you ask a web analytics tool how many female customers came to your website yesterday? Not really unless you are a total Adobe ninja with incredible data wrangling and tagging skills
So, what is wrong with basing your customer data strategy on a web analytics tool? Well here goes…
1. Leading web analytics tools don’t create robust multi-channel customer profiles. They will report on sessions, page views, campaigns and transactions but they won’t tell you that Alice is a lunchtime browser, an evening buyer and has an increased frequency of visit lately. Actionable customer profiles, as opposed to reports of your digital audience en-masse, are at the heart of a good data strategy, and web analytics tools don’t deliver.
2. Identity resolution is also not a core capability of the web analytics tools, and if your goal is to create the data needed to power exceptional customer experiences, you really do need to build a rock-solid profile of each individual. That means, when you are sure, consolidating data from different silos and looking back into previously collected data to build that profile.
3. Typically, web analytics tools are not used in channels such as ATMs, kiosks or credit card terminals and these systems also generate valuable data. Deploying a tag-based mechanism across disparate channels is very difficult to deliver cost effectively, but unless you collect data from all channels you run the risk of becoming incoherent.
I could go on for hours about the lessons I learned over the past decade working with web analytics tools, but I am sure you get the point: re-purposing data from one system to 'make do' won’t get you very far.
I had a lovely example of this recently where an AI powered recommendation project was failing, and the reason was the data. The system was using web analytics data that showed a recommended product had been rejected, and this was being fed back into the model to improve it. However, the web analytics data recorded a 'page view' not a 'page seen for more than 5 seconds by the targeted customer'. In reality, many recommendations had not been seen by customers, but the analytics tool had no way of determining this. As a result the AI was being force-fed false negatives because it assumed that customers who had not seen the offer had rejected it! Poor data will flummox the smartest AI.
I chose the headline for this post as 'Garbage in means garbage out – it’s an old but relevant adage' because it's a lesson I see many organizations relearning in the Age of the Customer. If you do not feed the clever AI-powered recommendation and personalization engines great data you get rubbish results, and if you chose the wrong data technologies you are doomed to this fate!
There is however a solution, that 33% of our respondents have discovered: use a Customer Data Platform to create the data you need to power your customer initiatives. You can learn more about our Customer Data Platform here, as well as download the Forrester study from Celebrus/Resources/Research & Insight, here.
LinkedIn: Matthew Tod