On November 9, the first edition of Untagged, an international Web Analytics event, took place at the Google Campus in Madrid, which we attended as sponsors. It was an event held entirely in English and at a high level, both due to the profile of the speakers and the audience. Since it is not easy to summarize more than six hours of talks in a single post, here we collect the main ideas of each speaker's presentation, as well as the links to the talks that have been made public by the authors. Index of contents 1. Brian Clifton: “The Current State of Digital Analytics” Brian gave a brief tour of web analytics from its beginnings to today, and his talk mainly revolved around the importance of data quality. He gave us an essential question that must be asked before performing any analysis: Can you trust the data provided by your web analytics tool? It's a basic question, especially if you invest more than $100,000 in a premium web analytics tool. And it is that, according to Brian's research.
Only in the basic implementation of the tool half of the websites do it wrong, with pages where the measurement code is missing or duplicated. Other data that he offered us: only 21% correctly track form submissions, 26% measure transactions well, 25% measure campaigns correctly and only 4% segment their users correctly… To know if correct conclusions can be drawn from the data, it is essential to audit the quality of the data before carrying out any other process number list and it is not an easy process. An audit can take more than three working days. In any case, the recommendation he gave us is to never create a dashboard until the data quality index exceeds 50%. Twitter: @BrianClifton 2. Kristoffer Ewald: “Innovations in Digital: New Models and Metrics” In an environment where data is everywhere, Kristoffer made a reference to the “Cluetrain Manifesto”, which begins by stating that markets are conversations and those conversations create dialogues. Concepts such as the six degrees of separation are no longer valid, for the simple fact that with social networks we are now much more closely related to each other than before. We have gone from a world where information was centralized with a single node.
To a world where it is totally decentralized, where popularity is not influence and where the web is not the center but just another channel. Currently, in the digital world we find ourselves with challenges such as the following: We know what we have to segment, but we don't know what the criteria are: among a sample of users, we can segment them by innumerable criteria: those who wear hats, women, those who wear green sweaters, those who wear blue pants... which one? is the correct segmentation we need? Even today, many tools still attribute the conversion to the last click, when we are in a world characterized by multi-channel and overlapping devices: the challenge is to find the right model. In standard analytics, we usually lack a fundamental metric: impressions… how can we then measure conversion correctly? The expiration time of a cookie is ridiculous, since many of the conversions happen beyond 90 days. It can even take years. If we buy a car, it makes very little sense that they do remarketing to us a few days after buying it, but it does make sense that they do it years later when it is presumable that we are considering changing the car again.