What we learnt at MeasureCamp 6

By Florian Giudicelli, posted 23rd March 2015 in Conference Review.

On March 13 & 14 2015, L3 Analytics attended MeasureCamp VI, the web analytics unconference. As with the previous events, it kicked-off with a series of training workshops the day before the unconference. More than 220 campers coming from various places in Europe gathered in London to share their knowledge during a full day.

MC Kick Off Session

The L3 team enjoyed learning new things from the Digital Analytics community so we would like to share with everyone, in return, a summary of our favourite tips from the weekend.

 

Google Analytics API

During the training workshop, Tahir Fayyaz (@tfayyaz) shared an awesome trick to enhance the Google Analytics alert feature available by default in the user interface.

MC API  SessionIt consists of using Google Sheets to extract your data via the API and push a notification via email if the value of the selected cell matches your desired rule. The great thing is you can use Zapier to send this alert to third party applications such as Trello.

With this approach, you can automatically create new tasks in your project management solution based on your GA data and be even more “agile” to improve your website performance!

Attribution

Four sessions on Saturday dealt with attribution. L3 Analytics attended two of them: Is last click dead and device attribution.

At L3, we have a pretty strong opinion about attribution since Peter explained why it is broken back in December 2012 at Emetrics. The key issue is that it doesn’t work across several devices as the logic is based on cookies. It is now clearly possible to link the same user on different devices thanks to a visitor ID and we had a great demo from Adobe about this. However, there are still several caveats due to not-logged-in users who cannot be identified.

MC Attribution SessionAnother takeaway is that the standard 30 day conversion window should be updated depending on your offer. The purchase process can be longer/shorter from one industry to the other.

For example, it should be at least 90 days for insurance and no more than a few days for a food delivery website. Following that, the average basket value within the same sector matters. The conversion window should be longer if you buy the last generation of TV than if you buy a cheap USB key.

Online/Offline Measurement

Google ran a session describing how to measure the ROI of your marketing campaigns within a multichannel environment. This topic is critical for retail websites as one of their primary goals is to drive customers to their physical stores. Hence, marketing departments require methods to estimate how much revenue is made in store thanks to online advertising.

MC Online/Offline measurement SessionThere is no perfect answer for this unfortunately but one option with GA is to use the measurement protocol in a smart way. Customers can be tracked within their multichannel journey through a common user ID. For online customers, the user ID is known when logged-in.

For offline customers, the user ID is known on purchase, when a loyalty card is used. Thus, it’s possible to link online/offline users and identify which online source drove a customer to buy in store.

As you can guess, the biggest caveat is that it’s impossible to identify all not-logged-in users. Google recommends being creative to encourage to people logging-in. The limit is that it could have undesired effects. A prospect might not wish to enter an email address just to look at a product description. This why customer feedback is still highly recommended to answer this type of business question – the easiest way of finding whether people came to your shop because of your website is simply to ask them!

Google Tag Manager

L3 is always learning new tips and tricks for Google Tag Manager (GTM) so we were excited by listening to Simo Ahava talking about Tag Management System (TMS) in the big room.MC Simo and GTM Session

The presentation was great as Simo demystified important facts that every advanced GTM user should be aware of. L3 especially enjoyed two reminders:

  • There is no magic connection between GA and GTM. A tag container is simply made of JavaScript code so it doesn’t matter whether you use the tag container from Google or from another company – you should be able to send the same information to Google Analytics.
  • Tag Management should always rhyme with IT and the reason is obvious: “GTM is all about IT as it’s a freaking JavaScript injector!”. JavaScript is a powerful language which can do many things including breaking a website if the code is wrong. Hence, marketing and IT must always work together and define the internal process to deploy new tags on the website.

R Programming Language

R is a powerful programming language which can be useful to treat a large set of data. A discussion was run by Guillaume Lombard and Richard Fergie to introduce this language to beginners.

The key benefit is that many libraries are available allowing the user to save time to export a complex database.

MC 'R' SessionIt is often used for data modelling, predictive analytics and to create some online dashboards. L3 enjoyed the example showing how to extract some data from the GA API as it shows how the tool can be used shortly after a brief learning curve.

BlackHat Analytics

Phil Pearce ran another fascinating discussion dealing with the malicious techniques faking your web analytics data. For instance, competitors could try to deteriorate your conversion rate to prevent you from taking the right business decisions. Hence, it matters to know what’s possible and how to avoid the most common issues.MC BlackHat Analytics Session

The key points to keep in mind regarding BlackHat Analytics are:

  • A simple and useful tip is to exclude JavaScript bots in your web analytics solution settings. In GA it’s only a box to tick but you could also configure additional filters.
  • A fake transaction with a huge amount of revenue could mess up your e-commerce metrics. The workaround is to declare the transaction as refunded.
  • Accidental PII in page names should be avoided as it violates Google’s TOS and your account might be deleted.
  • Flash cookies and Ever cookies can be malicious and are difficult to disable in your browsers.
  • DNT doesn’t stand for “Do Not Track” but “Do not Target”. To clarify, if this browser feature is enabled, the traffic is still captured in web analytics solution, it just means that users aren’t retargeted by online ads.
  • It is critical to add on your website a clear privacy policy in compliance with your web analytics solution and your different marketing partners.

Funnel Analysis

At the end of the day, Xavier Colomes shared some tips to analyse a website funnel. The method consists of using a strong page naming convention to identify the different stages more easily. In GA, a simple option is to create a view of your data with some filters to group pages by category (i.e. homepage, category page, product list page, product page, etc.).

A longer term solution is to specify the desired page name directly in the code. L3 Analytics completely agrees with this method as we believe that a strong page naming convention is critical to answer most business questions.MC Funnel Analysis Session


So we had a fantastic couple of days learning from our peers. What about you? Did you make it to MeasureCamp VI? And if you did, what were the highlights for you?