Digital Marketing Analytics – 5 Key Takeaways from the 2014 Melbourne Google Analytics User Conference

Yoke’s marketing team got the wonderful opportunity to attend this year’s Loves Data Google Analytics user conference at the Melbourne Arts Centre.

2014 Google Analytics User conference Melbourne

The annual conference has always hosted the world’s, and Australia’s, top experts in digital marketing analytics and this year was no different. The 2014 edition of the event sported several digital marketing analytics heroes, among them, non other than Google’s own Digital Marketing Evangelist Avinash Kaushik and Analytics Evangelist Justin Cutroni.

The intensive eight-hour day provided a bucket load of useful insights and actionable tips for both agencies and in-house marketing teams. Although we wouldn’t be able summarise the entire event in a single blog post, our recap of the five most important and talked-about concepts in the conference will hopefully give you an understanding of where to concentrate your digital marketing analytics efforts in 2014/2015.

1. Start using Google Tag Manager now

Definitely the most mentioned term throughout the conference was Google Tag Manager, which is Google’s answer to easier analytics implementation for marketers.

By tackling the everlasting challenge of combining dry analytics data with real business objectives, Tag Manager is a game changer; a marketer can simply define the ideal action a visitor would take on a website (such as subscribe to a newsletter) and define this as a rule in tag manager. After this, each time visitors subscribe, Tag Manager records this as a goal completion in Google Analytics.

The beauty of Google Tag Manager is that it lets marketers manage their analytics tags more efficiently without the need to bother the webmasters every time a Google Analytics tracking code needs amendments. This naturally speeds up the process and also prevents instances where tracking codes may not function properly or slow up the site.

So often was this Google product mentioned that conference attendees were, in fact, encouraged to take part in a GTM drinking game and have a drink (i.e. sip of their coffee) every time GTM was mentioned. We swear that by the time the clock struck 5 PM, the room was full of jittery analytics geeks high on caffeine.

2. Employ the Digital Marketing & Measurement Model to improve your digital marketing efforts

Just in case we didn’t emphasise this enough, we’re pretty big fans of Avinash Kaushik. With a strong business perspective that influences both his talks and his posts on analytics, you always get a multitude of concrete, actionable tips for your business.

The points he made during the conference were all delivered with his slightly unconventional and entertaining mode of self-expression, which included, among other things, blocks of chocolate, a higher than average use of expletives and frequent shouting. This method worked exceptionally well.

He put particular emphasis on how important it is to deploy a detailed measurement model when determining the success of marketing campaigns, such as the example below demonstrates.

Avinash Kaushik's Digital Marketing Measurement Model
Avinash Kaushik’s Digital Marketing Measurement Model.

One of the key points of Avinash’s talk, and what is also evident from the measurement model, was that conversion rate is but one metric of success, and it shouldn’t be used as the sole determinant of a campaign’s success or failure.

In fact, depending on how a company defines success, there can be any number of various success metrics, such as a website visitor, adding a product to a wish list, writing a review or subscribing to the company newsletter.

Here at Yoke, we will definitely look into employing this model in the near future. For a great tutorial on how to construct your own digital marketing measurement model, check out Avinash’s step-by-step guide.

3. There is no perfect attribution model (but some are better than others)

Attribution models were discussed on multiple occasions by the conference speakers. The question of how much value to give to each customer touch point in their conversion journey has been a problematic one, and to-date, there are no definite answers as to which model works the best.

In an ideal world, there would be only one touch point before a customer makes a purchase, but usually it’s not that simple. Customer shopping and search behaviour is a lot more complex and a single transaction can involve different devices, time lags, channels, IP addresses and online/offline interactions. This presents a real challenge for data attribution; which interaction pushed the customer to make that final purchase decision – was it the first click, the last click or a click in-between?

Currently, Google Analytics gives you seven different predefined attribution models to choose from, plus an additional option to create your own model. Avinash expressed his distaste towards the Google Analytics default model of last-click attribution due to its inability to paint a realistic picture of the conversion process and the roles the various channels play in it.

Instead, he recommended the linear and time decay models that give analysts a better understanding of which channels work well in converting visitors. A further tip was provided by one of the Google Partners panellists, as she pointed out that comparing the first-click model and last-click model could serve as a great top-level marketing sanity check.

4. Place less emphasis on benchmarking

Benchmarking involves comparing your company’s performance to that of your competitors’. It was all the rage back in the day until it went away, and now it’s making a comeback.

While benchmarking definitely gives you a nice overview of the current competitive landscape, a few of the Google Partner panellists thought of it mainly as a tool to get the management’s attention. Some argued that benchmarking encourages a wrong business strategy, as simply keeping up with competitors doesn’t mean a company has a competitive advantage.

Furthermore, when it came to benchmarking conversion data specifically, Avinash was strongly against the idea due to no companies being exactly the same, hence rendering any comparisons between conversions obsolete. This means that while you think you are comparing apples to apples, you might, in fact, be comparing apples to oranges.

5. Data puking – it’s not ok

Finally, what many marketers and analysts sometimes forget is that top-level management is not necessarily as fascinated by hard core data as marketers and analysts are. Marketers and analysts can build the prettiest presentations of rainbow-coloured charts and infographic-style layouts, but at the end of the day, the only one deriving any value from these works of art will be the marketers and analysts, not the top level managers.

Many speakers made strong cases against committing a so-called ‘data puke’ on the higher-level management and presenting only the major insights instead. Avinash went as far as advising the conference guests to use only words in English and to omit the numbers. Michael David from Internetrix also pointed out that delivering more insights than questions is the most important part of data reports.

Avinash presented a few beautiful examples of data visualisations that may be of value in reporting complex data. Needless to say these made the majority of guests salivate.

Sunburst diagram by Avinash Kaushik
Sunburst diagram by Avinash Kaushik.


Chord diagram by Mike Bostock
Chord diagram by Mike Bostock.

Obviously the day provided us with much more information than the points presented above, but in terms of wider business and marketing analytics strategies, these were definitely the five key takeaways from the conference.

In short, in 2014/2015 digital marketing is clearly more data-driven than ever and the subsequent analysis and reporting have become more sophisticated and complex.

What marketers and analysts now need to concentrate on is perfecting the entire chain of analysis; determining what and how to measure, constructing a valid measurement model, drawing the important insights from the analysis and then communicating the findings more effectively to the top level.