Do you want our logo?
Do you want our logo descriptionKnow our brand.
Do you want our logo?
Do you want our logo descriptionKnow our brand.
"Never make predictions, especially about the future" - Casey Stengel, baseball manager
In January 2015, we predicted the rising trends of in-site Digital Analytics. We were fresh from assisting some colleagues with their consulting clients about their data strategy. This is a checkpoint of how much we learned since then and how true the predictions were.
[caption id="attachment_1235" align="aligncenter" width="400"]
Accuracy vs. precision - The dartbox analogy. Source: Wikimedia[/caption]
Out of the original 5 predictions, 4 can be deemed true. We found that the trends described in 2015 are still strong right now, hence the title of this post.
The predictions were qualitative and bullish. The paradox of unquantified predictions is that, their being true, they cannot claim to be accurate nor precise. Using the dartboard analogy illustrated below, the top right hand target shows low accuracy and poor precision but true grouping.
It would be fair to say that this one was quite on the spot: while Google Analytics increased its usage by 41%, Omniture SiteCatalyst did saw its numbers fall by -32% both among the top 1 million properties on the World Wide Web. Some large companies who have the muscle (and vision) to fork out $150k USD for GA Premium are simultaneously planning to phase Omniture out. A few of these processes evolve into a Data Lake situation: they implement a Big Data approach consisting in grabbing all the raw on-site and off-site data on-site and actioning it with ad-hoc tools. This is something that our friends at Stratio are very familiar with.
Some observers like Blair Reeves claim that “Adobe’s Marketing Cloud SaaS business is hitting its stride” but the real contribution of Omniture to the revenue (and costs) over time is unknown.
It can be argued that Google is edging above former leaders SAS and Adobe in terms of the integration ecosystem. See the predictions below about Google Tag Manager and BigQuery for an illustration of this integration.
Google Analytics used pageviews as its core metric and it built its platform around it. Pageviews are becoming less relevant in the context of audience measurement (think about notifications on your smartphone, for instance: they are usually more valuable than visits to a web page). Events-based tools like Mixpanel or KISSmetrics are better at managing identities and calculating Customer Lifetime Value, CLV, for instance. This metric is more relevant for calculating the return of paid campaigns than most of the mainstream KPIs used in the advertising industry (CPM, etc).
[caption id="attachment_1241" align="aligncenter" width="512"]
Platforms with the ability to drill down into individual users. Adapted from Heap analytics[/caption]
Another contingency of page-based analytics is referral spam. Referral spam is the occurrence of fake visits from spam domains in the traffic reports. They typically show up with domains names such as “social-widget.xyz” or “share-buttons.xyz”. One of our clients with significant volumes of daily traffic saw their metrics for 2 consecutive days render useless due these pesky spammers.
The only way to fight this is curating a list of spammers to filter out on an ongoing basis but it is an annoying chore for skilled analysts. Googlers recently confirmed that they realize the spam referral in Google Analytics is an issue and they are working on solving the problem.
Score: The prediction was OK. I am however growing increasing concerned about the quality of the data on Google Analytics. The nature of direct traffic and the growing referral spam deteriorates the quality of the reports.
The adoption of Google Tag Manager in 2014 did accelerate through 2015. The penetration of GTM among the top 1 m websites by volume exploded from 20,000 in early 2015 to more than 70,000 nowadays.
Signal (former BrightTag) kept growing too but is being left behind by GTM. I found that Business Intelligence departments are the ones demanding these solutions while agencies follow and adapt to it. The advertising platforms that are not integrated in Google Tag Manager are working to do so. GTM is the de-facto standard.
Score: This is mostly correct yet hardly surprising: Google's strategy to commoditise and dominate the industries it breaks into (web search, PPC advertising, mail services and mobile operating system) worked like a clockwork.
Read BigQuery as Google's and other large scale data analytics like Amazon's Redshift. They all fulfil one of the premises of Cloud platforms: allowing the quick development of experiments and pilot projects. Their pricing models are flexible because they are roughly based on-demand usage with no little or no long-term commitment.
Score: OK, it was an easy guess being one of trends we heard about at the 4th edition of Big Data Spain. Take a look at our post comparing some of the public cloud services, in Spanish, for further information.
Behavioural and optimisation tools are doing a great job at making testing of web pages easy for virtually everyone. You do have to hold a degree in Statistics to carry out simple experiments with forms, navigation elements, etc.
Score: This was good. The irony is that while gathering, storing, analyzing and acting upon data is cheaper and more convenient than ever, the number of skilled people to make sense of it all is ever so scarce. You can train decent pros and empower them with experience but you simply cannot scale curiosity, intuition and stamina. The availability of great data analysts is the bottleneck of this industry.
I failed on this one: none of the onsite vendors I mentioned changed hands. The industry is booming and there is room of growth for every value proposition: visualisation of behaviour, split testing, etc.
Score: Fail! Are investors sitting patiently for their opportunity?
“I never think of the future - it comes soon enough” - Albert Einstein, theoretical physicist.
At Paradigma, our clients push us to ever higher levels of excellence also in Digital Analytics. Complex projets require to source and interpret data from diverse sources. This skill is the main reason why we got the trends right one year ago. Despite our decent record as oracles we will spare the readers from our predictions for this year. I will claim instead that analytics will be even more fun. I recommend Arjun Chopra's 7 Digital Analytics trends that will dominate in 2016. I find his considerations about the return of building an in-house analytics team timely and pertinent.
It is time for you to consider where are you standing with your data strategy in 2016. We help many clients to get more value from their digital properties and investments. Contact us to discuss how we can help you.
Comments are moderated and will only be visible if they add to the discussion in a constructive way. If you disagree with a point, please, be polite.
We are committed.
Technology, people and positive impact.