Note: This post has moved from Leapthree.com to Ayima as part of the 2018 acquisition.
Again and again over the past few years I have referred people to Eric Peterson’s blog post on “The Coming Bifurcation in Web Analytics Tools”, written back in Feb 2010. This blog post describes a potential future where the majority of people in companies use Google Analytics to answer day to day business questions while a very powerful data warehousing tool such as Adobe Insights is used by the analysts to answer the hard questions.
I think we need to go beyond that, that organisations actually need four separate digital analytics tools or user interfaces to meet business needs. This is required to allow team members of varying analytics skills and confidence levels to get the information they require. So, what are these four tools and how do they differ?
Note I use Google Analytics and Adobe Analytics as examples throughout this blog post but that is simply as they are the tools I am most used to.
Standard Reporting
The first tool is quite simply a reporting layer. It is intended for use by non-analysts (you know, like the vast majority of people in any organisation) and eliminates the need to log into intimidating digital analytics interfaces (yes that includes Google Analytics). While reporting is not the most respected aspect of analytics, I believe it to be essential. Non-analysts need to understand how the business is performing and they need to easily access the information they need to do their jobs.
It should include both performance and diagnostic reports. They can be daily, weekly, monthly or ad hoc time periods. The reports must be automated and easy to read. They can be delivered via excel, tableau, a self-built online interface or other online dashboarding tools. It is not impossible for all this to be provided via the Digital Analytics tool but I do not personally believe these are flexible enough – including Google Analytics & Adobe Analytics.
So, the first tool is a set of reports available to everyone in the organisation which provides them with all of the information they need to look at and take actions on provided on a regular basis. Management understands performance. Workers have the data to take actions without asking the analysts for help. Good outcomes.
Simple Analysis
The second tool is a simple analysis tool, again intended for use by non-analysts. These will only be a subset of the reporting tool users, those that are comfortable with numbers and exploring data themselves. The key requirement is the tool has to allow simple analysis (apply key segments, change date range, trend data, apply a breakdown) but to stop at that, it needs to come with a limited range of features and a limited set of reports. By reducing the options, more people will be confident to use the tool.
This tool could be Google Analytics, probably with only Dashboards & Shortcuts visible plus with features like secondary dimensions and the ability to create your own segments/custom reports removed. Or Adobe Analytics could work with a customised menu structure to provide only core reports. It is nearly possible with Excel or any data visualisation tool. Again, don’t really care on the delivery method, as long as it gets more people using data to take actions.
Ad hoc Analysis
The third tool is then for ad hoc analysis by digital analysts. It needs to have a full range of reports and analysis capability so users can really drill into the data. Think Google Analytics without sampling and whatever features they decide to add over the next 6 months. Or SiteCatalyst Discover Adobe Analytics Ad-hoc. Both tools where, if you know what you are doing, you can really explore the data. Not to answer everything but enough that you can answer most business questions that are thrown at you.
Deep Dive Analysis
Which leaves deep dive data analysis, the final digital analytics tool required by an organisation. It is for analysis that will take days or weeks to answer a single question. The big questions/projects. I will even allow users to call themselves data scientists – as long as they return insights that trigger actions that improve performance – not a stack of really interesting information/statistically valid data with no business application.
The tool could be Adobe Insights or the Google Analytics Premium integration with BigQuery. Other options might include Snowplow Intelligence, iJento or a self-built tools. Basically any data warehouse tool that allows full slice and dice capability with the raw data.
Update: I forgot to mention, I would include tools like R, SPSS, etc within this method of accessing/manipulating the data.
Final Thoughts
Ideally all four tools would use the same data source/s. This is not essential – I believe consistency of usefulness of data is more important than consistency of exactness - but it is good to have a single source of the truth. Update: If using a single data source, these four tools can then be described as four methods of accessing this data.
But there you have it. In my opinion, the usefulness of these tools actually follows the above order. Yes you need a tool for analysts but if they are the only people in the organisation who can use the tool, it is data that is unlikely to be used. And while “big data” may be the buzz word of the moment, there are so many business changes that can be made using “small data” with more impact and at a lower cost.
All of these tools are just about possible with Google Analytics, Adobe Analytics or likely any Web analytics tool. Except for that first tool, the reporting layer. It is generally the least regarded but, as able to be used by everyone in the company, capable of being the most powerful for exposing data to drive change.
Update: I forgot to mention what I think this means. Nearly every organisation has an ad hoc analysis tool. Some have invested or are investing in the deep dive analysis tools. I think the priority should be to invest in the reporting layer first, get that right. Then ensure there is a simple analysis option with education provided so users can access the information they need. Only then should the deep dive analysis tools be considered.
