Monday, November 17, 2014

Can End-Users Really Use Data Analysis Tools Without Data Experts?

There are some incredibly good end-user oriented data analysis software systems these days. Really user-friendly, lots of great graphical drag-and-drop functionality and one can get all sorts of analysis from them. Even the new CRM systems provide decent user-oriented tools which can be learned by non-technical staff in a comparatively short amount of time.

But herein lies the rub: if we make it so easy for end-users to be able to use what is actually quite sophisticated software, then how do we know they are using it correctly? Or more to the point: how do we know they are getting the correct data, getting the correct data which they really want, and interpreting the correct data correctly?

Let's take the simplest of examples: a user wants to know how many supporters they can mail for their next event. This means knowing how the database and data defines such supporters, knowing how to exclude deceased/gone-aways etc, knowing other opt-in and opt-out codes, knowing the opt-in and opt-out policies, knowing the meaning of the values within a specific field, knowing if other groups of users might want to approach the same supporters at the same time. And knowing exactly where and how all this is held in the database.

Maybe it isn't that easy after all… Add on any further level of complexity such as last raffle gift (not just any gift), average gift over the last 3 years, if they came or were invited to the same event last year, and you're starting to tax even the data experts.

Which is why so many charities have centralised, specialist data teams who do such work. Or at the very least, highly trained users in each fundraising team. With this approach, we can be sure that the specialists know not only the data and all the above conundrums but they also know what additional questions to ask the fundraiser who wants to know such answers.

All of which pains me a great deal to have to write. I want end-users to be able to run such queries themselves, I want them to be able to be empowered to ask 'what-if' type questions so they can see data patterns, I want them to be more self-sufficient so that it will help them and take some of the strain off the database team.

But to get such analysis software to a point where this is possible, to ensure the data is so easy to understand that anyone and everyone can do, to ensure that internal policies are fully understood - all that is not necessarily easy.

Which doesn't mean we shouldn't try to do it and it doesn't mean it can't be done. We just have to ensure that appropriate software and data and policy training is given to those who need to know - and probably to limit the sort of reasons for why they are doing it in the first place. i.e. As long as the users are doing such queries just to get rough counts, to get some idea of data, to be able to do a first data sweep of a data mining exercise, then I think that should be possible. But sadly I don't think even the best software in the world yet means that a end-user can do their own segmentation and create their own mailing files for their next big event. We still need the data specialists for that.

2 comments:

shaun williams said...

Hi Ivan

This is a subject area really close to my heart. At PDSA we have been a big user of Apteco Faststats discoverer for many years (and a previous winner of the best use of Faststats award). In the early days a wide range of marketers and fundraisers had access to our Faststats system and many used the software as you describe – for train of thought analysis, understanding core campaign volumes and getting a feel for data structures and supporter relationships. However, many marketers also used the tool for their own data selections which over time became more and more of a problem for accuracy, consistency, and delivery of a coordinated data strategy.

So we are starting to go the other way in that we are reducing marketer’s direct access to these sorts of tools (but expanding the Data & Insight team to compensate). Data is just becoming too complex for many none technical specialists. You identify some of the issues in your article, but when you layer on multi-channel data from the web, email or the call centre and add in complex segmentation and propensity modelling and you really need people with the right technical and analytical skills to interpret the data and deliver insight – allowing marketers more time to make better decisions.

I believe the answer is to develop a broader partnership between the marketing teams and data, analytics & insight. At PDSA we are starting that journey by bringing together analysts, data planners, and researchers as one team to support a newly formed CRM team.

Shaun Williams
Head of Data & Insight
PDSA

Ivan Wainewright said...

Thanks for your time and feedback, Shaun. Very interesting to read your final paragraph about developing the partnerships - I'm sure that's a good idea. And equally interesting to see your note that you have a newly formed CRM team - I suspect that is something we may see increasingly in charities with more and more data pressures.