Instead of Customer Insight as a topic, on which I debriefed a year ago, this time the topic for my table was “Real-time Analytics”.
It was encouraging to again see a good attendance from Data & Insight leaders, across many sectors. That was particularly impressive this year, as this event took place on 28 Feb. This was when ‘The Beast from The East‘ was really beginning to freeze the UK.
So, special praise to all who battled the snow & ice to talk about analytics. It was also great to see such a buzz amongst attendees & recognise many familiar faces. Data & analytics leaders, working in the UK, really are beginning to feel like an active community.
DataIQ Discussion Events Structure
Under the watchful eye of ever witty chairperson, David Reed. The structure of these annual events is simple but effective.
Delegates have opportunity to elect 3 tables they wish to attend (out of 10-15 possible topics). During the morning, they attend 2 of these tables, with a coffee break between, then another table after lunch.
At each table, over the course of 75 mins, a chairperson facilitates discussion and a relevant vendor is there to answer questions.
As with my experience of discussions at breakfast briefings hosted by MyCustomer, it is good to see more of this type of event. Enlightened vendors are beginning to see the value of content & question led meetings. Without any ‘hard sell’ or even pitches, there is great benefit in creating space for leaders to think about relevant problems.
In the case of this event, one of chief benefits of discussing real-time analytics was to dispel myths & confusion. Like the over used analogy with teenage sex, real-time analytics is still relatively rare in practice.
But, some useful lessons emerged, from the collective wisdom of the table. So here are a few lessons I learned…
Real-Time Analytics lacks a consistent definition
As guests shared their questions & experience, around the table, the first reason for confusion became clear. Different leaders & organisations were defining ‘real-time analytics’ in totally different ways.
Some appeared to be focussed on monitoring or real-time BI alerts. Others were considering, the use of predictive models, during real-time interaction with customers. But digging a bit deeper, it appeared in all cases that these were actually pre-scored models.
So, as is often the case, real world applications were far less sophisticated than the focus of many vendors & even ‘Gartner like’ researchers. This does not matter per se, as I’ve shared before, you often get more value from less complex solutions.
Where it did matter, during our conversation, was the confusion caused by lack of a consistent definition. When using the same term, different leaders meant & expected different things.
So, to aid our conversation, we reached a working definition. This is based on the idea of range of capability, simplified into 3 groups or ‘buckets’. Listed from the least complex to the most sophisticated, these are:
- Monitoring trends and alerts for BI reporting, operating in real-time. Examples shared included social media monitoring and reporting for call centres. Both focussed on spotting ASAP an issue requiring intervention or proactive communication.
- Use of pre-scored predictive models, during real-time interactions with customers. Examples shared included both prompts for advisers in call centres & dynamic pricing on websites. Both used available models scores (often propensity models), as relevant, to guide personalisation.
- Real-time scoring of individuals, using existing models but newly available data. I have seen this done, using new data items captured during conversation, but not previously available. However, none of my table guests had reached this level of sophistication.
Real-Time Analytics needs to focus on current needs
It was instructive, for me & vendors, to hear that none of my table guests had even reached level 3 of sophistication. In fact, many still had to overcome significant barriers to move beyond level 1.
I say that, because it seems that vendors are in danger of once again being carried away with industry hype & grand plans. Social media & websites will reveal their focus appears to be more on moving beyond even level 3.
To some extent this is understandable. I have been aware of solutions, for even level 3 sophistication, for at least 15 years. However, as suppliers seeking to help businesses generate more value, they should stay focussed on what is needed now. Rather than trying to bewitch leaders with ever more impressive demos.
A number of barriers to more sophisticated real-time analytics use were shared at my table. Here are a few of the most common:
- Single Customer View: It may be unfashionable, but having a single record (even virtual) of all variables for a person is needed for many models.
- Data Quality: We’ve raised before your data quality responsibilities under GDPR. But, it is also critical to model building & deployment. To many organisations are still limited by ‘Garbage In Garbage Out‘. Invest in data cleaning & at least, the metrics to reveal your real data quality today.
- GDPR: Beyond just Data Quality concerns, a number of leaders were concerned about consent & transparency. Some of the applications of real-time analytics previously planned would now be building on sand given wider GDPR issues. Focussing on compliance, also meant simplification, for many less is now more.
- No ‘killer app’ or use case: For some leaders, real time analytics (certainly beyond level 2) still feels like a solution looking for a problem. Perhaps because this area has been too vendor-led, they did not see a compelling use case. For this reason, they struggle to create a robust business case for investment.
With regards to that fourth barrier, they may be right. What I mean is, they may not have a need more real-time analytics.
That might sound like tech-heresy, but it brings me to the third theme that emerged at our table.
Not everyone needs Real-Time Analytics
Amongst each of the 3 table discussions, that I facilitated, we reached this point in the conversation. Someone raised the question, did they even need real-time analytics.
Now, before that just sounds Luddite, consider this. This was a conversation amongst capable data analytics leaders. None was tech-phobic & many had innovated use of technology, in many other ways, in their businesses.
Rather, this realisation came as we each considered how such a capability might help our customers or improve our business. Focussing on the desired outcome, or identified need, rather than the technology, helped clarity.
One of the reason, that almost half of table guests decided they might not need real-time analytics, was alternatives.
Practical alternatives that made more sense for a number of businesses and their customers are:
- Near-time analytics. Use of overnight, or intra-day, scoring and next best actions was sufficient for many businesses. The technology costs of moving to true real-time were not justified.
- Greater empowerment of people. For a few organisations, more improvements could be made, to customer experiences, by educating and empowering frontline staff. Greater simplification, or visual presentation of information, to inform their decisions, was more useful.
- Staying at current level of sophistication. For others, their current applications of either level 1 or 2 ‘real-time analytics’ was sufficient for their business need. That ‘three bucket‘ spectrum should not be seen as an aspirational journey for all. For many businesses, it seems, level 1 or 2 may be where there is a compelling cost/benefit.
Finally, I should mention the concern about ‘creepiness‘. A number of leaders also cited their experience as consumers. Quite a few has experienced real-time personalisation of content, based on recent web browsing or purchase & found it creepy.
It is always important for insight & analytics leaders to step back & take a human perspective. How will your intended application feel for your customers. Is it appropriate or too much of an Orwellian vision of surveillance.
Real-Time Analytics – what would you like to say
I hope that debrief was useful. I am grateful to all those who participated in our discussions and shared their experiences.
Now, over to you our readers. What do you think about real-time analytics?
Do you have any experience, implementing real-time analytics solutions, that you’d be willing to share? Are there any other questions you would have raised if you were there?
Hope to see you at next year’s DataIQ Discussion – just remember to pick my table 😉