For a long time, analytics has held out the promise of helping businesses to gain true insights from their data. But the need for it has never been more pressing than now, at a time where many of our past assumptions are no longer true.
That’s because even as a normality of sorts is slowly returning, life after the pandemic is fundamentally different than before in many ways. Old data that predates Covid-19 will tell a business little about how customer habits or spending patterns have changed.
The need for speed
As a company that works with large anonymised data sets, CKDelta has seen how Covid-19 is driving a need for much more current and more recently refreshed data and analysis than ever before. Previous machine learning modelling and simulations have got to be completely re-learned, because a lot of the underlying home, work and social behaviour they were based on is not valid anymore.
A transport company, for example, would need to know about the number of journeys people are now making because despite some easing of travel restrictions, many businesses are asking their people to work from home where possible. Passenger data such as ticket sales or journey flows from 2019 would not reflect the situation we’re now in or allow modelling of possible future scenarios.
To pick another sector, retailers now need to completely reassess all of their plans in light of everything that has changed. Consumer behaviour is radically different than before Covid-19, the interaction between physical and digital retail has changed. Brands need to reassess their omnichannel strategies and watch the shift in buying patterns towards online. In the physical stores, there are new rules around sanitisation and signage which are subject to updating and changing. These factors affect just about every aspect of a retailer’s business, from marketing plans and service levels to store opening hours and staff resourcing.
Going to the source of the data
Traditionally, companies doing analytics have trained their sights on data within their own organisation, but it’s increasingly clear that the future of analytics will involve a more complex set of data from multiple sources. Often, the source of knowledge lies outside the company, or by combining various sets of data. For example, in a retail environment, changes in commuting patterns may affect the numbers and types of people visiting your store and their reason for visiting – there may be value in incorporating data sets about journey flows into the retail optimisation.
In a similar way, if I wanted to understand trends in the electric vehicle market to anticipate growth, I would need to combine data from historically separate sources such as urban planning and the electrical grid as well as data from the automotive industry.
In a situation that’s constantly changing, the data needs to be as timely as possible. To continue the retail example, footfall levels are changing hugely right now, as are customer demographics, so stores will want to know how those factors are reflected in till receipts in order to understand how best to maximise return on potential marketing campaigns or other changes.
And it’s live
The key with this kind of ‘live’ analytics is that it lets the retailer plan a marketing campaign, make some assumptions on what they expect to see, and then test if it happens. Depending on the early results, they can adjust the plan while it’s still underway, rather than waiting to the end to discover whether it’s working or not. This could drive decisions like changing where to advertise or which customer segment to target.
As a related aside, CKDelta takes inspiration from a team with diverse backgrounds from Formula One motorsport to FMCG, where our aim was to apply the analytics we were getting in real time and adjust the strategy as circumstances changed – while the race was still going on. Now, businesses can take the same approach.
In reality, however, this approach is very different to the old ways. Many businesses aren’t ready for it. If the old model was ‘plan and execute’; in the new environment, it’s more a case of ‘plan, execute, and adjust on the fly’.
Analytics needs an adaptive approach
It’s all too easy to get focused on the data, but as I alluded to earlier, the biggest challenge here is not the analytics: it’s operational design of a campaign so that the people running it can learn from early results and adapt mid-stream if necessary. This part of the project is a people rather than a technical challenge: to ensure the people and teams involved are flexible enough to be able to respond quickly if a change is needed. This approach calls for linking planning and operations more tightly than ever before.
Another reason why I emphasise the people aspect is because analytics is not about fully automating decision making. In our experience, most of the value in this new kind of analytics is in supporting people to make better decisions, not removing them from the process. It’s about helping people to see further ahead or have a clearer view of the best set of decisions.
I use the words ‘set of decisions’ deliberately, because too often data analytics is presented as if it gives a single right answer when that’s not the case. No model or simulation can do that: it gives a range of different outcomes to a degree or probability.
The economist John Maynard Keynes is supposed to have said: “when the facts change, I change my mind.” Whether or not those are his words, it’s hard to deny their meaning. We all rely on information to help us make good decisions. For businesses, that’s increasingly critical as they look to understand the new reality. We are now in a much more fast-moving environment and there are growing numbers of data sources to tap into, so it’s clear the best decisions often lie outside of people’s cognitive horizon. Data analytics can help to unlock that understanding and drive better decisions.
Jim Newton, director – business development, CKDELTA