We have been working on some very interesting – and challenging – data projects lately. Underlying many of these projects is our customer organisation's desire to utilise AI either in their sales and marketing operations, or in business development, reporting and analytics.

Mapping out AI opportunities is heavily dependent on data, so we always kick things off by getting to know our customers’ data systems as well as how they collect and use that data.

We have instigated many fascinating conversations, witnessed moments of grand discovery and cleared up a number of misconceptions during these projects. I'll share a few examples to drive home why it’s worth talking about your data needs and goals out loud, often, and with a wider team.

Intriguing findings from customer data projects

Example 1:

Working with an established SME, we discovered that they had as eight separate databases in daily use – with zero integrations. Critical customer information was scattered across the different systems and a lot of data was manually duplicated from one place to another. Few people knew which data was up-to-date and where to even look for it.

There was no master data system and dozens of additional and supporting systems had been purchased to fill this gap. We helped identify the need to upgrade specific data systems and prepared a phased development plan.

Data management maturity level: poor. AI readiness: non-existent.

Example 2:

Another organisation that was highly committed to improving customer experience, had customer data as a shared interest for many different teams. Data was actively collected but they needed help leveraging.

Data analytics was considered to be in great shape, but they were disappointed when their data views failed to spark new development ideas. At best the organisation had 1,200 different Power BI reports. 80% of these were deleted without anyone ever asking after them.

We saw that the organisation's main problem was that there was no structure for using data for business development. None of the teams had designed an operational model for generating improvement ideas based on data analysis. Together we designed an approach where the responsibility for interpreting data and making data-driven decisions is shared between different teams and integrated into day-to-day activities.

Data management maturity level: satisfactory. AI readiness: satisfactory. Maturity level of data-based decision-making: poor.

Example 3:

While mapping the AI capabilities of yet another of our customers, we brought sales and marketing together with development and IT to discuss their challenges and goals.

During the discussion, the sales manager said: "We don't bother giving quotes for all the leads because the tool is so slow and cumbersome..." This information came as a surprise to the head of development, who commented: "Well, we'll need to address that right away, don't we."

When the conversation turned to analytics, the marketing manager said her team uses three separate analytics systems but none of these show how their efforts impact the sales. The IT representative added that such a view actually exists, but it is in yet another analytics environment. 

Data management maturity level: satisfactory. AI readiness: moderate.

Examples 4 and 5:

A large food producer wanted to monitor its marketing performance using an intelligent dashboard that would utilise AI. The challenge was that the operator's advertising and media agencies owned all the campaign data. Results were only monitored through separate PDF reports. 

To enable reliable, transparent MMM and ROMI tracking, we first needed to gather all marketing performance data into our customer's own databases.

We encountered a similar challenge with a beverage industry corporation. The company spent millions of euros a year on marketing, but they didn't have a clear view to how this impacted their business.

We combined all necessary data sources and built a centralised data view in Power BI to enable sales funnel monitoring, total sales forecasting, more precise targeting, and performance improvements.

Data management maturity level: poor x 2.  AI readiness: poor x 2.

The link between data management and the ability to use AI

As the above examples show, even in organisations that understand the importance of collecting and storing data, data management practices may be missing completely. In such situations, the use of data is fragmented and sporadic at best, leaving no real opportunity or rationale to applying AI to enrich and analyse data.

AI capabilities need to be built on top of reliable data management practices. Often this also requires the development of a coherent data ecosystem. Sometimes integrating the data systems and centralising data into a single database means just a few easy fixes. It might not take even half a year to get a specific section of the data in order so that intelligent analytics and AI can be tested. However, if essential data systems need to be replaced, the timeline will stretch out. The key thing to remember is that your data ecosystem development will always be an ongoing project.

To recap, utilising data is more than playing around with AI. You will still need people to interpret analytics, to formulate observations and to make actual development decisions. AI can surface insights, sure, but it cannot independently decide how to develop your business or services. Not for a while yet. Not in most organisations in any case.

The structures and practices for managing and using data that enable data-based decision-making – whether AI-assisted or not – need to be integrated into the day-to-day operations of an organisation. Such practices can in theory evolve organically, I suppose, in a very proactive and flat organisation. However, it is faster to get on track by implementing a development project with the help of a trusted and experienced partner.

If you need help unravelling your data, we are ready to help. Contact us!