Coming into this, I was fully aware that the old-fashioned days of selecting a sample of 10 and then writing reports were just not going to work anymore, I knew we had to change our approach and was convinced Data Analytics would form a part of it.
I am lucky enough to have accomplished similar changes at previous organisations, therefore, I had a good idea that the starting point for me was to first understand the systems this organisation was using, the architecture in place, and the underlying data that me and my team had to work with. What I will say though is that the start of my journey did prove quite challenging, this was due to the fact we were in the middle of a pandemic which meant that getting face-to-face time with key stakeholders was somewhat of a rarity, so I remember having to be precise and to-the-point with my requests and approval meetings.
Thanks to my previous experience, I already had a good idea of the areas I wanted to target, such as Expenses and Purchase to Pay, but ultimately, I would say the approach I took was heavily influenced by the systems, the data we had at our disposal and whether or not that data was of reasonable enough quality to be able to gain any meaningful insights or results.
I think so. We started out by obtaining organisational Expenses data which we ultimately shared with yourselves (DataConsulting) and then over time we discussed and agreed on the type of tests we wanted to perform. I remember we looked at it from more of a fraud angle, which I think is often the case when just starting out, I suppose you could say we focused on typical expense fraud risks.
It was an interesting time to start due to the pandemic, it was not a typical type of year or period of time where you would expect to see regular expenses flowing throughout the system. Much of our activities involve the attendance and supporting of events and the delivery of first aid training – which all ceased during the lock-down period.
Because of that, you could say it was a strange time to start, but I thought it presented a good opportunity to look at and potentially identify anomalies in the data. Myself and my team had already created the usual graphs, bar charts and pie charts but for me, we needed to create a story. This for me is the important aspect, to make sure this type of work is successful with management you need to talk about the story behind the numbers and thankfully, because of the situation (the pandemic), we did have a story to tell. Coming out of the pandemic we started to see expenses picking up again which enabled us to really explore where the organisation was spending the most amount of money which enabled us to offer senior management the assurances they wanted around those fraud risks that we began to explore at the start of this journey.
You must remember that with the situation, everybody working from home, people in difficult situations themselves, the heightened risk of fraud was clearly there and thus a concern to senior management. So being able to provide assurance around that was very well received.
Where we had success was picking an area that management was interested in and one that was going to add value, then presenting the results in such a way that a story could be told to provide the assurances my management needed. I will also add, the solution was not developed over night, there were many iterations and constant changes to see if we could accommodate this test or that one, there were question marks over the data’s quality but also its integrity for example was it all there, was anything missing. I have worked in organisations before that utilised more structured data from systems such as SAP and Concur, but here it was very unstructured and so perhaps not quite at the level I had worked with before which was unexpected.
The key thing for me is to start small and aim for that low hanging fruit. Look for an area where you know you are going to get some sort of result, it could be Expenses, Purchase to Pay, Accounts Receivable, your CRM.
To start with, do not take on too much and stick to one area of the business where you know most about the systems and its data. Also, one that will be of interest to management, as they will be the ones who are going to be assigning budget for future work.
Think very carefully about your model, I have had a situation in the past where I have had an in-house resource that could do much of the technical work (Data Analysis), so it was essentially a key person dependency, but they moved on after a few years meaning I was back to square one. So again, I would urge you to think carefully, especially your smaller Internal Audit functions, about how you are going to approach this. Do you want to produce this in-house? Do you want to use your co-source? Do you want to enlist external expertise like DataConsulting to support you? I would suggest thinking about a proof of concept first, that is how we (us and DataConsulting) started out.
Keep the momentum going, you may not succeed the first time, or even the second, but just keep having a go, using different datasets until you get some positive results.
If it has been successful let people know, speak to anybody that will listen, not just your management or your finance team, whenever you are having meetings with perhaps your comms teams, project management teams, data insight teams… tell them you have started utilising Data Analytics and have had some success.