On Wednesday last week I visited Big Data London at London Olympia, and I wanted to take a few minutes to talk about some of the important bits that I took away from it.

I ended up getting to five of the talks throughout the day, with some really great speakers and a lot of great insight shared. I really do love an event like this – both to exhibit and attend – and there aren’t too many that top Big Data LDN in our line of work.

Firstly:

Data Governance

This was everywhere throughout the show. Now data governance isn’t new, by any stretch, and getting your ducks in a row when it comes to integrity and compliance has been a focal point for virtually every organisation for a long time, but unless your company truly is a data driven organisation, you did data governance because you had to, not least of all from a regulatory standpoint.

But now that every person and their dog want to be using AI, you must get it right. We all know the old adage “put rubbish in, get rubbish out”, but it’s never been as true as it is today. You simply will not get the best out of any internally deployed LLM if your data governance isn’t up to scratch.

Golden Records

Next it’s golden records – a single source of truth.

We, as a company, work with some of the biggest names in retail, and if you think of a company that has multiple brands under their banner, most of the time there will be disparate customer data sets in various data silos across the place, likely with slightly different formats of home address or phone numbers with international codes or without them, so it can be really tricky to get a view of information like shopping history of a single person, when it’s very hard to match them as records.

The message was, if you want to be getting full value out of your data, and again we’re talking about AI – possibly for analytics – you need a single source of truth to leverage, and when we’re looking at consumer data, golden records are a good place to start.

The final spot on my list is:

Anonymisation and Pseudonymisation

For those that aren’t overly au fair with these terms, anonymisation is the removal of personal identifiers from your customer data, and pseudonymisation is the process of replacing identifying information with random codes of numbers, symbols and codes.

Both of these were cited by the European Privacy Officer of Acxion as key safeguards when making sure compliance is adhered with not only with existing standards like GDPR, but also the New Digital Information and Smart Data Bill which is one of the 40 legislative plans announced by the Labour government in the last few months.

Just as a slight aside, if you’d like to know more about our methods of pseudonymisation, take a look at our JetStream page.

 

So, to recap: -

  • Data Governance – if you want to use AI effectively, get it right
  • Golden records – get your single source of truth sorted, again for AI’s sake
  • Anonymisation and Pseudonymisation – make sure your consumer data is protected, especially when considering new legislation

 

By Steve Clarke – Commercial Director at The Ark

 

If you'd like to have a chat about anything we've covered today: