DUMP Big Data, DEMAND Better Data!Published: Friday, 15 September 2017 by Matthew Tod, Head of Analytics
I have seen a number of clients whose Big Data initiatives, like many, are floundering and destined to fail without corrective action. Lack of experience of working with high business expectations and the noisy, messy, real data in the commercial world is often behind the challenges the teams face.
Given the learning and capability development we are all going through I’m not sure floundering equals failure, it’s a form of accelerated learning, so I take my hat off to them for getting really stuck in.
When asked for my opinion on what needs to happen to help them breakout, one topic I frequently talk about is data quality.
Why? Have a look at this quote from somebody way more experienced than most of us:
"More data beats clever algorithms, but better data beats more data."
Peter Norvig, Head of Research. Google Inc.
This is such a shrewd observation that rings so true with me. Focusing on clever algorithms is so tempting, get ever more data seems logical but better data? Surely more is better? It’s all about Big Data, right?
In the bad old days, we defined the data we wanted, carefully designed a collection strategy, painfully crafted the collection mechanism and got a small amount of high quality data. It worked, but it became limiting as analytical tools and technologies became more powerful. Now we can collect everything, and we see a new problem - volume has increased at the expense of quality. This lack of quality means many projects struggle to progress because of the amount of work needed to make the data fit for purpose.
So, what’s the answer? Better Big Data!
I confess to being an advocate of collecting more data, see my post Collect Everything Once for details of doing it cost effectively. But I am an even bigger advocate of Better Data, in fact I’m told I can be pretty demanding about the quality of data!
I DEMAND Better Data that is:
- Compliant and legal as a base level, so that means ensuring data protection rules, and shortly GDPR, are followed
- Customer centric, channel agnostic and device agnostic are essential for the work I do. Digital data silos are a ridiculous hang up from the past
- Complete and that has been deduplicated to stop me wasting effort
- Structured to save me time, and contextualized to make it more useful
- Behavioral and experiential, one without the other is very limiting
- Detailed and granular so deeper analytics can be done
- Enriched with other data sources to facilitate better insight and has key features already built in
Did I miss anything? Please comment if you can think of anything else you want from Better Big Data.
When it comes to developing our Customer Data Platform, Celebrus spend a lot of time focusing on creating Extraordinary Data, that is data which enables our clients to deliver extraordinary results. We put many hours of time into shifting the data quality issue into collection so that our clients don’t have problems as the start to analyze, model and operationalize our data.
LinkedIn: Matthew Tod