image image 2nd December 2015

How to increase the ROI of a data warehouse

Data warehouse projects are expensive!

Depending on how you implement your project, it could well take over a year to deliver, even with a small army of business analysts, project managers, developers, testers, data analysts, report builders, data modellers (and someone to make the tea). That’s before you even consider which flavour of (hugely expensive) database management software to use, where to host it (in your own data centre or on the cloud somewhere) and ongoing maintenance (insert a few DBAs here for good measure).

So – how can you improve your ROI on this huge investment?

Well, the first one is obvious – make it cheaper.

Many businesses I visit utilise the same database management software they signed up to 10 years ago, and don’t even consider moving platform. I appreciate that there are various costs associated with this such as potential re-training of staff, which is seen as prohibitive… I’m not going to address this point fully in this blog post (that’s one for another time!), however I’ve seen quite a few instances where a change in platform has resulted in some retraining of staff (expensive) with significant reductions in staff attrition rates, and therefore reductions in recruitment costs (very expensive!). Sometimes, engaging your techy staff with new technology can have some very significant, unintended results!

So – you can consider moving platform. Do you really need Oracle Enterprise licensing? Or could you actually run MySQL (I’ve heard it runs at pretty large scale – ever heard of Facebook or Pinterest?). For some workloads, using a Hadoop distribution rather than 4-socket, half a terabyte of RAM monster servers may help.

How about sticking it into the cloud somewhere, and run a process where you can spin up and down servers as you require them?

To be honest, there are HUNDREDS of ways to reduce cost… but this isn’t where you’ll make the biggest improvement to ROI.

The second option is potentially the best – get people using the data!

From our experience, the success measure of a data warehouse project is delivery – i.e:

Congratulations! You now have a big database, nicely modelled on an expensive server! Plus is was only delivered {insert number here} months late!

As I’m sure is obvious, this is not a measure of success. A data warehouse isn’t any use if its not being used. If your query servers are idling at 15% CPU utilisation, then you’ve failed. These should be being hammered all day, every day by your hungry users.

Once, I worked for a financial institution where we discovered that, by increasing the lower bounds of a single variable in a credit-risk scorecard, we could save the company upwards of £7million in bad debt per year. This analysis took somewhere in the region of a day to produce, and a day to implement. Without a suitable data mart, this analysis would have been taken many days longer – or may never have happened at all.

So – one day, one user, £7million. Well, that’s more than paid for the initial data warehouse project.

How many potential users do you have who currently don’t have a way of accessing your data warehouse? How many don’t have the freedom to really delve into the data? It sounds obvious – if you users aren’t investigating every corner of your data warehouse, and throwing hundreds of requests to get new measures added, new structures modelled to aid analysis, or new data sources added then you’re not making the most of your initial investment.

Data warehouse projects are cheap, as long as you don’t view just having one as the end game.

We Love Data

Want to know more?

Drop us a line – we’re always happy
to chat – we promise we’ll keep the
geek speak to a minimum (unless
that’s your bag in which case we’ll
happily comply)!