Signs of a data project gone wrong
Trade Secret: Too many BI and analytics projects fail.
I saw this data on smart meter rollouts and while it's interesting in itself, I think it also illustrates the warning signs of a data project gone horribly wrong.
Any career in analytics will expose you to some less than successful projects. The story of Britain's smart meter rollout has three signals I've learned to look out for:
1. Insight nobody can action
If a project offers numbers without actionable insight, it'll suffer the worst fate of all: Being ignored.
My smart meter shows the energy I've used. It doesn't tell me which appliances to switch off to meaningfully impact my bill. So I ended up unplugging it.
2. Painfully slow delivery
Data projects must answer questions quickly, or the question will change before you're done.
At current rollout rates, smart meters won't be universal until 2038. The problem they solve will likely look very different by then.
3. Scope is too broad
Data projects with wide ranging aims tend to just drown us in numbers. It's better to concentrate on one solution that will definitely add value, rather than ten which might.
Smart meters promised a lot: Instant insight on energy efficiency, demand balancing, realtime billing, saving Planet Earth.
They produce a mountain of numbers but thus far they only helped billing.
What signals do you look for to keep a project on track?