Last night at Sapient Nitro’s London HQ October’s Great UX Debate took place. Lead by Giles Colborn, the panel – comprised of Lee Sankey, Cennydd Bowles, Daniel Harvey and Windahl Finnigan – spent around 90 minutes debating various UX-related topics.
Towards the end of the night the debate really did kick off in regards to whether design (not just UX, but design in general) needed more representation at a higher level in business and organisations. While this was probably the most engaging part of the debate (with strong points and rebuttals being made on both sides) for me the most poignant points of the evening were several made by Cennyd Bowles.
To paraphrase Cennyd (hopefully not too wildly):
I’d rather ship at 80% perfect and get to work on something else, and get that to 80% as well, rather than spending three weeks getting the first thing to 98%
Cennyd also called ‘bullshit’ on this belief we need to user test everything. He qualified this point by arguing that – if experienced enough – UX designers should know what will work and what won’t. The power of heuristic evaluation – perhaps a forgotten process these days?
This goes back to his 80% v 98% analogy: if you’ve designed a working login process before, you should be pretty confident doing so again (and confident that it will work for at least 80% of users). To get it to 98%, it might take weeks of work (not to mention additional costs for testing and development). 98% is great but in that same time what if you could get a search system implemented at 80%?
I know what I’d rather have. I think there’s far too much fear and anxiety about launching products and features, and not just from some in the UX world – this extends to developers, project managers, clients etc.
Of course you don’t want to ship something that’s broken, but we’re thinking about 80% here. That should be well beyond the point of minimum viable product. It’s not perfect, but it’s built and out there – it’s alive and not languishing behind closed doors having thousands of tiny details polished to a ridiculous degree.
We have a huge luxury in the software and online world: the final product can be updated after it’s been shipped. So we can iterate on our 80% easily – and moving beyond 80% on a live product is even easier as you can analysis usage metrics that can sometimes give you insight far greater and truthful than 5-10 users in a lab in hour-long sessions. (Edit: as Fabien Marry pointed out, this will only give you quantitative feedback, not qualitative feedback).
I’m not for a minute saying don’t user test. But we need to look at this in the big picture and use our expertise to make a professional call in many situations. With that in mind, my final take away from the great UX debate? The adage of the software engineering realm: ship early, ship often.