Shitty Experiences Won’t Die

We’ll just get better at spotting them automatically!

I’ve become convinced that much of what I’ve talked about recently — in terms of software quality, defects and device compatibility — isn’t soluble simply at the production end.

Let’s be clear — I’d like the solution to be that we engineer better software with less defects but our assembly lines aren’t geared towards that. Velocity and risk often play a part during times of growth and with egos involved, testing or carefully measuring stuff is simply skipped. With disastrous or costly consequences.

If cars were designed like digital experiences, there’d be a national outcry. Imagine if we went back to ‘non safety glass’ or ‘no seatbelts’ or ‘crumple zones’ or ‘crash ratings’ for cars — these compromises are unacceptable in terms of safety and comfort for modern vehicles. So why hasn’t software improved in ‘delivered quality’ when it comes to these products?

Well — the answer is — it’s HARD. Building an app or experience for one device, one substrate, operating system or browser is work enough. Building software that works flawlessly on every device, browser or viewport — is really really tough. When it works, I want to break into song — but that’s a rarity and often only on products I’ve designed all the interactions for (laughs sarcastically).

Rather than despair at our inability to rewire quality into the assembly line like car manufacturing, we will use machines. Yup.

Oh and before you tell me the tools already exist — I know. Vendors who collect behavioural or analytics data can now run models and algorithms that weren’t possible before.You can now start to match patterns, even if the machines don’t fundamentally understand them.

A machine can sense or notice a trend or cliff edge in collected data, across multiple dimensions of customer or device attributes, far faster than your poor brain can do it retrospectively in an analytics package. And all this in real time too.

There’s one slight problem though — all these nice bits of software are missing something — a meta layer. That’s where all my work of the last 15 years comes in.

Blind intelligence isn’t enough here — but training a tool around the specific examples our experiential knowledge has given us, is far better than a tool which has no knowledge but raw data. Spotting patterns is very nice but only certain patterns are useful or exploitable so much feedback comes with noise or triggers that contain false negatives or false positives.

We need someone to help us with the models and ML work — but we will supply a software platform, a team and the meta knowledge to train a better class of watchdog. For this tool’s job is to monitor, spot, isolate and explain customer bugs, defects, ux issues — in a smarter way than those trained merely around patterns and anomalies. It’s also going to tell you if you manage to break something you already fixed too ;-)

If you’re interested and want to work with a team of the top analytics, conversion and UX brains on the planet, we’d like to talk to you. We have some wonderful ideas which we’re working on for 2018 and we’d love to have your help.

Please DM me @OptimiseOrDie or on Linkedin:

Like what you read? Give Craig Sullivan a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.