Research methods for developing services that use machine learning

Existing services are challenged when AI becomes involved. Testing a system's design, functionality and content before and during development reduces risk and improves usability.
Image of a robotic hand making a chess move

How AI grew up playing games

In 1997 "Deep blue" became the first machine to beat a reigning world chess champion. Though Kasparov went on to win the series 5:1, it was a significant milestone.

AI is packing the games away and getting a job.

"We will be gamed" - How we don't always win when games are used to develop real world systems.

The research plan

Plan for each phase, each sprint and update the plan A research plan can be one of the outputs of a design sprint, when there is one.  If not, it's still helpful to write one as soon as possible. To be accessible to both team and stakeholders,…
analysing and communicating usability issues

Tips for usability testing and recording results

Reviewing and recording results takes headphones and about as many hours as was spent testing.
Not all surprises are pleasant

Not all surprises are welcome

A good reason to start researching sooner rather than later is to address the questions and inform the decisions teams have at the start of a project.
Image of a group of research participants evaluating a webpage

Turning opinions into results

Previous posts in this category describe how Neilsen's popular methodology of Discount Usability Testing can identity around 80% of a site's usability issues. It's quite "doable", typically taking a day, with 5 people coming in for…

Challenges with designing local government services

A recent assignment with local government brought home the challenges of introducing digital, agile and service design into a traditional service provider; into an environment that's neither digital first nor by default. In the UK local statutory…
How gov landing pages are laid out

A language of visual design – the grammar of digital engagement

Researching for a county council involved looking at other local government websites to see if there was a structure to the visual design, if they used a visual language. Unlike which is very uniform and accessible, local authority sites…
What people say is different from what they do

Formative research – techniques and tips

The post "Different types of research", categorised user research as either formative or summative (aka generative/evaluative): - Formative investigates environmental and human factors, constraints, opportunities, behaviours,…
Patient contact with health services


"More time is spent researching, analysing results and gleaning insights, than communicating the results" - discuss. That detracts from the impact of research, and doesn't help answer the important question - "So what?". Visualising_data helps…
Post it notes stuck to a wall

Taking down a wall of sticky notes – visualising_data

The patent on sticky-notes apparently expired in the 90s, some time before UX took off. That's 3Ms loss, as user research consumes vast quantities for collecting and visualising_data. What to do with a great mosaic of comments ? A wall…
Venn diagram - Diagram

Different types of research

Who researches what In the world of UX and UR, research can be crudely  categorised into - Formative - What's useful Summative - What's useable The first helps to create the product by investigating its context, environment,…
Title printed by Dymo

Analogue, digital, visibility, affordance and Klingons

It might not zap Klingons but nonetheless I know what to do If you don't remember the original Star Trek, you'll probably not recognise the quote or think the green Dymo printer resembles a phaser gun. The idea…

A checklist for evaluating webforms

Heuristics At some point, looking for a definition of "heuristic" turns up: "rule of thumb", which might or might not be helpful depending on whether you know that means "rough and ready", widely accepted, approximations that are "good…

Preparing for 5

Understanding usability testing 101, describes how "discount testing" with 5 people and no lab goes a long way toward identifying usability issues. The practical benefits are: - 5, 1 hour sessions constitute a good day's work Its…

Undertaking usability testing – a simple methodology

In the late 90s Jakob Nielsen and Tom Landauer established 80% of usability issues could be identified by 5 users. So a day of usability testing, 5x1 hour sessions, has the potential to dramatically improve a site's usability, and so it's…

Internet access by age demographic

Data from the UK's Office for National Statistics represented in an Excel radar (spider) graph. It shows 37% of those surveyed aged 65+ had accessed the internet in a three-month period; which seems high for the demographic. It would…