March 22 2019


Monday and Tuesday:

Still iterating the prototype. Feels much saner. There’s still one thing left that’s not sane, but Wednesday we’re hoping that we can prove it’s either just about workable (for MVP and seeing what happens in a pilot) or another thing to throw away.

We had a little “sort of pivoting” planning meeting too, which felt very positive — after knocking some work out of the sprint, we looked at bringing in things that can help our feature delivery but also the wider product.

A good chat came out of that meeting about some design work to be broken down. It felt constructive to be critted on work that had been sort of left fallow for a good while. And typically I felt far more able to articulate decisions and reasoning than I do while these things are fresh. Story of my life really.

This is good, a nice bit of API hacking from Andrew Duckworth. Smooches.


Our next research session. This iteration of the prototype feels way better. People get through it easily enough.

We’ve still got things to consider:

  • ways into this specific journey
  • we’re still asking tricky questions that we shouldn’t
  • how to sell in a referral

Ways in

How might people actually find our journey? We haven’t yet paid huge attention to this, concentrating instead on making the journey itself workable.

We do know that relying on our search to get people into an emergency prescriptions pathway won’t be at all reliable.

Most of the users in front of the prototype searched for the symptom that the prescription is meant to deal with, as opposed to something like “emergency medication”. We do know from analytics that terms like this are things that people search for, so this is one way in.

What other ways are there? Next sprint we’ll put a lot more energy into working that one out. Especially as we’ll pilot with a geographical area, just to add to the fun.

Tricky questions

We quickly proved that asking people to name the components of their prescriptions is a sure way to confuse them and get them stuck. Not to mention that as an organisation we already know this. So that’s gone from the prototype.

So in a step back from there, we’re asking whether people are on controlled drugs. If you’re on controlled drugs, there’ll be a different journey. So we’re attempting to filter users into that different journey.

But there’s a risk of false positives. People who are on controlled drugs seem to know about it. People who aren’t are way less certain.

Our prototype asks, provides helpful details around what might or might not be controlled, and allows the “don’t knows” to continue along the main journey.

But again, there’s uncertainty. And we already know!

Selling in a referral

This one’s a bit of a fucker to be honest.

If you refer yourself into particular services that might be returned via DOS, you are guaranteed to pay NHS prescription prices. You do that by providing some info in order for an electronic referral to take place between the 111 service and the pharmacy.

What that referral does is:

  • provide a recorded linkage in a healthcare journey between 111 and a pharmacy
  • enable NHS prescription charges (as opposed to it being a private service)
  • move clinical responsibility onto a specific pharmacy, who is then tasked with helping the user

It’s a bit of a classic in that:

  • it’s counterintuitive, doesn’t fit the user’s mental models
  • it uses the patient as one factor in an auth mechanism
  • it doesn’t guarantee help at the place you’re being referred to
  • it doesn’t get you nearer your actual goal even though we’re asking you to work


In order to decide to refer into the service, you need to be able to make sense of that service first:

  • what is it?
  • where is it?
  • is it open long enough to get there?

Once you’ve decided in the affirmative, then ideally we’d like you to refer in. But at this point you may well decide to simply phone them or just go there.

We shouldn’t force a referral like this for a couple of reasons:

  • we shouldn’t deny access to the service (in this instance) just because the user doesn’t refer
  • not all services that can help will offer the referral
  • not all services will be of the same type (see last week about agnosticism)

So we’re struggling with ways to show users both the service we’re recommending and tempt them with the myriad advantages of referring themselves. So far it’s been a bit reminiscent of my banner ad days.


Lots of calls, essentially schlepping our prototype between different bits of health and care. From NHS England through to areas we might be able to run a pilot with.

Seems the consistent rule to services is variation. For instance, the controlled drugs question we’re asking to filter users into other journeys may or may not apply in different service delivery areas.


We had a playback of research to the wider team from the last couple of sessions we’ve done with users.

Otherwise deep despair on my part about ever getting anything done. So it goes.