One of the first products Agile Stationery released was a deck of "Size and Sentiment" cards, a bundle of cards for, frankly, a pretty vaguely defined audience. We knew some people used T-Shirt sizing for estimating, and others used planning poker™ to pick a size. We also knew some people did squad health checks and other exercises where it was important to capture sentiment. The Size and Sentiment deck addresses all those together in a cost effective way, but summing up what the deck is for takes a long paragraph, like this. It felt awkward in marketing copy and sales and click-through rate statistics both disappointed.

Focusing on lights

Our next step was/is to focus on squad healthchecks and make a truly excellent squad health-check deck. The original Spotify health-check deck is licenced CC-By-SA, so there should be no barriers to making a commercial version. There is also value to be added in refining some of the text and getting them printed and boxed professionally.  We were keen to start.

In addition, we got some timely feedback that the particular green and red we'd picked for sentiment cards were not great choices for those with red-green colourblindness.  As a developer with 14+ years of experience I confess to being quite surprised by how little I knew about colour blindness. At university the message had been diluted (by the professor or possibly by beer)  to "avoid putting red and green together", but of course red and green go together in traffic lights. My education, something I usually look back on fondly, had let me down. 

I started googling around and reading up on the topic of traffic lights and colour blindness. I knew from the feedback - thanks Matthew - that the particular red and green had been tweaked in some way, possibly involving blue. I wasted hours looking for for the hex values of real UK traffic lights and lingered at the lights on the nursery drop off dozens of times peering at the red and green. I considered taking photos but thought I would not get an accurate result that way. Eventually I stumbled upon a colour blindness simulator, and it started to make a bit more sense.

Looking at the simulators and reading around about the various conditions made it very clear that not only were there different kinds of colour blindness, but the impact on the patient is very different in each case. It is also the case that the prevalence of different kinds of colour vision deficiency vary enormously. True colour blindness - achromatopsia -  is exceedingly rare.

I had worked out that in real UK traffic lights the red was a little orange and that the green was very blue, this was reinforced by photographic reports like this one that a bluey green is helpful. In Japan green lights used to be blue and are still called "blue", though presumably in Japanese.

Smiles for Solutions

A lot of experimentation in the simulators, and a bit of work in a colour picker showed these shades were good options for accessible colours, and we treated these as a design starting point:

We already decided we did not want to move too far away from the familiar traffic light analogy, rejecting the Nova Scotian solution of using different shapes and also the suggestion of putting smileys over the colours. For us this lacked seriousness, and we already worry that some Agile techniques suffer from that perception already.

We needed to put the needs of the many before the needs of the few, or of the 1%, while still showing everyone respect. We decided to take a similar approach as I would to system availability - quantifying accessibility as a percentage and aiming for as many 9s as possible without fatally compromising the user experience for the majority.

It was suggested that one place we could apply smiley faces - or emoji in fact - was alongside the traffic light. We replicated the design style of playing cards where the suit is added to opposite corners both ways up, so there is no "right way up" for the card. Happy, neutral and sad emoji would fit intuitively alongside the green, amber and red lights. On they went.

Subjective Issues

There are couple of very subjective design decisions which we made that are interesting. We decided to stick with traffic lights over other symbols and we decided to invert the sequence. We are showing Green over Red on theme cards and canonically, in marketing, will show Green Amber Red (the order of actual cards is immaterial for obvious reasons). 

We could have thrown out colours completely and supplied a deck of happy neutral and sad faces. That is still a great product idea and we might just do that, but it would not be a good choice for a truly excellent squad health-check deck. The reason is that we are not really capturing sentiment. To be honest, our older product has a bad name. Expect it to go on sale soon!

Instead we are prompting users for an assessment of whether current arrangements suit the empirical process of delivering concrete value to stakeholders. There will be sentiments expressed about these assessments but the assessments are not primarily emotional. Emotion is expressly secondary (except possibly when the theme is "fun"). Traffic lights, since they pertain to real world safety concerns, are a much better analogy.

At the same time, we are certainly not in the business of promoting negativity so when it comes to listing outcomes on the theme cards we make the inverse decision, we put the negative phrasing second for essentially emotional reasons. We have done this despite sympathy for users with achromatopsia, who will certainly be more used to seeing traffic lights shown the other way up.

This is probably not very scientific, but so be it.

How we think we did

Evaluating our design was a paper exercise. We don't yet have a large contactable customer list and we are not sure polling it for guinea pigs is a sympathetic approach if we did. Instead we went back to the sources and simulators.  We added up the population groups which the simulators tell us we managed to satisfy. We tweaked the standard for "satisfied" according to the two parts of the health-check deck.

Theme cards

These cards are conversation starters and are read and discussed as a group. We assumed that teams discussing these together would be supportive of any members with a colour vision deficiency. Amber is not used on these cards.

The needs of those with deuteranomaly, who are 3.2% of population, have been addressed by giving them some better colour shades. Only 1.0025% of the population should, we think, have any problem with the colours. Protanopia and deuteranopia sufferers (1%) see two reasonably distinguishable browns. Through our network we located one color blind individual who reported a consistent experience - thanks Chris for arranging that. 

We didn't add emoji to the scenario cards as we felt this would make them look crowded for 98.9975% of users who would not need them at all. The position of the lights would also be helpful for users with protanopia, deuteranopia and  achromatopsia, just as it is with real traffic lights.

Only the last 0.0025% of users with achromatopsia will have to struggle as they will just see two very similar greys. Although we do have sympathy for those at the wrong end of our chosen compromise, given the team context and prevalence (0.0025%) we feel sure that this is an acceptable compromise for the product. 

Traffic light cards

For the traffic lights cards (the bulk of the deck) the emoji are present on those cards to support all users with colour vision deficiency. Those with protanopia and deuteranopia should feel quite comfortable, so we think we have achieved four-nines -  99.9975% of users with a comfortable experience of this part of the deck.

Those with achromatopsia will rely on the smileys and should also get by just fine. There is no uncomfortable compromise for the bulk of our deck, which certainly makes us feel better. 

Conclusion

Obviously it is difficult to apply utilitarian calculus to something which is in part an emotional concern. I am not even much of a fan of utilitarian philosophy, but we do feel that by addressing the bulk of the problem for the colour blind population we had made a good product. The market will be the ultimate test of that, as is only right for a commercial product.

The next step is to get a small batch produced and on sale. After that we hope that people with colour vision problem will get in touch, via reviews or social media, with this post acting as a lightning-rod for your opinions. 

Thanks to Matthew and Chris for their assistance.
Header image is © i_yudai