Max MorrishThe Playbook

How the work gets done, chapter sixteen

Honesty is part of the job

Normal vision Kiln stock this week Huila Copper espresso Night Shift decaf Morning Roast in stock low sold out Sold out 4.10:1 red text on the ground normal text: fails AA large text: passes AA Protanopia Kiln stock this week Huila Copper espresso Night Shift decaf Morning Roast in stock low sold out Sold out 3.09:1 red text on the ground normal text: fails AA large text: passes AA Deuteranopia Kiln stock this week Huila Copper espresso Night Shift decaf Morning Roast in stock low sold out Sold out 4.74:1 red text on the ground normal text: passes AA large text: passes AAA Tritanopia Kiln stock this week Huila Copper espresso Night Shift decaf Morning Roast in stock low sold out Sold out 4.05:1 red text on the ground normal text: fails AA large text: passes AA In two of the four, the red and green dots are hard to tell apart, and the board uses colour alone.
The graphic

It stays in your browser. Nothing is uploaded or sent anywhere. Images only, up to 10 MB.

Two colours to check
Contrast of the text colour on the background, in each vision. WCAG 2.2 asks for 4.5:1 for normal text and 3:1 for large text at AA, and 7:1 and 4.5:1 at AAA.
VisionRatioNormal textLarge text
Normal4.10:1fails AApasses AA
Protanopia3.09:1fails AApasses AA
Deuteranopia4.74:1passes AApasses AAA
Tritanopia4.05:1fails AApasses AA
Kiln's stock board for the week, an invented example, as seen with normal colour vision and with protanopia, deuteranopia and tritanopia. The board marks stock with colour alone: green for in stock, amber for low, red for sold out. With protanopia or deuteranopia the red and green dots turn into two shades of the same olive, so a reader can't tell which roast has sold out. Under each panel is the contrast of the red "Sold out" text on the dark background, which fails WCAG AA for normal text in three of the four. The fix is words and shapes beside the colour, not a different red.

Those questions do not make the internet more fun. They make it more accurate.

From my draft of the essay in Chapter 10, 7 June 2026.

The idea

I think accessibility, true claims and disclosure are part of the work, not a check someone runs once it's finished. A chart has to read for someone who can't tell red from green. A video has to make sense on mute. Alt text, captions and plain words aren't extras. They're the same job done for more people.

Numbers work the same way. If I can't say where a number came from, it doesn't run, however good it sounds. Spec work gets labelled as spec work, an ad gets labelled as an ad, and the people whose ideas I use get named.

See it as they do

The same graphic four times: once as most people see it, and once for each of the three kinds of dichromacy, where one of the eye's three colour receptors is missing. Pick a graphic from this Playbook, or one of your own, and two colours to check against each other.

Captions, for people watching on mute

The clip below has no sound track at all. That's the point. People watch with the sound off, by choice or because they can't hear it, so the words have to be on the screen. The button switches the captions on and off, and the transcript under it does the same job for anyone who can't play the video.

Transcript

This clip has no sound at all. If the point lived only in the audio, anyone watching on mute would miss it. Captions put the words where everyone can read them. The transcript under the video does the same job without it.

What you see: three of the marks from this site drawn as light, one after another, over twelve seconds.

Before a number runs

Three questions I ask of any statistic before it goes into a deck, a post or a page. If nobody can name the source, it doesn't run. If the source says something else, the sentence goes. Anything still unsure gets re-sourced first.

An invented example. Put any statistic you're about to use here.

Where is it from? Can you name the original source, not a post quoting a post?
How old is it? Is it recent enough for what it describes?
Does it say what we say it says? The same people, the same measure, the same wording?

Verdict

Run, re-source or drop

No named source: drop it. A source that says something else: drop the sentence. Three yeses: run it, with the source beside it. Anything else: re-source it first.

Where I used it

  • The links on this site. They share the text colour, so the underline is the only thing that marks them, and I kept it on purpose: WCAG asks that colour isn't the only signal. A design linter I run flags the underline every time, and a comment in the stylesheet says why it stays. Solo
  • The UVA Food Assist redesign, a group project. The strategy playbook I wrote for the group followed Feeding America's language guidance, so the copy talks about "neighbors facing hunger". The lesson I took from building the website was about consent: real people's photos and details shouldn't go on a page, even a class project, until they've said yes. Group work, my section
  • The disclosures on this site. The Lululemon page opens by calling itself spec work, "Not affiliated with, commissioned or endorsed by Lululemon or Strava", and every chapter of this Playbook tags its examples as solo, group work or spec work. Behind the Playbook sits a set of IP notes that sorts what I was taught, what's public and what's mine, so each chapter credits the right person. Solo

What I'd do now

Some numbers from my earlier work shouldn't come back. Before this Playbook reused any of them, I went through each one, and here is what happened to them. The dropped numbers aren't repeated, since repeating a bad number is how it spreads.

The audit of numbers the Playbook might have reused
The numberWhere it wasWhat happenedWhy
An average order value lift for loyalty membersMy Lululemon spec briefDroppedI wrote it as a fact and it was an assumption. No source.
A conversion-rate benchmarkThe same briefDroppedNo source I could find for the range.
A reply-rate targetA cold-outreach planDropped as a benchmarkA target is a guess about the future. If it comes back, it comes back labelled as one.
Social platform user countsCourse slidesRe-source before any useAbout five years old. The platforms publish current figures themselves.
An attention-span comparisonCourse slidesDroppedI couldn't trace it to a study.
Open and click rates for personalised emailCourse slidesRe-source or dropVendor figures with no method attached. Chapter 8 uses none of them.
Engagement counts on social mockupsMy mockups for a spec brand conceptKept only as labelled inventionI made them up to fill the mockup. If a mockup appears, it says so.

The rule the other chapters follow comes out of this. Every number on a Playbook page is either something I measured, with where I measured it, or an illustration that says so on the page. Each chapter's safety notes keep a table of every number and which kind it is.

Contrast and the use of colour as WCAG 2.2 (W3C) defines them. The colour-vision simulation uses the matrices from Machado, Oliveira and Fernandes, "A Physiologically-based Model for Simulation of Color Vision Deficiency" (2009). DEI in marketing as taught in a digital marketing program, 2026.