04Google and Waymo · 2018–2021
Eighteen reports a year was the ceiling
The story is a ceiling that breaks. Nobody asked for a fix; I built BloQ in under three weeks, and production went to 200 a year, self-reported.
How this was made
Each square is one report a year: 18 before BloQ, 200 after. Each figure is one designer on the dataviz team. The red sliver is 2% of its bar, and the two retainer bars are drawn to scale, $400K against $5M.
All of these come from slide 59 of my own portfolio deck, so they are self-reported. The colour blocks in each square are BloQ's Mondrian language, not the reports' content.
Cutting the time to insight
Three weeks, or one screen. Answers used to wait 2 to 3 weeks for a boutique data firm's handmade booleans and dashboards, self-reported. Now one screen answers, through three lenses: Competitors, Cross Shop and Retention.
How this was made
Each square is one day: 14 solid for two weeks, 7 hatched for up to three. The 2 to 3 week wait for a boutique data firm's handmade booleans and dashboards is my own account, self-reported. The dashboard that comes back is a symbol, not a capture.
The three screens are Vision 2.0's Competitors, Cross Shop and Retention, redrawn in vector from the Design Showcase, slides 24 to 31. Their brands are fictional and their figures illustrative, as on the case page.
The home crowd came third
Two audience maps from 2019, a Bundesliga club’s and a Formula 1 world champion’s: 10,856 accounts and 306,501 links, drawn to show who a brand is really talking to. In 2026 I recovered both and rebuilt the map as a live explorer.
How this was made
Each dot is one account in the club’s 2019 map, at its own position from the recovered Gephi export; nothing is sampled. The columns stack the same dots by the country group the map gives them. Each line is one of the 6,395 links between the UK and Spain groups.
The club and the champion are not named. Every count is as in the files, recovered from Maven Road’s archive in October 2026.
Seventy-seven out of a hundred, and still no price
Twenty open-data reports on 3,210 small-business websites, each in its industry’s own look. On the score the sites look fine; on what a customer asks first, what it costs, most stay silent.
How this was made
Each dot is one website in the 20 reports (index.json, October 2026). Each industry sits at its own median score. A lit dot is a site that answers that industry’s price question: its share of the sample, rounded, from each report’s data.json. The insurance report asked no price question.
The questions are read from each home page and the pages it links to; a check passes only when the scan finds it.
Ten analysts a week, or one screen a day
The weekly ritual: ten analysts filling a sheet by hand every week, then one screen, refreshed daily.
How this was made
Each row is one of the ten analysts who gathered each country's data every week (deck slide 48). Each bar is one quarter, Q1 2017 to Q2 2019, read from the dashboard captures on slides 41 to 48: Twitter total impressions and Instagram average impressions per post, each panel on its own scale.
The Q4 2017 changes are the ones the capture prints, +296.1% and +111.5%. The 25% to 45% engagement lift comes from slide 48 and was not measured by Google, so it is self-reported.
Send the website, not the proposal
Old site in, new site out: a crew of five agents audits a site, redesigns it and writes the pitch.
How this was made
Each square is one page of Local Friend Tour's site: 41 mapped, 39 with a Spanish twin. The times, the score of 61 of 100 and the checks before going live follow the Clikiti film's Local Friend Tour story (components/home-v3/line/story.ts).
The old site's page in the film is staged; the new site is live at localfriendtour.com. The cats are Clikiti's crew: Whiskers scouts, Shadow checks, Tiger designs, Paws writes and Sphinx decides.
Thirty minutes to know who was turning on Google
An earthquake on the cover: one post, a ripple through the people who shape opinion, and an editor on every alert.
The featureThe files