BERG: Schooloscope

Turning official government data about schools into easy-to-read language (and faces) so parents could quickly get a feel for what a school is like.

A Schooloscope page for Brockenhurst Church of England Primary School, with a smiling house, a plain-English summary and a map

A Schooloscope school page: a plain-English summary, a smiling 'house face', and the data underneath — all answering one question, ‘what is this school like?’

Schooloscope turned official government data about schools into easy-to-read English — and smiling faces.

Team: Matt Webb, Jack Schulze, Matt Brown and Tom Armitage. Funded by 4iP.

The question a parent actually asks isn’t “what is this school’s contextual value-added score?” — it’s “are the kids happy here, and is the teaching any good?” Schooloscope (known inside the studio as ‘Ashdown’) tried to answer that question, pulling together data from Ofsted, the Department for Education and Edubase into a single, friendly page for every school in England. It was built by BERG and funded by 4iP, Channel 4’s digital investment fund.

The honest part of this story is that the two people who did the heavy lifting weren’t me. This page exists to point at their work.

Toiling in the data mines

Most of BERG’s ‘material explorations’ were with tangible stuff. Here the raw material was the data — and you can’t design with a material you can’t yet feel. Tom Armitage did the material exploration: the patient, unglamorous work of getting deep into the datasets — what’s in them, what’s significant, what stories they want to tell — building throwaway models and quick visualisations until the rest of us could see what was there. It’s a brilliant, candid piece of writing about what data exploration actually feels like: the overwhelm, the overnight processing runs, the rhythm of coding bursts and waiting. The product only became designable once Tom had done that.

Humanising data

The other leap was Matt Brown’s.

We’re wired to find faces everywhere — a quirk of perception called pareidolia. (I’d been collecting examples for years in a Flickr group called ‘Hello Little Fella’.) The brain’s fusiform face area lights up for a face-like pattern in about 165 milliseconds — faster than we can read a number.

A grid of everyday objects — a meter box, a power socket, a plate of food — that all look like faces
Pareidolia in the wild: a few of the 500+ ‘faces in places’ from my Flickr group, Hello Little Fella. We see faces everywhere — and read them instantly.

There’s a visualisation technique that exploits exactly this: Chernoff faces, invented by Hermann Chernoff in 1972, which map the variables in a row of data onto the features of a face, so patterns in many dimensions jump out at a glance.

Chernoff faces representing 2005 National League baseball statistics, one cartoon face per team
Chernoff faces applied to 2005 baseball statistics — each face is a row of a spreadsheet you can read in an instant.

Matt took that idea and ran with it, designing ‘Chernoff Schools’: every school drawn as a little house with a face, its features driven by the data. He worked it out as code-on-paper first — a Face class whose roof height, wall bulge and eye-spacing were all variables.

A notebook page sketching a 'class Face' with roof height, bulge and eye-spacing as parameters
Matt Brown working it out as code-on-paper — a ‘Face’ class whose roof, walls and eyes are driven by a school’s data. 16 November 2009.
A printed grid of around eighty house-faces exploring the parameter space, with a legend mapping features to data
Exploring the whole space by hand — and the mapping underneath it: roof height = overall Ofsted grade, smile = value-added (CVA) score, eye distance = gender mix, and so on.
Rendered, coloured house-faces with their attributes listed, plus a colour-value matrix
The house-faces realised in colour — each one a legible little portrait of a real school.

The faces did something a league table can’t: they let you scan a whole neighbourhood and feel the difference between places before reading a word.

A 'Schools Nearby' list, each school shown as a house with a happy or worried face and a small pie chart
Schools nearby, at a glance — a cheerful or worried roofline tells you more, faster, than a number ever could.

Over time the site shifted from being pretty-but-heavy information design towards something warmer: plain natural language, friendly faces, and a tone that treated parents as people rather than analysts.

Schooloscope launched in 2010 and ran until May 2012. The faces got the attention — but they only worked because of the brilliance behind them. Thank you, Tom and Matt.