One Plot, One Site Analysis, Every Decision Logged

Rule a page into four columns and head them: Finding. So what. Ruled out. Keep or cut. That page is your site analysis. The layered base plan, the sun path, the arrows fanning out from a bus stop — those are the presentation of it, and they are the easy part.

When a tutor says the analysis didn't influence the scheme, what they mean is that columns three and four are empty. You recorded things. You never let any of them stop you doing something, and you never threw any of them away. A site analysis that contains everything true about a site is a survey. A site analysis is what's left after you've worked out which of those truths has a consequence.

So here's one, worked end to end on a single plot, with the reasoning left in.

The plot

The site is a composite — the kind we build for teaching, because a real client's plot comes with real constraints I can't publish. Call it the Kiln Street corner. Everything in it is a condition you will meet on a real job.

A former builder's yard on the corner of Kiln Street and Mill Road in a market town. Roughly 26m by 18m. Entirely hardstanding — a cracked concrete slab with a weed line down the middle. The brief is a small community workshop and café, around 300m², two storeys if it needs to be.

The four edges:

• North-west: the blank gable of a three-storey Victorian terrace. Party wall condition, no openings.
• North-east: Kiln Street. Quiet, residents' parking, and the pavement everyone uses to walk to the primary school two streets up.
• South-west: Mill Road. Bus route. Narrow pavement, constant traffic.
• South-east: the rear gardens of four houses, close-boarded fence, first-floor windows looking straight in. A mature lime stands just inside the neighbour's boundary with about five metres of canopy hanging over the site.

The ground falls gently towards the Kiln Street corner. The Environment Agency's surface water flood map shows ponding at that same corner.

Notice what that geometry does before a single diagram gets drawn: the sun comes from the side with the neighbours on it. The quiet side is the dark side. That is the whole project.

Three questions every finding has to survive

Before anything goes on the board, each finding gets asked:

1. So what? Name the design move. Not "informs the massing" — name the move.
2. What does it stop me doing? If it doesn't close a door, it isn't a constraint, it's trivia.
3. Would the scheme be different if this finding were false? If the answer is no, cut it. This is the deletion test, and it's the one nobody runs.

We teach it in that order because the third question is the only one that's hard to cheat.

The log

The blank gable. Three storeys of brick with no windows in it, and a legal party wall condition attached. So what: the building goes hard against it, and everything that doesn't need daylight — stair, WCs, plant, store — stacks along that edge. What it rules out: the pavilion in the middle of the plot, which is what everyone draws in week one, and which here would waste the one boundary you can build up to for free. It also pins the tall mass to the north-west, which is the only place a three-storey element casts its shadow onto the street rather than into someone's garden.

The overlooking. Four sets of first-floor windows on the south-east, about twelve metres away across the fence. So what: no first-floor openings on that elevation, eaves kept low along it. What it rules out: the move the sun path diagram was begging for — a glazed south-east wall over two storeys. Which is the first real conflict, and I'll come back to how it got resolved. (The anatomy of an overlooking constraint is its own subject; we've gone through it properly in the seven constraints that shape a tight urban infill site.)

The lime tree. An arboriculturalist's BS 5837 survey gives you the root protection area as a drawn circle, and that circle lands on the south-east corner of the plot. So what: nothing with a ground-bearing slab inside it, which deletes about forty square metres of buildable area. But the canopy also shades that corner from late morning through the afternoon in summer, and drops its leaves in October. So the café sits at the edge of the canopy, where it gets dappled shade in July and low sun through bare branches in January, and the tree does the solar shading a brise-soleil would otherwise have to do. One finding, one deletion and one gift.

The school run. Walk the site at 8:40 and again at 3:15, and Kiln Street is full of people. Mill Road isn't — it's a road you cross, not a road you walk down. So what: the entrance goes on the Kiln Street side, pulled to the corner, because the bus stop sits at the junction and the pedestrians arrive from the other direction, and the corner is the only point both flows touch. What it rules out: service access from Kiln Street. You are not reversing a transit van across a school route twice a week.

Which pushes deliveries onto Mill Road — the noisy side, the side you'd instinctively keep closed. And that turns out to be fine, because the programme sorts itself: the workshop is the noise-tolerant part of the brief, so the workshop takes the Mill Road frontage with its loading door, and its mass becomes the acoustic buffer for everything behind it. The noise finding didn't move the building. It reorganised the plan.

The low corner. Surface water collects exactly where the entrance now is. That is an annoying coincidence and it was not resolved by moving the entrance.

When findings fight

That corner is the useful bit, so let's be explicit about how it was adjudicated — because "weigh the constraints" is advice that means nothing.

Three tests, in order, and this is what we teach rather than anything you'll find in a standard:

Is it movable? The root protection area, the party wall, the neighbours' window positions and the boundary line are fixed. Where the water goes is a function of how you grade the ground, and you are grading the ground anyway. Fixed beats negotiable, every time.

What's cheaper to fix later? A threshold level is a detail. You raise the finished floor, you regrade the external surface away from it, you move on — it costs money and a drawing. An entrance in the wrong place cannot be fixed later at any price, because it's wrong relative to how people move, and you can't move the school. So movement won and drainage got demoted to a detailing problem, with the cost consequence written down rather than quietly forgotten.

How many other decisions hang off it? Orientation of the main mass touches everything downstream — structure, daylight, overlooking, the lot. The position of a rooflight touches nothing. Resolve the high-fanout decisions first and let the low-fanout ones absorb the compromise.

Run the solar-versus-privacy conflict through the same three tests and it answers itself. The neighbours' windows are fixed; the sun can be caught another way. Glazing on a south-east elevation is one route to light, not the only one — so the south-east facade stays closed above ground floor, and the light comes in through a north-east clerestory and a run of rooflights over the workshop, which wanted even diffuse light rather than direct sun in the first place. The sun path diagram didn't lose. It just stopped dictating the elevation it had no business dictating.

The findings that got cut

This is the part that gets marked, and almost nobody shows it.

And one that nearly got cut and shouldn't have. The historic OS maps showed a kiln on the plot, which produced a genuinely lovely overlay drawing and changed absolutely nothing about the architecture. Then the site history was read again next to "former builder's yard", and the finding came back from the dead as a contamination desk study — which changes the slab build-up, the foundation strategy and the programme. The drawing was useless. The fact behind it wasn't. Findings can survive for a completely different reason than the one you collected them for, so read your own cuts twice before you bin them.

Proportionality, and the data that doesn't exist

The honest rule we give students for whether to commission a measured or topographic survey: if you can't name the design decision the survey will settle, you're not ready to pay for it. The moment your finished floor level depends on a number you'd otherwise be guessing, you commission it. Before that, a tape, a laser measure, a level and a long afternoon will get you close enough to test a massing idea.

And when the data genuinely isn't there — no topographic survey, a boundary nobody can agree on, an easement rumoured but not documented — the answer is not to stop. It's to design so the scheme doesn't depend on the unknown. Keep the building off the contested 300mm. Assume the sewer runs where the statutory undertaker's records suggest and keep foundations clear of that corridor. Then write the assumption down, with the date, the method and the tolerance, in an assumptions register you keep with the log.

That register is worth more later than it feels now, because the four-column log is the spine of your design and access statement. A planning officer doesn't want your sun path. They want the word because: the entrance is here because of where people walk, the south-east elevation is closed because of four first-floor windows twelve metres away. You already wrote all of that. You wrote it in column two.

The test

Before the crit, ask yourself one question: name the thing you wanted to do that the analysis stopped you doing.

If there isn't one, you didn't analyse the site. You decorated a decision you'd already made — and everyone in the room can tell, because the arrows all point towards the scheme you turned up with.

Get one good deletion in the log. It's worth more than ten diagrams.