How to Write a Site Analysis Report People Actually Read

Site analysis gets rejected at the sentence level, not the survey level. The maps are usually fine. The sun path diagram is geometrically correct, the contours came off a real topographical survey, the noise sources are plotted. What's missing is the line underneath each one that says what it means for the building you're proposing — and until that line exists you haven't done analysis. You've done data collection with a decent colour palette.

That's why you keep hearing "it doesn't say anything". It isn't a criticism of your research. It's a criticism of your prose.

Every Observation Owes You a Consequence

Here's the only rule that matters. No finding in the document is allowed to end as a description. It has to travel: what's there → what that does → what the design must therefore do, or what it risks if it doesn't.

Three steps. If a paragraph stops at step one, delete it or finish it. In practice, if you apply this honestly to a typical student site analysis, somewhere around half the content disappears, because half of it was there to show you'd been thorough rather than because it changes anything.

Worked examples. Weak first, then the version that does work:

Weak: The site slopes from north to south.

Decisive: Say the survey shows a 2.4m fall from the north boundary to the south. A single-level ground floor would mean cut and fill across the whole footprint and retaining close to the shared boundary. We've taken a split level at the mid-point instead, which keeps retaining under a metre and keeps excavation away from the party wall.

Weak: Sun path analysis shows the rear garden receives afternoon and evening sun.

Decisive: The rear garden gets sun from early afternoon until dusk in summer, which puts the living spaces at the back and the service rooms at the front. It also means the full-width rear glazing the brief asks for faces the afternoon gain, so the overhang depth and the glass spec stop being an aesthetic question and become a comfort one. Flagged as a risk; resolved at the next stage.

Weak: The site is within a conservation area.

Decisive: The site sits within a conservation area, so the character of the street is the thing the proposal will be judged against rather than the plot in isolation. The prevailing rhythm is a two-storey brick frontage with a recessed entrance and a parapet line that holds across the terrace. The proposal keeps the parapet datum and the entrance recess; the contemporary element is held behind the ridge where it isn't read from the street.

Weak: There is a bus stop approximately 200m from the site and a railway station within walking distance.

Decisive: The site is served by a bus stop at the end of the street and a station within an easy walk, which is the basis on which we're proposing reduced parking. If that argument isn't accepted, the alternative is a crossover and a hardstanding to the front, which removes the only mature tree on the plot.

Notice what the strong versions share. Each one names a thing that changes in the building. Each one admits a cost. And each one gives the reader something to agree or disagree with — which is the entire point, because a reader can't engage with a fact, only with a judgement.

Put the Answer on Page One

The conventional report sequences itself like a research process: context, climate, topography, vegetation, circulation, legal, conclusions. Everything builds to the findings at the back.

Nobody reads to the back.

A planning officer has a caseload and a target date. A client wants to know what it means for their house and their money. A tutor is standing up, reading at distance, forming an opinion in the first fifteen seconds. All three of them are skimming, and all three of them will form their view of your competence from whatever is nearest the front.

So invert it. Findings first, evidence second.

1. One-page summary. The whole argument, standing alone.
2. The findings, ranked. Most design-determining first, not in category order. Each one gets a short section: the observation, its consequence, the design response.
3. The evidence. Maps, surveys, sun studies, photographs, policy extracts. This is the appendix and it should be treated like one.

The third section is where everything you gathered goes to live a quiet, useful life. It isn't wasted. It's just not the argument — it's the proof you'd reach for if someone challenged the argument.

The One-Page Summary Sheet

One side of A3 or A4. One annotated site plan, and the following, in words:

The site in one sentence. Not the address. The condition. "A long, narrow back-land plot with a single pinch-point access, falling away from the street and overlooked from two sides."

The three findings that shape the building. Not five, not twelve. Three, each in two lines, each ending in a design move.

The risks. The things that could change the scheme if they go against you: a drainage asset under the proposed footprint, a boundary position that doesn't match the deeds, a daylight relationship with next door that will need a formal assessment.

The open question. The one thing you haven't resolved and need a decision on. Putting this in writing feels dangerous when you're a Part 1. It isn't. It's the single most senior thing on the page, because it tells the reader you know where the edge of your knowledge is.

If it helps, lay the findings out as a table — it's genuinely tabular content, the same three columns repeated:

That third row is the kind of thing that makes a contractor relax and a tutor sit up, and it comes from the site visit, not from a dataset.

When Two Findings Disagree, Somebody Has to Lose

This is the part nobody teaches and the part that separates a real analysis from a tidy one. Constraints conflict. The best solar orientation points the living space away from the street frontage the planning officer expects you to address. The quietest corner of the site is the one with the worst ground conditions. The view is in the direction of the overlooking.

A weak report presents both findings on separate sheets and lets the contradiction sit there unacknowledged, which reads either as naivety or as hoping nobody notices. Neither is good.

Name the conflict. Say which one you've prioritised. Say why — and the why should be something external: a policy test, a brief requirement, a cost, a comfort standard, the client's stated priority. Then say what you've done to mitigate the thing that lost.

"Solar orientation and street frontage pull in opposite directions here. We've held the frontage, because the street's continuity is what the conservation area designation protects and breaking it would put the application at risk. The south aspect is recovered through a rear courtyard and a high-level opening over the stair rather than through orientation of the main volume."

That's the paragraph that gets you hired.

Three Readers, Three Documents

Same findings. Different emphasis, different vocabulary, different length.

The planning officer is writing a report of their own, under time pressure, and the best thing you can do is write sentences they can lift. That means using the language of the policy framework they're assessing against, referring to the immediate context rather than the abstract site, and dealing with the neighbours' amenity before they have to ask. What we've found works: address the obvious objection in the analysis, not in the rebuttal. An officer who can see you understood the problem is more inclined to believe your solution.

The client needs the consequence in money, time or comfort. They do not need the word "permeability". "The ground falls away at the back, which means a retaining wall and a deeper foundation on that side — it's a cost, and it's also the reason you can have the ceiling height you asked for." Two sentences, both honest, no jargon.

The tutor wants the reasoning visible, including the bits you rejected. Show the option you discarded and the finding that killed it. A crit is an examination of your thinking, not your conclusions, so the discard pile is evidence rather than clutter.

Captions Are Where the Argument Lives

Most annotation labels things. "Prevailing wind." "Existing tree." "Pedestrian route." That's a key, not a caption, and it tells the reader nothing they couldn't see.

Write the caption as a claim instead. "South-westerly wind funnels down the alley — the entrance can't go here." "Mature sycamore; its root protection area rules out the southern third of the plot." "Main pedestrian desire line cuts the corner; the frontage should acknowledge it rather than fight it."

The test: cover the drawing and read only the captions. If they still make an argument, the sheet works. If they read like a shopping list, you have a graphic with no voice. There's a whole separate craft in making those diagrams legible at distance, which we've covered in how to turn site analysis findings into diagrams that read on a crit board — but legibility won't save a caption that has nothing to say.

Go Back to Site and Photograph the Problems

The findings that make a report feel like it was written by someone who has actually been there almost never come from a dataset. They come from standing on the plot and noticing something inconvenient.

What's worth recording, in my experience, is the stuff that will later become a sentence: the width of the access at its narrowest point, measured, not estimated. The view from the neighbour's window back towards your site — take that photograph, because someone will raise it. The position of every drain cover and inspection chamber. What the site sounds like, and at what time of day. Which way people actually walk, as opposed to where the footpath is. The condition of the boundary that the title plan draws as a confident single line.

Date and time every photograph. A shadow claim you can date is evidence. A shadow claim you can't is an opinion with a camera.

The Folder Isn't the Analysis

The analysis is the page that tells someone what to do. Everything else is supporting material, and supporting material has never once persuaded anybody of anything on its own.

If you take one thing into your next report: write the summary page first, before you've drawn a single diagram. If you can't fill it in, you don't yet know what your site is telling you — and no amount of further mapping is going to fix that. Go back to site and look harder.