UK Estate Agency & Lettings

Estate agency is the only sector in this series that runs two economically incompatible revenue engines inside one branch, staffed by overlapping people and reported as a single business. Sales is a contingent one-off commission paid on completion. Lettings management is a recurring monthly percentage of rent. Averaging them describes neither.

What the sector publishes

• Residential transactions, first five months of 2026449,000Observed
• Year-on-year movement over that period−6.2%Observed
• Sole agency commission range, plus VAT1.0–1.8%Observed
• Lettings full management, share of monthly rent10–15%Observed
• Redress scheme membership, per office per year plus VAT£269.86Observed
• Modelled instructions reaching a completed, billed sale50%Modelled
• National realised commission rate—Not established

The two engines differ on every dimension that matters. Sales cash arrives two to four months after the work and must be entirely re-won each period; lettings cash arrives the same month and recurs for as long as the tenancy runs. A branch reporting £679,000 at 71% recurring is a fundamentally different business from one reporting £427,000 at 21%, and no single margin figure captures the difference — which is why a branch whose profit swings sharply quarter to quarter is usually a branch with a small lettings book rather than a badly run one.

The lettings-led branch is modelled at 22% central normalised EBITDA against the sales-led branch’s 12%, on 1.6 times the revenue, and three mechanisms compound to produce that. The revenue recurs, so the branch starts each month with most of its income already won. There is no fall-through, so nothing is lost after full cost has been paid. And portal cost falls to roughly half the revenue share — a modelled 7% against 14% — because lettings stock turns over far less often and consumes far less listing exposure per pound of revenue.

Estate agency has the best-measured demand denominator in this series and one of the hardest to read. HMRC publishes monthly completions, and stamp duty changes plus a two to four month completion lag have made any individual month meaningless.

The two engines pull in opposite directions on almost every demand axis. A vendor has high timing discretion and high fee sensitivity: moving house is postponable, comparison-shopped, and the fee is a visible four-figure number on a transaction where every other cost is also being scrutinised. A landlord has neither — the property is let or it is empty, and management is bought to avoid work rather than to save money. Composition matters as much as volume: a book weighted toward probate, divorce and relocation holds when discretionary movers pause, and shows up in no volume count.

Two engines, reported here separately because they cannot be added meaningfully without first being understood apart. The sales engine is contingent and re-won every period; the lettings engine recurs and is not.

Fee ranges are advertised conventions, not surveyed data. What is actually charged after negotiation is collected by nobody, so the national realised commission rate is marked Not established. Conversion figures are modelled: no body publishes instruction-to-completion rates.

No-sale-no-fee is presented to vendors as a guarantee of alignment; economically it means the agency carries the entire cost of every failed transaction. Photography, floor plans, an energy certificate, portal exposure for the full listing period, accompanied viewings and weeks of progression all happen before any fee is earned, and half of that spend recovers nothing. The instinctive response — win more instructions — increases the absolute quantity of unrecovered cost in proportion. Winning fewer, better-qualified instructions and converting more of them improves the same profit line from the opposite direction, and costs less to do.

The same sector produces very different economics when conversion, recurring share and portal cost change. Principal time is normalised to a market replacement rate before economic EBITDA, and recurring revenue share is reported alongside margin because the two together describe the business and neither does alone.

Base central model: 140 instructions · 50% modelled completion conversion · £295,000 average price · 1.2% commission · 320 managed units at 12%

Portal cost is modelled as its own line rather than absorbed into marketing spend, and the three reasons together make it the most strategically important cost in the sector. It is large: trade press puts it at up to 13.5% of monthly sales commission, which on the sales-led archetype is a bigger line than premises. It is unavoidable, because an agency absent from the dominant portal is not visible to buyers at all. And it is regressive, so the smallest operators pay several times what the largest pay for an identical product, independently of how well anyone is run.

Instructions the branch can actually service to completion bind before either vendor demand or staff headcount. An instruction that will not sell consumes the same listing cost, viewing hours and progression effort as one that will, and half of them do not.

Agencies almost universally describe fall-through as a feature of the English conveyancing system rather than as something they influence, which is partly right and substantially incomplete. The system creates the exposure — an accepted offer is not binding until exchange — but what the agency controls is how quickly problems in that process are discovered. A chain stalled at a solicitor for three weeks and one stalled for three days present the same way to a branch that is not actively chasing, and are completely different to resolve. Where nobody owns progression, the branch discovers a collapse rather than preventing one.

Estate agency does not usually have a valuation problem. It has a conversion problem at two separate points, and the second one destroys work the branch has already been doing for months.

Ask a prospect for these twelve figures. Most agencies can produce valuations, instructions, completions and units under management. Valuation-to-instruction rate by valuer, fall-through rate with reasons and realised commission against headline are the ones that are missing, and they bracket both conversion losses and the pricing leak.

The modelled waterfall loses twenty-four points before an offer is even accepted and a further twenty-three to fall-through afterwards — roughly equal in size and completely different in character. The first is largely a pricing and qualification loss: instructions that never sell are disproportionately those taken at an over-optimistic valuation, and the agency knew or could have known at the point of instruction. The second is a progression loss and far more painful, because a sale collapsing at week fourteen has consumed listing cost, portal exposure, viewings, negotiation and months of progression effort, and returns nothing at all.

Around 23,000 estate agency businesses operate in the UK, and every trading branch must belong to a redress scheme, so the operator population is unusually knowable. What separates them is not size but recurring share and conversion, and the ladder below is ordered accordingly.

Rungs one to three improve the sales engine. Rung four changes what happens when the sales engine stops, and in a sector where transaction volumes fell 6.2% in the first five months of 2026 that is not a hypothetical distinction. The test is specific and calculable: if sales revenue went to zero, would recurring lettings income cover the fixed cost base? An agency that answers no faces the sector’s characteristic cycle — it cuts staff in the trough, loses the negotiators and progressors it spent years training, and rebuilds into the recovery behind competitors who did not have to.

Modelled effect on normalised EBITDA of a one-standard-step move in each driver, multi-branch regional archetype, widest first. Average sale price ranks last, which is worth noting because it is the variable agencies discuss most.

The sensitivity ordering inverts the sector’s normal conversation: average sale price — the market — is the narrowest driver on the list, and instruction-to-completion conversion is the widest. Price moves slowly and is almost entirely outside the agency’s control; a ten per cent movement in local house prices is a substantial market event that takes a year or more, while five points of completion conversion is achievable within a quarter by one person chasing chains on a weekly cadence. Both halves of the conversion loss are decisions and processes rather than market conditions.

Technology creates value in estate agency only when it moves one of the drivers on the previous page. This is not a software catalogue. Five of the six interventions below operate on instructions the agency has already won and is already paying for.

An agency under pressure almost always responds by trying to win more instructions. It is the sector’s reflex, it is what most marketing sold into the sector promises, and on the modelled economics it is the wrong move: half of instructions produce no fee while consuming full listing, marketing, portal, viewing and negotiation cost, so winning more at unchanged conversion increases unrecovered cost in direct proportion. Five points of completion conversion on 140 existing instructions is worth roughly £25,000 a branch at close to full contribution, and requires no additional marketing, viewings or staff.

The sector model becomes commercially useful at the moment a named agency is scored against it. The eight dimensions below convert everything in this pack into a diagnostic that can be completed in a single working session with the agency’s own data.

The eight dimensions are scored against the agency’s own figures rather than against an interview, and in this sector that distinction matters: principals have a strong intuitive feel for their market and a much weaker one for their own conversion rates. Where a figure cannot be produced the dimension scores one, which is the finding rather than a penalty — an agency that can state completions but not fall-through reasons has told you precisely where the problem is. The output is usually two profiles, one for each engine, because an aggregate score would conceal exactly that.