How many comparables do I need, and how recent do they have to be?

Three to four genuinely matched sales is the working minimum, and more matters less than closer. Prefer registered sales from the last three to six months; stretch to twelve only when the building is thin, and say out loud that you have stretched. Two comps is a coincidence rather than a pattern. If the only way to reach four is to cross into a different tower, a different tenure or a different bedroom count, you do not have four comps — you have two comps and two distractions, and a bank valuer or an opposing agent will find the seam in minutes.

Can portal asking prices be used in a CMA at all?

Yes, but never as evidence of value. Asks belong in a separate section of the file headed competitive position: what a buyer will see alongside your unit, at what price, and for how long it has been sitting. That tells you where to list and how urgently to move. Value comes only from registered transactions. Mixing the two is the single most common way a CMA fails scrutiny, because an ask is a hope with a photograph attached and a registered sale is a settled fact two parties and a registrar agreed on.

How do I handle a service-charge difference between my unit and the comp?

Capitalise it. Take the annual service-charge difference in dirhams for the whole unit, then divide by a sensible capitalisation rate for that community to get the value effect, and adjust the comp accordingly. A building that costs meaningfully more to run every year is worth less at the same rent, and investor buyers price this even when owner-occupiers do not. Do the arithmetic in front of the seller rather than asserting a percentage — the calculation is short, it is checkable, and it makes every other adjustment in your grid look equally deliberate.

What if there are no recorded sales in the building at all?

Widen deliberately and label every step. Move out one tier at a time — sister towers by the same developer in the same sub-community, then the wider community — and record which tier each comp came from in the table. Then reduce your confidence statement rather than your price: give a wider band, name it as wider, and explain why. A thin cell produces a rate with error bars, not a rate, so never quote it to two decimal places. Cross-check with rental evidence and current asks, and schedule a review for the next registry refresh.

The seller points at a neighbour asking far more. What is the answer?

Ask how long it has been asking that. A price nobody has paid is not a price, and an ask that has sat for months is evidence against itself. Then show the registered sales in the same tower over the same period, which is what buyers and their banks actually respond to. Finish by making the ask useful rather than wrong: it tells you the ceiling of the current competition and it tells you which listing your unit must beat on presentation and on terms. That reframes a rival's optimism into your marketing plan.

How to Run a CMA in Abu Dhabi That Survives Scrutiny

A CMA that survives scrutiny in Abu Dhabi rests on three or four registered sale transactions from the same project, matched on bedrooms and within roughly 25 per cent on size, each adjusted line by line for floor, view, layout and age, then delivered as a price band with its evidence attached rather than a single confident number.

The golden rule of the whole method: every number in the file must be traceable to a registered transaction or to an adjustment you can explain in one sentence. If you cannot say where a figure came from, delete it — an untraceable number does not weaken the CMA, it discredits the rest of it.

What counts as a true comparable in Abu Dhabi?

Most weak valuations fail here, not in the arithmetic. A comparable is not a nearby property; it is a property a rational buyer would have considered as a substitute for yours on the day it sold.

The tier ladder

Work down in order, and never skip a rung because the next one is more flattering.

Tier	What it means	When you may move on
Project	The same tower or the same villa phase	Fewer than three matched sales in the window
Sub-community	Adjacent towers, same developer, same access and amenity set	Project tier is empty or unrepresentative
Community	The wider island or district	Only with an explicit note that you widened

The hard matching rules

Bedrooms match exactly. A large one-bedroom is not a small two-bedroom, whatever the sqft says. Buyers search by bedroom count, so the market prices by bedroom count.
Size within about 25 per cent. Outside that band, the rate per sqft stops being comparable because the fixed elements — one kitchen, one bathroom, one entrance — carry differently across the area.
One area basis throughout. Gross, net and suite areas circulate for the same unit in Abu Dhabi, and mixing them silently is the most expensive clerical error in this business. Fix the basis at the top of the file and convert everything to it.
Tenure matches. Freehold, musataha and usufruct are not cosmetic differences; they change what the buyer is actually acquiring and who can acquire it.
Ready with ready, off-plan with off-plan. A launch price and a resale price answer different questions.

What must be excluded

Your own live listings. Court-mandated and distressed sales. Related-party and family transfers. Bulk and portfolio deals. Partial-share transfers. Anything you cannot identify well enough to defend. Excluding a transaction is a professional act, so record it — a CMA that shows what was thrown out and why is far harder to attack than one that only shows what survived.

Why do recorded transactions beat asking prices?

Because they are different species, and only one of them has a buyer in it. An ask is what a seller hopes for. A registered transaction is what someone actually paid, agreed by two parties and recorded by the registrar. The gap between the two is not noise; it is the negotiation, measured.

Three practical consequences follow.

Asks are contaminated. Duplicate listings, stale prices left up for months, and units that quietly sold long ago all sit in the same portal search. There is no filter for optimism.
Asks tell you competitive position, not value. They belong in the CMA — in a separate section, headed as such — because they show what a buyer will see next to your unit and how long it has been sitting. That is a marketing input, not a valuation input.
Clean prices can still hide incentives. Furniture packages, fee absorptions and post-handover terms sit behind recorded figures the registry cannot see. When you know a comp carried an incentive, annotate it rather than dropping it silently.

One timing discipline matters more than any other. The registry records a transfer some weeks after the handshake that created it, so the newest month always understates itself and fills in later. Read your trend on settled months, label the newest one out loud, and do it before a client finds the gap for you. If a figure looks impossible, query it against the ADREC record rather than repeating it — half the time you have found a thin cell, and half the time you have found the month's real story early.

How should you adjust for floor, view, layout and age?

Adjust the comparable towards the subject, never the other way round, and change one variable at a time so the grid stays readable. The best source for adjustment sizes is your own market: find two sales in the same tower that differ mainly in one attribute, and the difference between them is your paired-sales estimate for that attribute. That is a figure you can defend, because it came from the building in question.

Where no pairing exists, start from a working hypothesis and label it as one.

Attribute	What actually moves value	Indicative starting hypothesis
Floor	Threshold effects, not smooth steps — the level where the podium or the neighbouring tower stops blocking the outlook	Small per-floor increments within a band, with a step change at the view break
View	Usually the largest single swing in a tower: full sea, corniche, mangrove or park versus internal, car park or road	Can reach a double-digit percentage between best and worst aspect in the same stack
Layout	Efficiency and usability — dual aspect, no wasted corridor, usable balcony depth, no intrusive columns, bedroom separation	Low single digits, larger where a plan is genuinely awkward
Age and condition	Building age, but more often fit-out: kitchen, bathrooms, flooring, whether the unit shows	Mid single digits for a full, tasteful renovation over original condition
Service charge	Annual running cost difference, capitalised at a sensible rate for the community	Calculate it; never assert it

These ranges are indicative starting points for structuring the analysis, not published market figures — calibrate every one of them against paired sales in your own building before it reaches a client. Nothing here is investment, legal or tax advice, and a formal valuation for lending or dispute purposes must come from a licensed valuer.

Two guardrails keep the grid honest. First, if the net adjustment on a comparable exceeds roughly 15 to 20 per cent, it is not a comparable — it is a different property you are trying to argue into the table. Drop it. Second, gross adjustments matter as much as net: a comp with a plus 12 and a minus 11 that cancel out is not a close match, it is two large guesses that happened to meet in the middle.

How do you turn adjusted comps into a defensible band?

Weight, do not average. A simple mean lets one small outlier drag the rate; weighting by transaction value — total dirhams divided by total sqft across the comp set — reflects where the money actually moved. Then weight your judgement towards the comps that needed the fewest adjustments, because those are the ones the market has already answered for you.

Deliver four numbers, in this order:

Evidence value — what the adjusted comps say, before strategy.
Recommended list price — evidence plus negotiating room, stated as a decision.
Expected-sale range — the band you genuinely expect to transact in.
Walk-away floor — agreed with the seller in advance, in writing.

Before you present, run three cross-checks. Does the implied gross yield make sense against current rents in the same building? Does the price per bedroom sit sensibly against the tier above and below? And does your number sit coherently among the live asks and their days on market? If all three agree, you have a band. If one disagrees, find out why before the meeting rather than during it. Finally, state your confidence explicitly — a four-comp, same-tower, same-quarter analysis and a widened three-tier reach are both legitimate, but they are not the same product, and pretending otherwise is what gets a CMA torn apart.

How do you present a CMA to a seller who wants more?

Do not argue the number. Ask what would have to be true for it, and let the table answer. The sequence that works is goal, evidence, band, strategy — never price first, because a price that arrives before its evidence is an opinion, and sellers argue with opinions.

Four moves carry most of these meetings.

Show the rejected comps. Naming the three sales you excluded, and why, proves the surviving four were chosen by method rather than convenience. It is the single fastest way to convert a suspicious seller into a participant.
Run the two-column honesty test. What this unit has more of than the comps; what it has less of. Sellers can see the second column exists — they just want to know whether you can.
Make the neighbour's ask useful. The unit asking far more down the corridor has not sold. Put its days on market beside your band and let the contrast do the work.
Offer a strategy, not a verdict. List inside the band for speed, or slightly above with a written, pre-agreed reduction trigger — no offers in 21 days, a defined cut, executed by calendar. The trigger converts a future argument into a signature you already have.

What does a scrutiny-proof CMA file look like?

One page for the client and an appendix for everyone else. The page carries the band, the four numbers, the strategy and the source line: registered transactions, the tiers used, and the date of the data. The appendix carries the full comp table with every adjustment shown as a line item, the exclusions and their reasons, the area basis, the current competing asks with days on market, and the assumptions you have labelled as assumptions.

Then treat it as a living document. Rebuild the comp set the week you go live even if the analysis is a month old, because one new transaction in the tower can move the band. Version and date every rebuild. Keep the file after the deal, too — when a bank valuation comes in short, the party holding a dated, itemised comp table with its exclusions written down is the party who brokers the landing. Pulling those transactions used to mean a morning of manual work per property; the reason Knownable exists is to make that pull take minutes, which changes how often a broker is willing to redo it.

The bottom line

A CMA is not a price. It is an argument, and its strength is entirely in what you can show. Match comps properly and reject the ones that do not qualify. Take value from registered transactions and competitive position from asks, and never let the two blend. Derive adjustments from paired sales in the building rather than from habit, and label every assumption as one. Deliver a band with a floor, not a point. Show the seller the working, including the parts that do not flatter the property. Do that consistently and the CMA stops being the awkward part of the mandate meeting and becomes the reason you win it.

Knownable's Abu Dhabi comp tooling is in closed beta. Explore-tier access is free, and early users come in through the waitlist.