Should I use the median or the average price per square foot?

Use the median when the sample is small, and say which one you used. A mean is easily dragged by one large penthouse or one bulk transfer, and small community-level cells are exactly where that happens. A value-weighted rate — total dirhams divided by total square feet — is the better measure once the sample is deep, because it reflects where the money actually moved. Whichever you pick, keep it constant month to month. Switching measures mid-series is the fastest way to manufacture a trend that never happened, and clients notice.

How many transactions do I need before I report a price for a community?

As a working rule, do not print a headline rate for any cell with fewer than roughly ten registered transactions in the period. Below that, report the count and a range instead of a point figure, or widen the window to a rolling three months and label it as such. The alternative is publishing a number that swings twenty per cent next month for no reason other than which units happened to register. Print the sample size beside every figure — n is not a footnote, it is part of the number.

Can I quote days on market from the portals?

You can, but only if you label what it measures. Portal days on market is listing age, not marketing time: relisting resets the clock, refreshes can reset it, and withdrawn or expired listings drop out entirely, so the surviving sample is biased towards stock that is still available. It also describes asking-side behaviour, never completed deals. Report it as "median days live on portal for currently advertised stock" and treat direction of travel as the signal, not the level. Your own brokerage's mandate-to-transfer data is far more honest, if smaller.

What do I do when the newest month contradicts what I told the client last month?

Say so, in the report, in one line. Registration lag means the newest month is always incomplete, so revisions are normal rather than embarrassing — but only if you flagged the figure as provisional when you first published it. Carry a standing "what we said last month, what actually happened" line at the top of every issue. It costs two sentences and it is the single strongest credibility signal in the document, because it proves the report is a record rather than a marketing piece.

How to Produce a Monthly Abu Dhabi Market Report Clients Actually Trust

A credible monthly Abu Dhabi market report is short, sourced and repeatable: median price per square foot and transaction counts by community and bedroom count, drawn from registered transactions, with sample sizes shown, the newest month flagged as provisional, and commentary that describes what happened rather than predicting what happens next.

The rule the whole document rests on: every figure carries its source, its period and its sample size. A number without those three things is an opinion that has learned to type.

What belongs in a monthly report?

Most broker reports fail for the same reason: they are written fresh each month, so the shape changes, and a client cannot compare issue three with issue eleven. Fix the skeleton once and the report becomes a series instead of a set of documents.

Block	What goes in it	Why it earns its place
Header	Period covered, data cut date, source, one-line method	Answers "what am I looking at" before any number appears
Last month's call	What you said, what actually happened	Turns the report into a track record
Headline table	Rate, transaction count, month-on-month and year-on-year change, per community	The only page most clients read
Bedroom split	Same measures by unit type	Community averages hide the product the client owns
Volume note	Counts and direction, with seasonality named	Tells them whether the price move has conviction behind it
Supply note	Launches and expected handovers in the client's cluster	The rent and resale competition they will meet in two years
Their asset	The client's own project, tower or portfolio	The reason they open the email
Method and caveats	Definitions, filters, provisional flags, disclaimer	The page a sceptical advisor reads first

Keep it to two or three pages. A monthly report nobody finishes is worse than a shorter one everybody does, and the discipline of a fixed skeleton is what lets you produce it in an afternoon rather than a week.

Which measures actually mean something?

Very few. Most of what circulates in market commentary is either asking-side data wearing transaction clothes, or a number too fragile to survive a follow-up question.

Measure	The question it answers	How it breaks
Median AED/sqft, by community and bedroom	What did buyers actually pay for this product?	Mix shifts — bigger, newer or better-located units transacting makes the rate "rise" with no appreciation
Transaction count	Does the price move have conviction, and can my client exit?	Seasonality; registration lag on the newest month
Year-on-year change	Is this a trend?	Endpoint choice — measuring from a trough flatters everything
Month-on-month change	Is anything moving right now?	Noise, in almost every cell smaller than a whole community
Ready versus off-plan split	Am I comparing like with like?	New launches and older stock are different products, not the same one appreciating
Asking rents	What is the letting market advertising?	Asking evidence only — never present it as achieved

Two habits follow from that table. First, always pair a price with its volume; a rate that rose on four deals is a rate that rose on four deals. Second, put the mix question in writing before a client asks it — "what changed in the mix?" is the first thing a competent advisor will raise, and the broker who raised it first owns the number afterwards.

Rents deserve their own line of care. Unless you have a body of signed tenancy contracts to work from, what you have is advertised rent, which is a landlord's hope. Label it as such every single time.

How small is too small a sample?

This is where most monthly reports quietly become fiction. Slice an emirate by community, then by bedroom count, then by ready versus off-plan, and you can reach cells of three or four transactions without noticing.

Set a floor and hold it. Below roughly ten registered transactions in the period, do not print a headline rate. Print the count, print a range, or say "too few transactions to report a reliable rate this month" — a sentence that has never once cost a broker a client.
Widen the window instead of lowering the standard. A rolling three-month median for thin sub-communities is honest and stable. Label it clearly so nobody compares it against a single-month figure elsewhere in the document.
Print n beside every figure. If a rate appears without its sample size, a reader cannot tell a market from a rounding error.
Round like you mean it. Quoting a thin-cell rate to the dirham implies a precision the sample cannot support. Round to the nearest ten, or to a band.
Hunt the distorters before you publish. Bulk transfers within one project, related-party transfers, court-ordered sales and unusually large plots will each move a small cell on their own. Flag or exclude them, and say which you did.
Treat the newest month as provisional. Registration happens days to weeks after agreement, so the most recent period is always short of its final count. Read trends on settled months.

If you take one thing from this guide: a suppressed figure costs you nothing, and a fragile figure that reverses next month costs you the client's confidence in every other number on the page.

Why do days on market and yield need caveats?

Both measures are genuinely useful and both are routinely misreported.

Days on market. Portal DOM measures how long a listing has been live, not how long a property took to sell. Relisting resets it, some refreshes reset it, duplicate listings fragment it, and withdrawn stock disappears from the sample entirely — which biases what remains towards the properties that have not sold. Report it as "median days live on portal for currently advertised stock", use direction of travel rather than the absolute level, and put your own brokerage's mandate-signed-to-transfer figures beside it. Your internal number is smaller but it measures the thing the client actually asked about.

Yield. A yield built from advertised rent over a registered sale price is two different species divided by each other. If you publish one, publish it as an indicative gross range rather than a decimal, and show the deductions that turn gross into net: service charges, vacancy allowance, letting and management fees, and maintenance. It is also worth noting for context that Abu Dhabi's municipality housing fee, typically around five per cent of annual rent, sits on the tenant side of the ledger and therefore shapes what a tenant will pay in total, even though it never reaches the landlord. Any figures of this kind are indicative and for information only — nothing in a market report is investment, legal or tax advice, and that sentence belongs in your method page verbatim.

How do you write commentary that ages well?

The commentary is the part clients quote back to you, sometimes a year later. Write it so that being quoted back is a good thing.

Describe, don't forecast. "Transaction volumes in the community fell for a second consecutive month while the median rate held" ages perfectly. "The market is set to accelerate into Q4" ages like milk.
Date-stamp every claim. Every sentence should be readable in eighteen months without the reader needing to know when it was written.
Attribute causes only when you can name them. A handover, a launch, a school opening, a road connection — name it. Otherwise write "we cannot yet explain this move" and mean it.
Use the three-sentence paragraph. What the number is. What changed since last period. What would change your reading of it. That third sentence is what separates analysis from a caption.
Give a review date instead of a prediction. "We will know whether this is a trend when the next two settled months are in" is both honest and a reason for the client to open the following issue.
Ban the vocabulary that dates you. Record, unprecedented, booming, soft, poised, red-hot. None of them survive contact with a revised figure.
Keep the archive. Never quietly edit a published issue. Correct it in the next one, in the standing "last month's call" line at the top.

The strongest sentence available to a market analyst is "I was wrong last month, and here is the corrected number." It is also the one competitors will not write.

What does the production routine look like?

Consistency beats depth. A two-page report that arrives on the second working day of every month builds more authority than a twenty-page one that arrives whenever there is time.

Pull (half a day). Same query, same communities, same measures, every month. If you are building the pull yourself, freeze the definitions in a document so the series stays comparable when you change tools — this is the discipline behind the community-level series Knownable publishes, and it is worth copying whether or not you use anyone's platform.
Quality check (one hour). Scan for cells below your sample floor, for single transactions large enough to move a median, for any community where the count changed by more than half, and for last month's provisional figures that have now settled.
Write (two hours). Headline table first, commentary second. Writing the commentary before the table is how forecasts sneak in.
Personalise (thirty minutes per key client). One paragraph on their tower, their project or their portfolio, appended to the standard document. This is the only part of the report that gets read twice.
Send on a fixed date. Same day each month, same subject-line format, same file name convention. Predictability is the product.

Then hold two figures on your own dashboard: how many recipients open it, and how many reply. A monthly report with a rising reply rate is a pipeline; one with a flat open rate is a newsletter, and a newsletter is not why you did the work.

The bottom line

Credibility in monthly reporting comes from restraint, not volume. Fix the skeleton so the report becomes a comparable series. Report medians and counts together, per bedroom, per community, and refuse to print a rate the sample cannot carry. Caveat days on market and yield in the same breath you publish them. Write commentary that describes rather than predicts, and open every issue with what you said last month and what actually happened. Do that for a year and you will own something no advertising budget can buy — a client who checks your report before they check the market. If you would like the underlying registry-based series to build on, Knownable's Abu Dhabi platform is in closed beta and the waitlist is open; the Explore tier is free.