The Belgian agency market, measured on the annual accounts agencies filed themselves with the National Bank. No survey, no panel, no gut feeling. Nobody was asked anything, so nobody could answer what they would have liked to read.
Twelve chapters · free · sign in with your email address
Barometer reading
52 / 100
The share of agencies that did not shrink in the last complete year-on-year comparison, 2023 to 2024. Three years earlier, the same calculation gave 68.
4,242
financial years filed since 2021, by 1,018 agencies
804
active agencies in reference year 2024
5
financial years, four year-on-year comparisons, one reference date: 16 August 2026
0
survey questions. An annual account has no socially desirable answer
The market in brief
Four years of decline
Median growth in the market fell from fourteen per cent to a little over one. Not in a single quarter and not in a single segment, but across the full width, four year-on-year comparisons in a row. The share of agencies shrinking rose from 32.1% to 48.0%, and passes the halfway mark in the provisional final year.
At the same time, the median margin has been recovering since 2023 — but not at the bottom. Those are the two movements the rest of this report rests on.
Growth in gross margin per agency, per year-on-year comparison
The line is the median agency. The violet band is the middle half of the market: from the first to the third quartile.
medianmiddle half of the market
Base: 2,549 valid growth pairs across four year-on-year comparisons. The last comparison is provisional: financial year 2025 is 78.4% filed, and late filers are rarely the agencies coming off their best year.
EBITDA margin on gross margin, five financial years
Watch the lower edge of the band: the first quartile keeps falling while the median above it recovers.
Between 2021 and 2024 the third quartile lost 6.7% of its margin, the first quartile 45.5%. The bottom of the market lost nearly half its profitability while the top barely moved.
Why this concerns everyone
A market that stalls is not only the agencies' problem
Agencies rarely stand alone in their own figures. What an agency keeps determines whether it can invest in its people, in its work, and in what it can deliver to you next year. Four parties carry the outcome of this series, each for a different reason.
Advertisers
Your agency is a supplier you are committed to for one to three years. One in seven runs at a loss, and an agency without a buffer does not carry that risk alone — a pitch you have to run again appears in no quotation.
And more broadly: an agency running on 7% margin does not invest. Not in its people, not in its work, not in what it proposes to you next year.
The people who work there
Median staff turnover is 34.6% of the workforce. A sector that does not grow on balance therefore replaces roughly a third of its people every year — and in a quarter of agencies, more than two in three walk out.
There is no link with margin or revenue behind it. That makes it a feature of the sector, not of the winners or losers within it.
Agency owners
The entire distribution dropped fourteen points, so the same performance produces a different outcome today than it did four years ago. An agency growing at +5% without changing anything now sits well below zero.
At the same time your position is not fixed: the rank correlation between two growth periods is 0.105. The ladder is not getting steeper; people climb it and fall down it.
The sector as a whole
Of the 18.2% rise in staff cost per person, 16.8 percentage points were automatic indexation. Virtually the entire squeeze therefore came from outside the business.
That shifts the conversation. It is not about agencies that became too expensive, but about a sector that failed to pass its cost increase into its rates.
IKAg advises on both sides of that table: advertisers on agency selection, agencies on growth, positioning and sale. That is precisely why this report flatters neither side — it says what the annual accounts say, including where that is uncomfortable.
You select an agency on the work, on the people in the room and on the click. That is how it should be, and this report changes none of it. What it adds is that the annual account is already there: free, public, and readable in ten minutes. And that you are co-responsible for it: that agency's next annual account is written in part by what you pay, how you brief and how you pitch.
85.4%
of Belgian agencies covered their own operations in 2024. The market is healthy by a large majority
26.3%
grew by a fifth or more between 2023 and 2024. In a market that stood still as a whole
4
of the five figures you need are already public at the National Bank. You need nobody's permission
This is not an investigation into your agency
It is a reading instruction for something that already exists, and that is a fundamental difference in how you enter the conversation. You do not arrive with a file, you arrive with a better question.
The thresholds below also measure continuity risk, not quality. An agency on 6% margin can deliver excellent work and think more sharply than its competitors — it simply has less buffer if something goes wrong. So do not use these figures to reject an agency. Use them to know which question to ask.
The share of agencies covering their own operations, five financial years
A positive EBITDA, so before depreciation, financial charges and tax.
More than four in five agencies are profitable, and that is the most important context for the table below: the odds that an agency on your longlist is healthy are high. But the series is falling — from 91.6% in 2021 to 85.4% in 2024 — and that is precisely why it is worth checking rather than assuming.
Five things to ask for
Ask for this
Healthy
Danger zone
EBITDA margin on gross margin
above 22%
below 7%
Gross margin per FTE
above € 76,000
below € 61,000
Three-year gross margin trend
growing or stable
two consecutive years of decline
Equity
positive and growing
negative
Share of largest client
below 25%
above 40%
The thresholds are the median and the first quartile of the market in financial year 2024. Everything between the two columns is the middle half of the market: neither confirmation nor alarm, and by far the most common outcome. An agency on 15% EBITDA margin sits exactly there — below the median, well above the first quartile, and entirely normal.
Four of the five you do not need to ask
The first four are in the annual accounts filed with the National Bank and are freely available to anyone. You need the company number and ten minutes.
This checklist is therefore not an investigation into your agency. It is a reading instruction for something that already exists, and that changes the tone of the conversation.
Only the fifth needs asking
The share of the largest client appears in no annual account. That makes the question doubly useful: the answer says something about the agency's dependency, and the willingness to answer says something about how the next three years will go.
Where those four are already worked out for you
IKAg maintains a factsheet per agency, with exactly these figures alongside the medians from this report — so you immediately see which side of the threshold an agency falls on, instead of a number without a yardstick.
By service, useful when your longlist is not yet settled.
By location — and note: a Brussels address raises revenue per person, not profitability.
By group — the axis most often skipped in a pitch: two agencies facing each other can hang from the same holding.
And if the question is not "is this agency healthy" but "is this a fair price"?
That is a different question, and this report cannot answer it. An annual account tells you what an agency earns per person on average. It says nothing about how much work sits in the quotation in front of you — and that is what you actually want to know when three agencies name three different amounts.
IKAg measures that with its own unit: PAR, where one PAR is roughly two agency days of work. The scope of an assignment is converted into a number of PAR, with the fee alongside. From those measurements: up to eight times the difference between agencies on a single quotation line, and only 1.1 times across the full scope. The difference therefore sits in how the work is distributed, not in the total — and a day rate never shows you that.
Five greens? Then you have a supplier that will probably still exist when your campaign ends. It says no more than that, and that is not nothing.
One red line? That is a conversation topic, not grounds for rejection. Ask about it. An agency that explains and gives context is more informative than one that talks smoothly past it — and that answer tells you more about the collaboration than the figure itself.
Even on the strictest signal in this report, two consecutive years of decline, nearly three in four agencies are still profitable. One bad year followed by a second can just as easily be a large client lost as a structural problem. It is worth asking, and the answer is usually illuminating.
And whatever you decide: from the signature onwards you are co-author of the next annual account. The revenue per person in chapter 7 is the fees advertisers paid: +9.8% between 2021 and 2024, against +18.2% payroll cost per person. Reading before you sign is one half. The other half sits in the fee, the payment terms, the brief, what you ask for outside the scope and how many pitches you have made for free.
For seven chapters this report hunts for the characteristic that separates the winners from the losers. Size turned out not to be it, location neither, cluster only partly. What remains is a decomposition into two terms — and only one of them moves.
The squeezed middle runs from 5 to 40 FTE
When your agency grows, there comes a point where it yields less than it did when it was smaller. The pattern is not a slide but a trough, and its lowest point sits between 15 and 40 FTE — the size agencies reach once they have just solved their first bottlenecks.
By size class, financial year 2024
What you charge per person barely moves. What you pay per person rises with every class.
Between 1–5 and 15–40 FTE, gross margin per person rises by 8.9% and staff cost per person by 28.6%. The gap between those two movements is the trough. Going from 8 to 25 people does not get you out of the trough, it takes you deeper into it.
The squeeze, in two numbers
What an agency takes in per person, and what it pays per person
The violet band is what remains between the two. It is not widening.
+9.8%
what an agency takes in per person, 2021 to 2024
+18.2%
what an agency pays per person, over those same three years
+16.8%
automatic indexation under joint committee 200. There was nothing to decide
What agencies charge per person has fallen seven points behind what the law obliges them to pay per person.
And that is why the lever is on what you charge
Split the agencies into four equal groups by what they take in per person, then look at what they pay per person. The top quartile earns 2.23 times as much as the bottom one and pays 1.36 times as much. What remains therefore differs by a factor of twenty.
Four quartiles by what they take in per person
Financial year 2024, medians per quartile.
takes in per personpays per personkeeps per person
The coefficient of variation of gross margin per FTE is 0.70, that of staff cost per FTE 0.33. Agencies therefore differ more than twice as much in what they charge as in what they pay. And the top quartile does not underpay: it pays the most of the four.
Enter at least your gross margin, your EBITDA and your workforce.
Four routes, and the right one depends on where you are
Route 1
Sell expertise rather than hours
From the third to the top quartile is +40.5% on what you charge and nearly three times more on what you keep. The only route that requires no change of size or structure.
Route 2
Grow without hiring
35.6% of growing agencies added no FTE at all, grew 31.5% at the median, and reach 28.6% margin against 25.0% for those that did hire.
Route 3
Leave the middle by growing through it
Beyond 40 FTE the margin recovers to 17.1% and the loss-making share drops to 8.3%. A merger is often a shorter path than organic growth.
Route 4
Sell the agency
The very position you want to leave in route 3 is attractive to the right buyer. What is your problem is, for him, an addition.
And the good news, which is a measurement and not encouragement
The starting position does not decide. The median grower has 7.1 FTE, the median shrinker 7.8. Not bigger, then, and not more digital or more Brussels-based either. The rank correlation between workforce and growth is 0.003.
And your position does not stick: of the agencies that grew by 10% or more between 2021 and 2022, 38.0% did so again two years later. Of those shrinking then, 41.9% grew again. The rank correlation between those two periods is 0.105 — and the same measurement on a second, independent pair gives 0.104. The ladder is not getting steeper. People climb it and fall down it.
The figures are solid because nobody reported them themselves. That imposes an obligation: you must be able to see how they were arrived at, and where they stop.
The main line
Agencies produce more with fewer people — and it is not outsourcing
This is the pattern the figures in this edition converge on, and it runs through four chapters. If you want one story out of this report, this is it.
More than a third of growing agencies hired nobody. 88 of the 247 agencies that grew by 10% or more added no FTE at all. Their median growth was 31.5% — no lower than that of the agencies that did hire.
They keep more of it: 28.6% EBITDA margin against 25.0% for the growers that added staff.
It is not a fringe phenomenon. At 78.9% of all growing agencies, gross margin rose faster than headcount.
And the work was not pushed to freelancers. That has been measured, below.
The test: did they push the work to freelancers?
The obvious objection is that those agencies outsourced rather than worked more efficiently. In that case item 61 — services and other goods — should rise. That item only appears in the full filing schema, so it is visible for 8 of the 88.
Financial year 2023 to 2024
no FTE added (n=8)
did hire (n=15)
median growth in gross margin
+42.4%
+21.4%
median change in item 61
−14.5%
+13.3%
item 61 as % of gross margin
129.9% → 89.9%
87.4% → 78.7%
The opposite turns out to be true. Five of the eight cut their outsourcing while growing, median −14.5%, while the growers that did hire bought 13.3% more of it. And those eight did not simply refrain from hiring: their workforce fell by 11 to 21 per cent while their gross margin rose by 42.4%. More margin, fewer people, less bought in.
Eight agencies, all large and on the full schema, three of them group entities where item 61 also carries a management fee. The test clears away the main counter-explanation where it is visible; it does not measure the whole market.
The most likely explanation is AI and automation. That is a reading and not a measurement — an annual account has no AI line item, and better rates or more work with the same team fit the same figures. But the counter-explanation you would reach for first, outsourcing, has been tested above and does not hold. What remains is a sector that became 9.8% more productive per person in four years, and where a third of the growth now happens without extra people.
What the figures do not show alongside it is a gap. The dividend lands across the full width of the market rather than with a vanguard — see angle 1. That is the sharpest version of this story: AI makes the sector more productive without designating winners within it.
Five angles the data supports
Each with its figure, and with the sentence that holds it up under scrutiny.
1
The AI dividend is not at the top, it is across the whole market
The figure: gross margin per person rose 9.8% between 2021 and 2024, and the ninth decile rose 9.9% — virtually identical. The ratio between the third and the first quartile has stood at 1.55 or 1.56 for four years and falls to 1.49 in the provisional 2025.
Why it is news: the common expectation is that AI opens a gap between agencies that embrace it and agencies that lag. That is not happening. The productivity gain lands across the full width — the top does not pull away, the bottom does not fall off, and the spread even narrows slightly.
That does not make it a softer story but a more stubborn one: anyone hoping early adoption yields a lasting lead finds no trace of it in four years of figures.
2
The margin squeeze comes from indexation, not from the market
The figure: between 2021 and 2024, what an agency takes in per person rose 9.8%. What it pays per person rose 18.2% — of which 16.8 percentage points were automatic indexation under joint committee 200.
Why it is news: virtually the entire wage increase was imposed. Agencies did not start paying their people more by hiring differently: the bill came from outside and there was nothing to decide. Anyone who does not pass indexation into their rates pays it out of margin.
The indexation percentages are the only figures in this report from an external source: the sector publications, not the annual accounts.
3
The mid-sized agency is starting to disappear
The figure: the share of agencies between 5 and 40 FTE hovered around 56 to 57% for four years and falls to 52.0% in the provisional 2025. A filing effect explains at most one and a half points of that.
Why it is news: that is precisely the class with the lowest margin in the market — 13.3% between 15 and 40 FTE. The middle is not only the least profitable, it is now emptying out.
This is the first movement of significance in this series and rests on a single provisional year. Hence a prediction, not a finding.
4
The sector's largest cluster is its least profitable
The figure: Digital & tech counts 204 agencies, a quarter of the market, and comes last with 16.4% margin — more than five points below the sector. And it is not a pricing problem: the cluster takes in € 75,439 per person, just below average. Its staff cost ratio is 83.0% against 78.5% sector-wide.
Why it is news: the corner of the market considered most future-proof keeps the least. Full-service presents the mirror image: normal costs, the lowest revenue per person of all six.
5
Last year's success predicts almost nothing
The figure: the rank correlation between an agency's growth in two periods two years apart is 0.105. Of the growers, 38.0% managed it again two years later; of the shrinkers, 41.9% came back — a shrinker therefore had a slightly better chance of growing than a grower had of continuing.
Why it is news: this measurement appears twice in the series, on pairs sharing no financial year, and gives 0.105 and 0.104. Four measurements, the same answer. The market re-sorts every year, which makes the bad news in this report less final than it sounds.
Six clusters: revenue per person against what they keep
Bubble size is the number of agencies. The dotted lines are the sector: € 76,097 per FTE and 21.8% margin.
With 204 agencies, Digital & tech is both the largest cluster and the least profitable, at 16.4%. That is a quarter of the market, not a niche. When quoting the 36.8% for Production & events, note: it is a cluster figure, the middle half runs from 17.2% to 57.9%, and the median moves across the series from 40.8% to 30.1% and back to 38.4%. Capturing this cluster in one number means implicitly choosing a year.
What this report cannot do
No transactions
No acquisition price, no deal value, no multiple. The National Bank publishes annual accounts, not sale deeds.
No causal direction
This report measures which figures move together. An annual account contains no causes, and a sequence in time proves none.
No normalised EBITDA
Directors' remuneration appears in 1.4% of filings. For the smallest agencies the margin is therefore an upper bound.
No Wallonia in detail
Fifty active agencies carry one regional total and nothing below it. Quoting that figure means quoting 50 annual accounts, not the Walloon market.
The weakest point, and it should come from us
The list of agencies was compiled by hand. There is no NACE code that lines up neatly with "advertising, communications and digital agency": the closest codes contain printers and call centres, and miss agencies registered under an IT or consultancy code.
The consequence: how complete this list is against the real market cannot be established. There is no reference population to set it against. What can be said: the list was not selected on performance, size or federation membership, and nothing was removed from it on the basis of figures.
Three predictions that can fail
A report that returns every year should commit itself. All three can be checked with the same measurement on the same bases. Where they hold, that will be stated next year. Where they fail, that too.
Prediction 1
The reading will land between 45 and 49
The 2027 edition measures the 2024 to 2025 comparison. It provisionally stands at 49.7, and that figure is flattered. Two historical calibrations put the definitive measurement between 46 and 48.
Prediction 2
The top does not break away
For four years the ratio between third and first quartile of gross margin per FTE has been 1.55 or 1.56. In 2027 it will not exceed 1.60. If that ceiling breaks, a pivot to AI-driven work is the first candidate explanation.
Prediction 3
The squeezed middle empties out
The share of agencies between 5 and 40 FTE hovered around 56 to 57% for four years and falls to 52.0% in the provisional 2025. A filing effect explains at most one and a half points. In 2027 it will stay below 55%.
The figures on one page
Every number with the base it rests on, so you do not have to hunt. The bases differ per measure and cannot be netted against each other.
Figure
What it is
Base
52 out of 100
barometer reading: share that did not shrink
738 pairs 2023→2024
+1.3%
median growth in gross margin
738 pairs
48.0%
share of agencies shrinking
738 pairs
26.3%
grew by 20% or more
738 pairs
21.8%
median EBITDA margin on gross margin
804 active agencies, FY 2024
7.1% and 42.3%
first and third quartile of that margin
804 active agencies
14.6%
ran at a loss (117 agencies)
804 active agencies
8.6%
negative equity
804 active agencies
€ 76,097
median gross margin per FTE
709 agencies with at least 1 FTE
€ 59,962
median staff cost per FTE
709 agencies with at least 1 FTE
+9.8% / +18.2%
revenue and cost per person, 2021→2024
709 agencies with at least 1 FTE
16.8%
compounded indexation JC 200, 2022–2024
external source, not an annual account
13.3%
lowest margin, between 15 and 40 FTE
135 agencies in that class
36.8% / 16.4%
highest and lowest cluster on margin
92 and 204 agencies
34.6%
median staff turnover
731 agencies with staff
0.105
rank correlation between two growth periods
497 agencies with both figures
35.6%
growers that added no FTE
247 with FTE on both sides
€ 190,004
median equity
526 agencies with balance sheet data
804 · 4,242 · 1,018
active agencies, financial years, agencies that filed
reference date 16 August 2026
How not to quote this
Six readings that come naturally and that the data does not support. They are here because all six have been made before.
"The Belgian agency sector is shrinking"
It is the median that passed through zero, and 48.0% shrank. That means the other half grew, a quarter of them by 20% or more. Write "half of agencies shrank", not "the sector is shrinking".
"Event agencies make 36.8% margin"
That is a cluster figure, not an agency type. The middle half of that cluster runs from 17.2% to 57.9%, and the median moves across the series from 40.8% to 30.1% and back to 38.4%. Capturing it in one number means implicitly choosing a year.
Netting columns from one table against each other
Subtracting staff cost per FTE from gross margin per FTE does not give the EBITDA margin beside it. These are medians of separate distributions, on bases of 804, 709 and 738 agencies.
A 2025 figure without the caveat
That financial year is 78.4% filed and late filers are rarely the strongest agencies. Every 2025 figure is flattered; by how much is stated in the footnote below.
"The Walloon agency market does X"
Wallonia counts 50 active agencies in this file. That gives one regional total and nothing below it. Quoting it means quoting 50 filed annual accounts.
A day rate without the assumption
The median runs from € 544 to € 761 depending on how many billable days per year you assume. That assumption shifts the figure almost as much as the difference between a weak and a strong agency. Without it, the number is meaningless.
Quoting, and where to reach us
Reuse is permitted with attribution: "IKAg Agency Barometer 2026, analysis of annual accounts filed with the NBB". For a 2025 figure, carry the caveat, and for the day rate always the assumption about billable days — without it, that number is meaningless.
This report never contains figures for an individual agency; everything is a median, quartile or share.
Every chart can be saved using the button beneath it, as an SVG with the figures inside — scalable, and openable in Illustrator or a browser.
The figures are also available as data, each with its base and with the caveats attached: plain text or JSON. That is the same series as in the table above, meant to be taken over without having to parse the layout — by an editorial system, a spreadsheet or an AI assistant.
Feel free to call for an additional breakdown, a conversation about the method, or to have a figure checked before publication. The full technical justification — item codes, formulas, de-duplication and sensitivity analyses — is available on request.
The barometer reading, and why it lags by a year. The reading is the share of agencies that did not shrink in the last complete year-on-year comparison. This edition stands at 52 because it measures 2023 to 2024: 52.0% grew, 48.0% shrank.
On the provisional 2024 to 2025 comparison, the same measurement gives 49.7. It does not count as a reading — financial year 2025 is 78.4% filed — but it does count as direction. On the first measured comparison the calculation gives 68; that figure was established retrospectively and was never a published reading, but it does represent the distance this market covered in three year-on-year comparisons.
Financial year 2025 is flattered for the same reason: late filers are rarely the agencies coming off their best year. Across two calibration years the median margin was one and a half points too high and the shrinking share two to three points too favourable. Carry that caveat if you quote a 2025 figure.
The full report
Twelve chapters, with the methodology
What you have read here is the summary. The report itself measures the market in nine chapters, and adds three for those with a decision to make. Each chapter states what financial year 2025 already says about it — and where that cannot be measured, it says so.
Free. You sign in with your email address and receive a link in your inbox — no password, no sales call.
1. Growth has stalled
2. The margin recovers, but not at the bottom
3. The squeezed middle runs from 5 to 40 FTE
4. Six clusters, two worlds
5. Brussels is the market of extremes
6. Growth and margin are the same symptom
7. Productivity rises, but modestly
8. What you earn, pay and keep per person
9. The growers' playbook
10. For advertisers checklist
11. Where the growth actually is: four routes
12. The value of your agency
— Outlook: three predictions
— Methodology and caveats
Frequently asked questions
Short and concrete
What is the Agency Barometer?
IKAg's annual sector report on the Belgian agency market. Every figure comes from an annual account an agency filed itself with the National Bank. The 2026 edition rests on 4,242 financial years filed by 1,018 agencies, with 804 active agencies in reference year 2024. No survey, no panel: nobody was asked anything, and nobody could therefore answer what they would have liked to read.
What does a reading of 52 mean?
The share of agencies that did not shrink in the last complete year-on-year comparison. Between 2023 and 2024, 52.0% grew and 48.0% shrank. The reading always measures the most recent complete comparison and therefore lags the publication date by a year. On the provisional 2024 to 2025 comparison the same measurement gives 49.7, and on the first measured comparison 68.
Why is this called a baseline?
The National Bank only supplies structured annual account data for filings from April 2022 onwards; everything before that exists only as PDF. This edition therefore has five financial years and four year-on-year comparisons, not ten. What is fixed here in definitions, thresholds and bases determines the comparability of every subsequent edition — which is why it sits in the methodology and not in a footnote.
Why gross margin and not revenue?
Gross margin is revenue minus goods for resale minus services and other goods: what the agency earns itself, after deducting media buying, production, freelance input and subcontracting. For an agency that passes costs through, revenue says little and gross margin says everything. For the same reason the EBITDA margin is taken on gross margin and not on revenue: that makes a media agency and a creative agency comparable.
Are there figures for individual agencies?
No. Everything is a median, quartile or share. Where an outlier influences the picture, that is stated — as with the Production & events cluster — but no agency is ever named. The fact that the underlying annual accounts are publicly available at the National Bank does not change that.
May I reuse figures?
Yes, with attribution: "IKAg Agency Barometer 2026, analysis of annual accounts filed with the NBB". For a 2025 figure, carry the caveat, and for the day rate the assumption about billable days — without it, the number is meaningless. For press enquiries or additional breakdowns: bart@ikag.be.
