Cut the ads: profit rises half the time.
Same as not cutting
314 consumer companies in five industries · 628 company-years · three years of audit reports
When times are hard, advertising is the first thing to go. It is the fastest valve to close on an income statement, and closing it is supposed to bring profit back. We took three years of audit reports for 314 Korean consumer companies, picked the 104 years in which a company cut its advertising by more than 20%, and checked whether operating profit rose that year. It did half the time (55%). In years when advertising was left alone it was 48%, and in years it was raised by more than 20%, 50%. The three are the same. Advertising spend was not a valve that controls profit. It behaved more like a thermometer that shows how the business is doing.
01What was measured
For externally audited companies in five industries (cosmetics, food, beverages, health functional foods and e-commerce), we read three years of sales, advertising and operating profit straight from each audit report. Operating profit is the money the business itself earned. Operating profit per 100 of sales is the operating margin. Advertising means the advertising expense line only.
We dropped companies whose industry code matched but whose business did not, such as drug makers and paint makers. We also dropped any company with sales under 3 billion won, or zero advertising, in any of the three years, because a change from zero cannot be expressed as a percentage. That leaves 314 companies, each with two changes (2024 and 2025), so 628 cases.
The 628 cases fall into three groups. A cut year is one where advertising fell by more than 20% and by more than 100 million won (104 cases). A held year is one where advertising moved by less than 20% (239). A raised year is one where it rose by more than 20% and by more than 100 million won (199).
02Cut it, and it is still half
| Cases | Operating profit rose | Operating margin change (median) | Sales change (median) | |
|---|---|---|---|---|
| Ads cut 20%+ | 104 | 55% | −0.19p | −5.2% |
| Ads held | 239 | 48% | −0.33p | +0.6% |
| Ads raised 20%+ | 199 | 50% | −1.05p | +7.1% |
Figure 1 · Share whose operating profit rose: 55% in cut years, 48% in held years, 50% in raised years. The thin lines show how far each number could wobble, and they overlap
In years when advertising was cut by more than 20%, operating profit rose at 55% of companies. In held years it was 48%. That looks like a 7-point gap, but 104 cases cannot tell a gap that size from chance. That is what the overlapping lines in the figure mean. So this piece claims only what was observed: even after cutting, half the time profit did not come back.
Raised years act as a mirror. When advertising goes up, sales rise 7.1% and the margin falls 1.05 points, a change too large to be chance. In cut years the margin actually slipped 0.19 points. Raising advertising visibly costs margin; cutting it does not visibly bring margin back.
In years advertising was cut, profit rose at half the companies. The same as in years it was not cut.
03Advertising is small; the business is big
Why? Taking only the years when advertising moved by more than 20%, we put two things side by side: how much advertising moved as a share of sales, and how much operating profit moved as a share of sales. Advertising moved 1.72%; operating profit moved 4.26%. Even in years of big advertising swings, profit moved 2.5 times as much. In three of four (72%), the rest of the business moved more than the advertising did.
Figure 2 · In years advertising moved 20%+, ad spend moved by 1.72% of sales and operating profit by 4.26%
Think of plumbing. To save water by closing a valve, the valve has to sit on a thick pipe. In these industries advertising is a thin pipe. The thick pipe that decides profit is the business outside advertising.
Split the 57 cut years in which profit rose and it gets clearer. In 38 of them, the business itself improved even with advertising set aside. Only in 19 did the cut make up for a weaker business and protect profit; that is 18% of the 104. Among the 66 cases where the business got worse and advertising was cut, profit rose 29% of the time. Barely one in four.
Figure 3 · The 104 cut years: 38 where the business improved, 19 where the cut held profit up, 47 where profit did not rise
Advertising moves 1.7% of sales; operating profit moves 4.3%. What sets the direction of profit is the business, not the advertising.
04You do not keep what you cut
If every won saved on advertising stayed as profit, cutting advertising by 1% of sales would raise the margin by 1 full point. It actually raised it by 0.62 points, comparing companies whose sales moved by the same amount. Leave the accompanying fall in sales in, and it shrinks to 0.22 points, because sales tend to fall in the same years advertising is cut.
Figure 4 · Cut ad spend by 1% of sales: if every won stayed, the margin would rise 1.00p; in fact it rose 0.62p
Where the remaining 0.38 points went, this data cannot say. The filings here do not include the cost-of-sales detail needed to follow it.
05Only big cuts move anything, and the price is sales
We split the cuts by size, as a share of sales.
| Ad cut ÷ sales | Cases | Operating margin change (median) | Sales change (median) | Operating profit rose |
|---|---|---|---|---|
| Under 1% | 42 | −1.00p | −3.8% | 45% |
| 1–2% | 15 | −0.27p | −3.9% | 53% |
| 2% and up | 47 | +2.11p | −5.8% | 64% |
| Of which 5% and up | 24 | +4.67p | −12.4% | 63% |
Figure 5 · Margin change and sales change by size of cut. Companies that cut more than 5% of sales gave up 12.4% of sales
The 42 companies that trimmed less than 1% of sales saw nothing happen; the margin actually fell 1.00 points. Among the 47 that cut more than 2% of sales, the margin rose 2.11 points, but even that is not large enough to separate from chance. What is certain is the price: sales fell 5.8%. The 24 that cut more than 5% raised the margin 4.67 points but gave up 12.4% of sales. The valve only closes when you turn it hard, and turned hard, the pipe starts leaking somewhere else.
Follow the 21 companies that cut more than 2% in 2024 into the next year: they left advertising where it was, sales grew 3.2%, and the margin rose another 3.99 points. Companies that cut and held kept the margin into a second year. With only 21, that is a direction, not a finding.
The raised side tells the same story by size. Spend 2–5% of sales more and sales rise 11.6% while the margin falls 2.75 points. Spend more than 5% more and sales rise 22.0% with the margin down 1.92 points. Advertising trades sales against margin; it is not a dial for margin alone.
Figure 6 · Sales change and margin change by size of increase. Spending more lifts sales and lowers the margin
Only big cuts move the margin. What is certain is sales. Cut more than 5% and sales fall 12%.
06Ad spend is a thermometer
In years advertising was cut, sales fell at 61% of companies. In held years 46%, in raised years 34%. Companies cut advertising when sales are falling and raise it when they are growing. Before it is a valve management turns, ad spend is a thermometer that reads how the business is doing.
Figure 7 · Share whose sales fell that year: 61% in cut years, 46% in held years, 34% in raised years
So when a set of accounts says "we cut advertising and profit came back," remember the 57 in section 03. In 38 of them, the business recovered and advertising happened to be lower that year; in only 19 did the cut create the profit. The causality also runs the other way. Companies that cut advertising in 2025 had already seen sales fall −0.8% in 2024 (versus +2.2% for those that did not cut). They were cooling before they cut. Because this selection effect cannot be removed, this piece states an observation ("even after cutting, half did not recover"), not a prediction ("cutting does not work").
Before it is a valve management turns, ad spend is a thermometer the business reads.
07So what do you look at
| What you look at | When | What you do |
|---|---|---|
| A margin target built on ad cuts | Every won cut is counted as profit | Multiply by 0.62 and recompute. If the cut is under 2% of sales, take it out of the margin target entirely |
| Ad cut ÷ sales | 5% or more | Put −12% sales in the same plan. A higher margin is not guaranteed |
| "We cut ads and profit came back" | Did the business outside ads improve that year? | If so, it was the business that recovered (38 of 57). Do not build next year's budget on that logic |
| Which ads to cut | Decided as a total | Cutting the total costs 5.8% of sales. Separate the orders that would have come without ads first, and you find what can be cut without giving up sales (post 15) |
| A plan to raise advertising | Increase of 2% of sales or more | Write the −2.75p margin into the plan. It is the price of +11.6% sales |
08What this piece is saying
Post 15 said advertising is a maintenance cost, not a growth cost. Post 16 said profit is not the same as cash. This one says advertising is not a valve that controls profit. All three point at the same place. A company that steers its total ad budget from one line of the income statement does not know which orders the advertising created and which would have come anyway. So when it cuts, half the time profit does not come back, and when it cuts hard, sales leave. The valve you can actually turn sits below the total. Attach ad spend and sales at the level of channel, campaign and order, and the sales that would have come without ads become visible; cuts made there do not give up sales.
Most of our work looks like this. Joining ad spend and orders at the same level comes before the analysis, and it is usually the genuinely hard part.
Joining them so you can see where the valve really is: that is what we do.
Appendix — the doubts we tried, and numbers for checking
Was advertising just moved to another account? If advertising was cut and the same money booked as sales promotion or sales commission, nothing was cut. Of the 55 companies with promotion expense in all three years, advertising plus promotion also fell at 91%; of the 28 with commission as well, 79%. It was not moved; it fell together.
Does the threshold matter? Replacing 20% and 100 million won with 10%, 30%, 50 million or 300 million won, profit rose in 55~58% of cut years, and the business outside advertising moved more than advertising in 67~75%.
Are loss-making companies just recovering? Looking only at companies profitable the year before: 47% in cut years (66 companies), 43% in held years. Same picture. By year, 2024 is 48%, 46% and 45% for cut, held and raised, the same; 2025 is 61%, 49% and 54%. The two years differ (48 vs 61). In four of five industries, cut and held years are within 10 points. Only beverages shows 71 vs 44, on 14 cases.
We first measured it another way and dropped it. The first idea was "of every 100 won cut, how much stayed as profit." That ratio blew up for companies with small cuts: the middle half ranged from −142 won to +185 won and said nothing. So we switched to the share that rose and the margin in points.
For checking. The wobble ranges (Wilson 95% intervals) for the 55%, 48% and 50% in table 2 are [45, 64], [41, 54] and [43, 57]. The raised-year margin change of −1.05p has a range of [−2.11, −0.20], below zero; the cut-year margin change straddles zero. In table 3 the under-1% margin range is [−2.56, +0.57] and the 2%-and-up margin +2.11p has [−0.76, +4.73], both straddling 0. On sales, 2% and up is −5.8% [−16.3, −1.2] and 5% and up is −12.4% [−26.0, −4.3], both below zero. The 5%-and-up margin is +4.67p [−1.18, +6.71]. Median ranges come from 4,000 resamples. The 0.62 in section 04 is a regression coefficient with sales change, industry and year matched and 2% trimmed at each end (628 cases, p<0.001). If every won stayed it would be 1.00, and the coefficient's 95% interval never reaches 1.00. Without matching sales it is 0.22. The 21 companies from 2024 in section 05 had next-year sales +3.2%, margin +3.99p, advertising −1%.
(Note: source — audit reports of externally audited Korean companies on DART, 5 industries: cosmetics, food, beverages, health functional foods, e-commerce; current, prior and second-prior years (FY2023–2025). Excluding companies whose business did not match the industry code and any with sales under 3 billion won or zero advertising in any of the three years leaves 314 companies, 628 company-years. The population is a snapshot frozen on 2026-09-15, minus 5 companies later identified as industry-code contamination. Advertising = the advertising expense line only. Cut year = advertising −20% or lower and by at least 100 million won; raised year = +20% or higher and by at least 100 million won. Mostly separate accounts. No company names. Zero internal data.)
Same method, different industry. The bottleneck differs every time — finding that difference is what this series does.
One industry, one bottleneck, public data only — straight to your inbox. Nothing else, ever.
This is as far as public data can see. Repurchase, incrementality, attribution — the numbers that change decisions live inside your own data, and making them countable is what Lambency does.
caffrey.w.lee@gmail.com