A peacock's tail is an ad budget: the maths of costly signals

Zahavi's handicap principle and the economics of advertising say the same thing: a signal is believed when only the genuine article can afford to send it. The peacock's own numbers add a twist.

An Indian peacock with its blue neck and long green train
Photo: Senthi Aathavan Senthilverl, CC BY-SA 4.0, via Wikimedia Commons

A peacock's tail and a brand's ad budget are the same kind of thing: a costly signal, believed because a weak sender could not afford to fake it. Zahavi's handicap principle in biology and the advertising-as-signal models of economists Phillip Nelson, Paul Milgrom and John Roberts reach the same equation: a signal separates good from bad when its cost is affordable to the high-quality sender and ruinous to the low-quality one. The peacock's own numbers add the twist marketers should notice: the 320 g train costs a male far less than people assume, and peahens mostly use it as a minimum bar, not a ranking.

Key takeaways
  • Handicap principle: a signal is believed when a weak sender cannot afford to fake it (Zahavi 1975, proved by Grafen 1990).
  • ₹3.9–15 crore is the credible-spend window for the worked D2C launch: below it, anyone could copy; above it, nobody profits.
  • Repeat gap \( r_H - r_L \) sets the window's width; with no repeat purchase, ad spend carries no information.
  • 320 g train, about 3% of daily energy to grow, and no measurable penalty on walking or take-off.
  • Inverted U: spend that looks excessive reads as desperation (Kirmani 1990).

The Indian peafowl, Pavo cristatus, is India's national bird and evolution's most famous puzzle. A long, bright, heavy ornament looks like the opposite of survival of the fittest.

The answer biologists settled on is also the best available theory of why expensive advertising works. This post lays out that shared theory, solves it with numbers for an Indian brand, and then tests how far the analogy holds using the actual peacock data.

320 gmass of a peacock's train, 6.9% of body weight (Askew 2014)
~3%of a male's daily energy spent growing it (Askew 2014)
₹6,199 crHUL's ad and promotion spend in FY25
~9.7%HUL ad spend as a share of its FY25 total income

Why would anyone believe an expensive signal?

Because cost makes cheating unprofitable. When a signal costs more to the weak than to the strong, only the strong send it, and receivers can trust it.

The story

On 3 April 1860, five months after On the Origin of Species came out, Charles Darwin wrote from Down House in Kent to the American botanist Asa Gray. He told Gray that the thought of the eye had once made him go cold all over, but he had got past that stage. What troubled him now were small details of structure. The sight of a feather in a peacock's tail, he wrote, whenever he gazed at it, made him sick.

The feather was a problem because natural selection should strip away anything that costs a bird and gives nothing back. Darwin's own answer, sexual selection by female choice, came in 1871. The sharper answer, that the cost itself is the message, had to wait another century for Zahavi. Every brand manager defending a big campaign is standing on that feather.

Darwin Correspondence Project: Letter to Asa Gray, 3 April 1860

From biology to economics

Amotz Zahavi proposed the handicap principle in 1975: ornaments such as the peacock's train are honest precisely because they are wasteful, since only a male in good condition can carry the waste. Biologists doubted it until Alan Grafen proved in 1990 that it works as an evolutionarily stable strategy, provided the marginal cost of signalling is lower for higher-quality individuals.

Economists had found the same logic independently:

  • Michael Spence: his job-market signalling model explained education as a costly signal.
  • Phillip Nelson, 1974: for experience goods, which you can only judge after buying, the content of an ad matters less than the fact that the firm paid for it.
  • Milgrom and Roberts, 1986: only a firm that expects repeat purchases can earn back a heavy launch campaign, so heavy spending reveals confidence in quality.
A signal separates good from bad when its cost is affordable to the strong sender and ruinous to the weak one.
ElementPeacockBrand
SenderMale peafowlFirm
ReceiverPeahenCustomer, trade partner, investor
Hidden qualityGenetic quality, health, conditionProduct quality, staying power
SignalLong train with many eyespots, energetic displayVisible media spend, celebrity, production value
Why fakes failA weak male cannot grow and carry it without cost to survivalA weak product loses buyers after trial and never recovers the spend
PayoffMatingsRepeat purchases

When does ad spend become a believable signal?

When it falls in the window between what a weak brand could ever earn back and what a strong brand will earn back. Here is the full solution for a simple launch.

Assumptions and variables:

  • A new product is an experience good. A visible campaign costing \( A \) brings \( N \) people to try it once.
  • Each order earns contribution margin \( m \).
  • A high-quality product (type H) gets \( r_H \) repeat orders per trier over its life; a low-quality product (type L) gets \( r_L < r_H \), because people try it once and leave.
  • Customers can see roughly how much is being spent, the "perceived expense" in the research below.
  1. Payoff if believed. A firm of type \( i \) that runs the campaign earns \( \Pi_i = N m (1 + r_i) - A \).
  2. Low type must not want to copy. \( N m (1 + r_L) - A < 0 \Rightarrow A > N m (1 + r_L) \).
  3. High type must still want to spend. \( N m (1 + r_H) - A \ge 0 \Rightarrow A \le N m (1 + r_H) \).
  4. The separating window. \( N m (1 + r_L) < A^* \le N m (1 + r_H) \). Any visible spend in this window is a credible signal.
\[ N\,m\,(1 + r_L) \;<\; A^* \;\le\; N\,m\,(1 + r_H) \]

Plug in an Indian D2C skincare launch (all numbers are my illustrative assumptions): \( N = 1 \) lakh triers, \( m = ₹300 \) contribution per order, \( r_H = 4 \) repeat orders, \( r_L = 0.3 \).

  1. Low type's ceiling. \( 10^5 \times 300 \times 1.3 = ₹3.9 \) crore.

    Calculated in this post with these example inputs (N = 1 lakh triers, m = ₹300 an order and rL = 0.3, all my illustrative assumptions, not facts about a real brand), using the separating-window logic of Milgrom and Roberts (Journal of Political Economy, 1986): 105 × 300 × 1.3 = ₹3.9 crore (1 crore = 107).

  2. High type's ceiling. \( 10^5 \times 300 \times 5 = ₹15 \) crore.

    Calculated in this post with these example inputs (N = 1 lakh triers, m = ₹300 an order and rH = 4, all my illustrative assumptions, not facts about a real brand), using the separating-window logic of Milgrom and Roberts (Journal of Political Economy, 1986): 105 × 300 × 5 = ₹15 crore (1 crore = 107).

  3. Verdict for the brand. A visible campaign of more than ₹3.9 crore and up to ₹15 crore cannot be profitably copied by a weak product, so it tells customers something true. A campaign of ₹1 crore tells them nothing, because anyone can afford it.
₹3.9 cr₹15 crAnyone can affordno informationOnly a good product recovers itcredible signalNobody profitsVisible campaign spend A →
The separating window from the worked example, from N m (1 + r_L) to N m (1 + r_H).

Your credible-signal window

The sensitivity check

Sensitivity is where the insight lives. The window's width is \( N m (r_H - r_L) \), so it depends entirely on the repeat gap. If the weak product also earns 2 repeat orders, the window shrinks to ₹9 to ₹15 crore.

If repeat purchase is irrelevant, as with a one-time purchase like a wedding venue, the window closes and ad spend stops being informative. That is why heavy brand advertising clusters in categories with frequent repeat purchase, such as soaps, snacks, telecom and payments, and why the biggest spenders in India's TV ad rankings are FMCG firms.

A peacock with its train raised and fanned in full display
A peacock in full display. The fan is the visible part of the signal; the shivering, the calls and the positioning in sunlight are the rest of the campaign.Photo: Trey Perry, CC BY 3.0, via Wikimedia Commons

Does the evidence say expensive ads actually persuade?

Yes, up to a point: customers infer quality from how expensive an ad looks, and beyond a level they start to distrust it.

Tim Ambler and Ann Hollier tested the handicap idea on advertising directly, citing Zahavi by name. In their study people rated TV ads online for perceived expense, message, brand familiarity, quality and reliability, and for how likely they were to choose the brand.

Ads that looked more expensive were significantly more persuasive, mostly indirectly, because viewers inferred higher brand quality.

The waste in advertising is the part that worksTim Ambler and E. Ann Hollier · Journal of Advertising Research 44(4) · 2004

Drawing on Zahavi's handicap principle, the authors hypothesised that visible excess in advertising signals "brand fitness". Participants evaluated TV ads online on perceived advertising expense, message, familiarity, perceived quality, reliability and likelihood of choice. Higher perceived expense raised persuasiveness significantly, mainly by raising perceived brand quality rather than directly. The limit: these are stated ratings from an online panel, not purchases, and "perceived expense" is what viewers guessed, not what was spent.

Read the abstract (free)

Where the curve turns over

The curve has a top. Amna Kirmani's 1990 experiments found an inverted-U: perceived ad cost raised brand perceptions until spending looked excessive, after which perceptions fell. Later replications explain the drop as viewers feeling manipulated, one of several inference shortcuts covered in Mindblog's cognitive biases ranked by evidence.

That maps neatly onto biology. A signal works when it reads as cheap for the sender; when it reads as a strain, it signals desperation. It is also the reason cheap-looking AI ads cost trust: if production looks effortless, the signal disappears.

How costly is a peacock's tail, really?

Much less than the theory assumed. The train weighs about 320 g, 6.9% of the bird's weight, takes about 3% of daily energy to grow, and measurably costs nothing to walk or take off with.

Graham Askew filmed five Indian peafowl taking off in 3D, with and without their trains. The power used in the first two wingbeats was statistically the same, and while the detached train doubled parasite drag in a wind tunnel, that drag was only about 0.1% of the total aerodynamic power (Askew, 2014).

A University of Manchester team then put peacocks on a treadmill with a respirometry mask, during the breeding season with a full 1.5 m train and after the train had moulted.

The peacock train does not handicap cursorial locomotor performanceThavarajah, Tickle, Nudds and Codd · Scientific Reports 6:36512 · 2016

Ten peacocks (mean 4.76 kg) were tested walking and running at 0.5 to 1 m/s with full trains in the breeding season, and nine (4.58 kg) after moulting. The fitted cost-of-transport curves differed by a constant 1.02 J per kg per metre, and it was the birds with full trains that used less energy, about 12 to 15% less across the speeds tested. Stride kinematics did not change. The authors suggest males may be adapted to lower costs exactly when display demands are highest. The limit: small samples of captive birds, and the comparison is across seasons, so other seasonal changes could contribute.

Read the paper (free)
Peacock cost of transport with and without the train
See the data as a table
Full train (breeding season) (J/kg/m)Rudimentary train (after moult) (J/kg/m)
0.5 m/s6.257.27
0.64 m/s5.686.7
0.75 m/s5.696.71
0.88 m/s6.247.26
1.0 m/s7.298.31

Source: Calculated from the fitted equations in Thavarajah et al., Scientific Reports 2016 · Lower is cheaper. The full-train birds were cheaper at every speed.

Cost gap, not cost size

This is not a refutation of costly signalling; it is a refinement that Grafen's maths already allowed. What makes a signal honest is the difference in cost between strong and weak senders, not the absolute cost.

A healthy male with a well-adapted body carries 320 g almost for free; a sick or underfed male cannot grow a full train of perfect eyespots in the first place. The brand equivalent: a strong brand pays for a big campaign from its repeat revenue, so the spend is cheap relative to its future; a weak brand pays from its last round of funding, so the same spend is fatal.

Strong sender: healthy bird, repeat buyersWeak sender: same signal, steeper costSize of signal (train, ad spend)Cost
Grafen's condition: what keeps a signal honest is the gap between the two cost curves, not how high either one is.
\[ \text{Honest signal} \iff \frac{\partial C(s,q)}{\partial s}\ \text{falls as quality}\ q\ \text{rises} \]

Put the two budgets side by side. Either way, the signal is a single-digit share of the budget: sustainable for the strong, a stretch for the weak.

BudgetSignal spendTotal budgetShare
Peacock (Askew 2014)Energy to grow the trainDaily energy during moult~3%
Hindustan Unilever, FY25₹6,199 crore on advertising and promotion₹64,138 crore total income~9.7%
An Indian peacock standing with its long train trailing behind it
Folded, the 1.5 m train trails behind the bird. Askew's and Thavarajah's measurements found it does not slow take-off or make walking more expensive.Photo: Moheen Reeyad, CC BY-SA 4.0, via Wikimedia Commons

Do peahens even look at the tail?

Mostly as a pass-fail check. Peahens reject males whose trains are badly damaged, but among males with normal trains they choose on other cues.

Close-up of iridescent peacock train feathers with their eyespots
Eyespots up close. Peahens appear to check that enough of them are there, then judge the male on display behaviour and other cues.Photo: Wilfredor, CC0, via Wikimedia Commons

The evidence is genuinely split, and it moved in three steps:

  1. Petrie et al., 1991. From a park population in England, peahens preferred males with more eyespots.
  2. Takahashi et al., 2008. Following a feral population in Japan for seven years, they found no link between mating success and train length, eyespot number or eyespot symmetry, and titled the paper Peahens do not prefer peacocks with more elaborate trains.
  3. Dakin and Montgomerie, 2011. Natural variation in eyespot number did not predict mating success, but males whose trains were clipped to remove many eyespots did worse.

Females also weigh display behaviour; Takahashi found that successful males shivered their trains more and longer.

What the threshold means for brands

For marketers this is the most useful part of the analogy. Visible brand presence works as a threshold: below some minimum of share of voice, production quality and availability, a brand is ruled out; above it, buyers choose on product, price, word of mouth and experience.

That is the balance argued in brand versus performance marketing: brand spend gets you into the consideration set, and the funnel decides the rest. The same threshold logic shows up in India's most valuable brands, which nearly all clear a high bar of visibility, then win on distribution and experience.

So, is a peacock's tail an ad budget?

Verdict: the link holds, with one correction

Both are costly signals governed by the same condition, proven by Grafen in biology and by Milgrom and Roberts in economics: spend is credible only in the window \( N m (1 + r_L) < A \le N m (1 + r_H) \), about ₹3.9 to ₹15 crore in the worked launch example.

The correction comes from the birds: the 320 g train costs a healthy male almost nothing to carry, and peahens use it mainly to rule out damaged males. Honest signals need a cost gap, not a huge cost, and they work best as a threshold. Spend enough to be clearly in the window; spending beyond it reads as strain, on Kirmani's inverted U.

The rule of thumb that follows from the algebra: before arguing about the size of a brand budget, measure your repeat rate. If your product earns repeat orders and a weak competitor's does not, visible spend is a signal you can afford and they cannot. If it does not, the money buys attention but not belief.

What if next?

Questions people ask

What is the handicap principle?

Amotz Zahavi's 1975 idea that a signal is honest because it is costly: only a high-quality individual can afford the waste. Alan Grafen proved in 1990 that it is evolutionarily stable when the marginal cost of signalling falls as quality rises.

Why do companies spend on ads that say nothing about the product?

Because for products you can only judge after use, the spend itself carries information. Nelson (1974) and Milgrom and Roberts (1986) showed that only firms expecting repeat purchases can earn back a heavy campaign, so the campaign signals confidence in quality.

Is the peacock's tail a burden?

Less than once thought. It weighs about 320 g, 6.9% of body weight, and studies found no penalty on take-off power and no extra cost of walking or running; growing it uses about 3% of daily energy.

Can a brand spend too much on advertising?

Yes. Kirmani's 1990 experiments found an inverted-U relationship: perceived ad expense improves brand perceptions up to a point, and excessive spend is read as manipulative or desperate.

Sources

  1. Thavarajah et al., Scientific Reports: The peacock train does not handicap cursorial locomotor performance (2016; source for body masses, the 1.5 m train and cost-of-transport equations)
  2. Askew, Journal of Experimental Biology: The elaborate plumage in peacocks is not such a drag (2014; source for the 320 g train, 6.9% of body weight, take-off power and drag)
  3. EurekAlert, University of Leeds: Peacock's train is not such a drag (Sept 2014; source for the ~3% daily energy cost of growing the train)
  4. Ambler and Hollier, Journal of Advertising Research: The waste in advertising is the part that works (2004; source for perceived expense raising persuasiveness)
  5. Kirmani, Journal of Consumer Research: The effect of perceived advertising costs on brand perceptions (1990; source for the inverted-U)
  6. Takahashi et al., Animal Behaviour: Peahens do not prefer peacocks with more elaborate trains (course summary) (2008; source for the seven-year Japanese study and shivering result)
  7. ScienceDaily: Peahens prefer peacocks displaying more eyespots, but rarely (Dakin and Montgomerie) (2011; source for the clipping result and other courtship cues)
  8. Lake Forest College review of peacock courtship research (source for Petrie et al. 1991, Loyau et al. 2005 and the state of the debate)
  9. BestMediaInfo: HUL's ad spend slips 4.5% to ₹6,199 crore in FY2025 (Apr 2025; source for HUL ad spend and total income)
  10. Darwin Correspondence Project: Charles Darwin to Asa Gray, 3 April 1860 (source for the story: the eye and the peacock feather)
  11. Wikipedia: Handicap principle (source for Zahavi 1975, Grafen 1990 and the link to Spence's signalling model)
  12. Milgrom and Roberts, Journal of Political Economy: Price and advertising signals of product quality (1986; source for the repeat-purchase signalling model building on Nelson)
costly signallinghandicap principleZahaviadvertising as a signalIndian peafowlbrand buildingseparating equilibrium
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Marketblog.

Keep reading in Marketblog

From the other blogs