In the early 1990s, Procter & Gamble noticed something odd in its nappy supply chain. Actual demand at the till barely moved week to week, parents buy nappies at a fairly constant rate, but the orders P&G received from distributors swung wildly, and the orders distributors received from retailers swung even more wildly than that. Nothing was wrong with any single link in the chain. The distortion was the chain itself.

This became known as the bullwhip effect: a small movement at one end of a chain amplifies into a much larger one by the time it reaches the other end, the way a flick of the wrist at the handle turns into a violent crack at the tip of a whip. It's one of the best-documented phenomena in supply chain management, and it maps onto B2B pipeline forecasting almost exactly.

What the bullwhip effect actually is

The mechanism is simple once you see it. Each party in the chain doesn't just pass along the order it received, it adjusts it. A retailer sees a slight uptick in sales and orders a bit extra to be safe. The distributor sees that slightly-inflated order and adds its own buffer on top. The manufacturer sees the distributor's already-padded order and pads it further still, because from where the manufacturer sits, it looks like real demand is climbing fast.

None of these adjustments is irrational in isolation. Each one is a sensible response to uncertainty, made by someone who can only see their own small window onto the chain. The problem is what happens when those small, sensible adjustments stack.

The one-sentence distinction

The bullwhip effect isn't caused by anyone behaving badly or forecasting incompetently. It's caused by good-faith caution compounding across a chain nobody can see the whole of at once.

The pipeline version of the same distortion

Swap nappies for leads and the pattern is nearly identical. A modest dip in marketing qualified leads one month, the kind that happens most months for entirely ordinary reasons, travels through a chain of people who each add their own reaction on top of it.

Link in the chain
What they see, and what they do about it
Rep
Notices fewer fresh leads to work this week. Starts flagging deals as "at risk" earlier than usual, just in case.
Sales manager
Sees several reps flagging risk simultaneously. Reads it as a broader slowdown and pads the quarter's forecast down further.
Marketing
Sees the revised forecast land on their desk looking much worse than the original lead dip warranted. Reacts by reshuffling budget toward whatever converted fastest last quarter, regardless of whether that's still the right channel.
Finance
Sees a marketing reallocation and a softened sales forecast arrive in the same week. Reads it as a genuine downturn and freezes discretionary spend across the board.

By the time it reaches finance, a single ordinary month's dip in MQLs has become a spending freeze. Nobody in that chain did anything unreasonable. That's exactly what makes the bullwhip effect hard to catch, there's no single bad decision to point to, only a chain of individually sensible ones that added up to an overreaction.

Why each hand-off adds its own margin of caution

Three things drive the amplification, and all three are present in most B2B revenue organisations by default.

1
Information delay

By the time a signal reaches finance, it's already several weeks old and has been reinterpreted twice. Nobody at that point is looking at the original data, they're looking at someone else's summary of it.

2
Batching

Forecasts and budget reviews happen monthly or quarterly, not continuously. That means small day-to-day fluctuations get bundled into a single number at each review, and the bundling itself exaggerates whatever direction the trend was already leaning.

3
Safety margins stacking

Everyone in the chain builds in their own buffer against being wrong. Individually reasonable, but a 10% safety margin added at four consecutive hand-offs doesn't stay 10%, it compounds.

Three ways to dampen the swing

Supply chain teams have spent decades on this exact problem, and the fixes translate directly.

Key takeaways
  • Share the same raw number across marketing, sales, and finance instead of letting each function report its own adjusted version, the more times a number gets reinterpreted, the more it drifts from reality
  • Shorten the chain of hand-offs between the original signal and the final decision, every extra link is another place a safety margin gets added
  • Agree in advance what size of dip actually warrants a response, so a normal month-to-month fluctuation doesn't trigger the same reaction as a genuine trend
  • Treat a single bad month as a data point, not a verdict, and wait for confirmation across at least two or three cycles before reacting structurally
Stop the signal getting louder every time it changes hands

Our attribution modelling work is built around a single shared source of truth for pipeline data, so marketing, sales, and finance are reacting to the same number instead of three increasingly distorted versions of it.

Explore Attribution Modelling →