Campaigns, Launches and the Lift Nobody Measured
Why the expected lift from a campaign or launch is almost always higher than the observed one, how to estimate it from your own history, and how to carry a first-time estimate honestly.
The lift a campaign or launch is expected to produce is almost always larger than the lift it produces, and the gap is systematic rather than occasional. Planning staffing on the expected figure produces a predictable pattern of over-resourcing that is rarely attributed to its cause.
The method in “Campaigns, Launches and the Lift Nobody Measured” is only as useful as the definitions and source records behind it. A team exploring a structured route to employee monitoring software with screenshots for employee monitoring software with screenshots can add timely work and project context, with transparent settings, limited access and a documented correction route around every consequential use.
The remedy is not scepticism but measurement: your own history of the last thirty campaigns is a better estimator than anybody's expectation about the next one.
Additional context for “Campaigns, Launches and the Lift Nobody Measured” is available from BLS employment projections. External material is most useful when it prompts a specific validation check rather than replacing local judgement.
Where the expectation comes from
The person running the campaign, usually before the detail is settled, usually from the business case that justified it. A business case figure is a target, and targets are set to be ambitious.
A business case figure is a target. It arrives in the planner's inbox looking like a forecast, and it is treated as one because nothing in the document says otherwise.
It is not dishonest. It is simply a different kind of number, and it arrives in the planner's inbox looking like a forecast.
Measuring your own lift
Join the send calendar to the daily series. For each send, compare the following three days to comparable days without one, controlling for weekday and for anything else in the calendar.
Thirty sends gives a distribution: a median lift, a range, and a sense of which kinds of send move things.
What the distribution usually shows
A median well below the expected figure, a long right tail where two or three sends did very well, and a meaningful number with no detectable effect at all.
The tail is what people remember. The median is what to plan on.
Lift decays with repetition
An audience contacted monthly responds less than one contacted quarterly. Where sends have increased in frequency, the per-send lift will have fallen, and a coefficient estimated three years ago overstates it.
Re-estimating annually catches this; it is a morning's work.
The shape of the lift matters more than the size
A 20% lift spread over five days and a 20% lift concentrated in six hours are different staffing problems entirely. The second one creates a queue; the first one does not.
Most operations know their total lift and not its shape, which is the wrong half to know.
Launches are worse than campaigns
Because they have no precedent, because the expected effect is larger, and because the demand they create is not a spike but a new baseline whose level is unknown.
A launch is the clearest case for two published cases rather than one number.
Carrying a first-time estimate
The sponsor's figure as the high case, something around half of it as the central case, and both labelled with their basis. Say explicitly that the central case is an adjustment applied because first estimates historically run high, and give the evidence from your own record.
Stated that way it is a method, not an insult.
The demand a launch creates that nobody counted
Support contacts, not sales. The business case forecasts orders; the operation receives questions, complaints and returns, which arrive later and last longer.
This is the most consistently missed element of launch planning, and it is estimable from the last launch if anybody recorded it.
Recording what happened
After each campaign or launch: expected lift, observed lift, shape, duration, and what was different. Five fields.
Thirty rows of that file is worth more than any external benchmark, because it describes your audience and your operation.
Feeding it back
To the people who make the estimates, without an argument attached. "The last eight sends of this type produced a median lift of nine per cent" is information most campaign owners are glad to have.
It usually improves their next estimate, which is better than the planner adjusting it afterwards.
The campaign that was cancelled
More common than anybody admits, and the forecast that carried it is then high for a fortnight. A confirmation step in the week before — "is this still going out" — costs a message.
Without it, a cancelled campaign looks exactly like a forecasting failure.
Lift that lands somewhere else
Worth checking rather than assuming: a campaign aimed at sales frequently produces contacts in a different queue entirely, handled by a different team who were never told it was running.
Measuring lift only in the obvious queue understates the total and misdirects the staffing.
The second wave
Many campaigns produce two peaks: the immediate response and a later one when the consequences arrive — deliveries, bills, cancellations.
The second is smaller, later and longer, and it is the one most often missing from the staffing plan because it falls outside the window anybody was watching.
What this adds up to
A forecast that treats promotional activity as a measured driver with a known distribution rather than as a series of announcements. The measurement exists in the data already; what is missing is the join between the calendar and the series, and that is a morning's work.