Recalibrating a stated-intent forecast: the 2026 World Cup as a complete case
Before the tournament, 8% of French people were considering a subscription. Afterwards, the real audience is known. Confronting the two is the only way to calibrate a stated figure.
An intent survey produces a figure before the event. Audience measurement produces another one after. In between, most plans are signed off, executed, and never confronted with what actually happened. The 2026 World Cup offers a complete case, with both measurements available.
The point is not to note that stated intent gets it wrong. It is to measure the bias, in which direction and by how much, so as to know how to read the next survey.
What was announced
Before the tournament, a WPP Media study of 1,024 people established that 8% of French people were considering a beIN Sports subscription to follow it, and that 36% had the wrong broadcaster in mind. Internationally, a YouGov study put 42% of the global adult population as planning to follow the event.
What was observed
M6 drew 20.2 million viewers for the France-Spain semi-final and 12.2 million for the Spain-Argentina final, at a 59.2% audience share. Across the whole competition, 60 million French people were reached, or 93% of the population.
The two series do not measure the same thing, and that is exactly the point. The first concerns a committing, paid act; the second a free exposure. Bringing them together without saying so produces a false conclusion, in either direction.
8% subscription intent before, 93% of the population reached after: two accurate figures, no possible comparison.
Three reading rules
- Compare like act for like act. A subscription intent is set against subscriptions actually taken out, never against a free audience.
- Treat stated intent as a ceiling. An intention expressed at no immediate cost overstates conversion. The figure bounds what is possible, it does not forecast it.
- Keep the question that was asked. A percentage separated from its wording and its base becomes incomparable by the very next campaign.
Why the gap is worth archiving
A gap measured once is an anecdote. Measured across several competitions with the same method, it becomes a correction coefficient applicable to the next survey. It is the only durable asset a post-event review produces.
That requires fixing, before the competition, what is being forecast and how it will be checked. Decided afterwards, the protocol always adjusts itself to the result.
Continuously observed behaviour does not have that flaw: it depends on no question being asked, and compares from one period to the next without reprocessing.
Sources
- Étude WPP Media (1 024 personnes, 18-70 ans) : 8 % des Français envisagent un abonnement beIN Sports, 36 % se trompent de diffuseur, Sport Business Club, 2026
- Étude YouGov, FIFA World Cup 2026 Global Brand Handbook : 42 % de la population adulte mondiale prévoit de suivre la compétition, Presse Agence, 2026
- Bilan d'audience M6 sur la Coupe du monde 2026 : 20,2 millions de téléspectateurs sur France-Espagne, 12,2 millions sur la finale (59,2 % de part d'audience), 60 millions de Français touchés soit 93 % de la population, The Media Leader, 2026
Frequently asked questions
- What did stated intent say before the 2026 World Cup?
- A WPP Media study of 1,024 people established that 8% of French people were considering a beIN Sports subscription and that 36% had the wrong broadcaster in mind. Internationally, a YouGov study put 42% of the adult population as planning to follow the competition.
- What do the observed figures show?
- 20.2 million viewers for France-Spain, 12.2 million for the final at a 59.2% audience share, and 60 million French people reached across the tournament, or 93% of the population.
- Can those two figures be compared directly?
- No. Intent concerns a paid, committing act; audience concerns free exposure. The comparison has to be made like act for like act, meaning subscription intent against subscriptions actually taken out.
- How can a stated figure be used despite its bias?
- By treating it as a ceiling, keeping the exact question and its base, and archiving the gap measured after each event. Repeated with the same method, that gap becomes a correction coefficient.
