The Festival di Sanremo is the last property in Italian media that still behaves like a mass audience. The 2022 edition, the seventy second, ran across five nights at the start of February and averaged more than 11 million viewers with a 58 percent share, according to the Auditel figures published at the time. The final went above 66 percent, peaked at 16,984,000 viewers during the evening, and touched an 81 percent share in the minutes when the winner was announced. For anyone planning media in Italy, those numbers describe a category with one member. Nothing else on the calendar aggregates an audience of that size, for that long, five nights in a row.

The detail that should interest a marketer more than the totals is the age curve. Across the five nights, the share among viewers born between 1997 and 2012 sat at roughly 70 percent, with the 20 to 24 bracket near 73 percent. Italian television spends the other fifty weeks of the year losing that cohort to everything else. For one week, it has all of it back.

An event that size does not stay on the screen it is broadcast on. It spills. For five nights the country watches and types at the same time, and the commentary layer that forms around the broadcast is, in pure reach terms, a second event running in parallel. The difference between the two is ownership. The physical event is ticketed, rights managed and sponsored at a price set by its scarcity. The conversation around it belongs to nobody.

A brand in the window, not on the stage

The subject of these notes was not the rights holder and not a broadcaster. It was a consumer brand investing in the Sanremo window, with no stage presence to leverage and no ticket to sell. That is the ordinary position most brands occupy around a tentpole event: the property itself is either unavailable or priced at a level that makes the arithmetic difficult, and the alternative on offer is a broadcast slot competing against every other brand that reached the same conclusion.

This plan went the other way. Instead of buying into the event, it bought into the six days around it, on the surface where the audience was already talking, with editorial rather than advertising.

One dimension is worth recording because it is the reason the window is strategically different from buying the same reach at another time of year. The distribution used here skewed heavily young and female, with roughly nine accounts in ten under the age of 35 in the audience profile and the single largest bracket between 18 and 24. That is the same cohort the broadcast was pulling back to television at a 70 percent share. The two surfaces were not reaching different people. They were reaching the same people twice, through different doors.

Nineteen posts in six days, with nothing behind them

The output was 19 posts, published from 31 January, the day before the opening night, through to 5 February, the day of the final. That is roughly three pieces a day on a schedule set by the broadcast rather than by a content calendar. The format was reaction content: short, native, assembled from the material the evening itself was generating, and written to be recognised instantly by anybody who had watched the same thing an hour earlier.

Two properties of that format matter commercially. The first is that production cost is close to nothing, because the raw material is the event and the work is the caption. The second, and the more important one, is that none of the 19 posts carried paid distribution. Every account this campaign reached, it reached because a person chose to pass the post along.

The numbers, and how to read them

Across the six days the activity reached 7,693,642 accounts and collected 1,090,360 likes plus 44,435 saves and comments. That is 1,134,795 recorded interactions against the accounts reached, an engagement rate of 14.7 percent, and an average of 404,929 accounts and 57,387 likes per post.

A rate near 15 percent needs a caveat before it needs a compliment. It is calculated on unique accounts reached, not on impressions. A single account can be served the same post more than once, so impressions are the larger number and the rate computed on them would be lower. What follows is therefore the generous version of the calculation, and it should be read that way. Reaction content is also structurally advantaged on this metric, because a post that lands on somebody who watched the same broadcast is asking for a reflex rather than a decision. The figure is real, it is very high, and it is not comparable to the engagement rate of a product campaign.

Bought distribution is flat. Earned distribution is not.

Thirteen of the 19 posts have per post insights available. Together they account for 6,233,668 of the reach, about 81 percent of the total, and they carry the split that explains the entire campaign: for each post, how much of the reach landed on accounts that already followed the publishing profile, and how much landed on accounts that did not.

Read the follower column on its own and it is almost boring. Across all 13 posts, reach from followers stays between 219,368 and 469,232. A spread of 2.1 times between the weakest post and the strongest one is what a stable distribution base looks like: broadly the same people see broadly the same volume of content, whatever the content happens to be.

Now read the other column. Reach from non followers runs from 8,359 to 856,154, a spread of 102 times. Same 13 posts, same week, same profile, same base. Everything separating the median post from the breakout post happened outside the follower list.

The strongest post of the set reached 1,325,386 accounts, of which 856,154, close to two thirds, were people outside the base. Remove that one post and the remaining twelve average 409,024 accounts with 18 percent of reach from non followers. Both facts should be held at once. The twelve are the evidence that the format performed consistently rather than luckily. The one is the evidence that the ceiling of a format like this is not set by the size of the audience you already own.

The save outnumbered the comment 24 to 1

The composition of the interactions is worth a look. Across the 13 documented posts the audience left 35,282 saves and 1,491 comments, a ratio of nearly 24 to 1. Reaction content inside a live window is usually assumed to be comment bait, so the reverse result is informative.

A comment is a public statement and it costs a little social risk. A save is private, and on this kind of content it generally means one of two things: the person intends to send it to somebody, or they want it back later. Both are forwarding behaviours, and forwarding is exactly the mechanism that produced the non follower reach in the chart above. The saves are not a vanity number sitting beside the likes. They are the closest available proxy for the thing that actually distributed the campaign.

What the format costs the brand

There is a trade here and it should be stated plainly. In a reaction format the brand is not the protagonist. The audience is reacting to the event, not to the advertiser, and the harder a post pushes the brand toward the centre, the less it behaves like the thing that earned the reach in the first place. What this shape of campaign buys is association and presence inside a national conversation at a cost per thousand no broadcast window can approach. What it does not buy is a controlled message, a clean intent signal, or a short path to a purchase. A brand that needs those three things from a tentpole week should be spending somewhere else in the window.

The second constraint is operational rather than strategic. Six days, three publications a day, reacting to a live broadcast, cannot be produced in advance and cannot be approved slowly. Most brands lose this format not because they cannot write it, but because the piece that would have worked at 23:40 goes out at 11:00 the following morning, by which point the audience has moved on. The approval chain is the real production constraint, and it has to be settled before the window opens rather than during it.

Sanremo sells its stage, its broadcast and its sponsorships, and all three are priced according to how rare they are. The conversation around them is the one part of the event nobody is selling, and for six days a year it is the largest unclaimed audience in the country.