Italy Instagram Retail Marketplaces Are Running a €57 Sales Ad Machine

The most useful number in this Instagram sales cohort is not the total spend. It is the median ad spend.

Across Italian retail marketplace advertisers on Instagram, the median sales ad spent just €57.01 over the last 90 days. The 75th percentile was €452.08, while the 25th percentile was only €3.25. In other words, this is not a market where most ads are being given enough budget to fully prove themselves. It is a market where many ads are being launched, allowed to touch the system, and then either killed, starved, or left as campaign exhaust.[^ads-spend-percentiles]

That matters because the cohort is not tiny. The data covers 3,167 Instagram sales ads from 7 retail marketplace brands and 74 ad accounts in Italy, with total tracked spend of €1,874,194.55 over the 90-day window.[^ads-cohort-size]

The headline is simple: Italian retail marketplaces are not behaving like a few brands pushing a few major sales campaigns. They are behaving like catalogue operators, affiliates, regional franchise systems, or marketplace media teams whose default motion is proliferation.

The spend curve says “test factory,” not “hero campaign”

Here is the spend structure that should make performance teams pause:

The gap between the 25th percentile and the 75th percentile is the story. A large share of ads are not being used as finished assets. They are being used as probes.

That is not automatically bad. Marketplace advertisers often have legitimate reasons to run many variants:

• product feed permutations
• regional offers
• price drops
• category-level promotions
• seasonal inventory
• creator or partner whitelisting
• retargeting cells
• language and copy tests

But a €57 median ad also creates a management problem. At that spend level, a dashboard can easily confuse “not enough signal” with “bad creative.” Teams then optimize against noise, especially when ads are segmented by product, placement, audience, and account structure.

The result is a familiar marketplace trap: the organization believes it is testing aggressively, but much of the test matrix never receives enough budget to become a real read.

Seven brands, 74 accounts, one fragmentation problem

The account structure is the second signal. This cohort has 74 ad accounts for 7 brands.[^ads-accounts-brands] That does not prove bad governance, but it does suggest that distribution is being managed through a wider operating system than a single clean brand account.

For retail marketplaces, that can be rational. Different accounts may map to countries, sellers, agencies, business units, catalogue feeds, or local promotional teams. The tradeoff is measurement complexity.

When spend is fragmented across accounts and ads, three things get harder:

1. Creative learning compounds slowly. A lesson learned in one account may not transfer cleanly into another account’s naming, targeting, or optimization setup.
2. Budget allocation becomes political. Teams defend pockets of spend instead of reallocating toward the few concepts that are actually scaling.
3. The creative library bloats. Many low-spend variants remain visible in reporting, making it harder to see the active winners.

This is where the median spend number becomes more than a benchmark. It becomes a diagnostic for operating model design.

If a brand has hundreds of live or recently live sales ads, but the median ad only reaches €57.01, the team probably does not have a creative volume problem. It has a prioritization problem.[^ads-spend-percentiles]

The missing CPM is its own warning label

There is a temptation to turn every paid-social benchmark into CPM. This cohort does not support that. The CPM fields are null, with 0 ads carrying CPM coverage in the returned data.[^ads-cpm-null]

That absence matters. It prevents the easy but often misleading conclusion that one set of ads is “cheap” and another is “expensive.” For this dataset, the cleaner read is spend distribution, not media-price efficiency.

That is a better lens for operators anyway. A low CPM does not rescue a messy test architecture. A high CPM does not condemn a sales ad if it is carrying high-intent traffic or retargeting scarce users. Without CPM, the cohort forces attention back to the thing teams can audit immediately: how many ads are being funded enough to matter.

Reach is present, but the cadence signal is weak

There is one reach metric worth using carefully. Among ads with EU reach data, median total EU reach was 33,044.[^ads-eu-reach] That means the median ad is not necessarily invisible in lifetime terms.

But the reach-per-day fields are not useful as a velocity benchmark here. The 25th, median, and 75th percentiles for reach per day are all 0.[^ads-reach-per-day] That can happen when many ads are inactive, paused, archived, or measured across windows that do not align with actual delivery days. It is a warning against overreading daily distribution from this pull.

So the story is not “these ads reached nobody.” The story is subtler: lifetime reach exists, but current or normalized daily delivery is not giving us a reliable comparison layer.

That distinction matters for marketplace teams reviewing old ad libraries. A past ad with reach may still be creatively useful. A current ad with zero daily reach may simply be dormant. Combining those without lifecycle tagging can produce bad creative decisions.

What the winners are probably doing differently

Because the format field returned as “unknown” for all 3,167 ads, this cohort cannot tell us whether Reels, Stories, feed placements, or static catalogue-style units are carrying the spend.[^ads-format] That absence blocks the most common format-versus-format comparison.

But it does not block the strategic read.

In a sales objective marketplace cohort with thousands of ads and a €57.01 median spend, the likely winners are not merely better at making assets. They are better at deciding which assets deserve oxygen.

The practical playbook is straightforward:

• Separate probes from contenders. Do not evaluate €3.25 ads beside €452 ads as if both had the same chance.
• Create a promotion ladder. Define what spend, reach, click, or conversion threshold moves an ad from test to scale.
• Audit account duplication. If the same offer appears across many accounts, centralize the learning even if buying stays distributed.
• Tag lifecycle status. Active, paused, evergreen, seasonal, feed-generated, and partner-funded ads should not live in one undifferentiated report.
• Review by concept, not asset ID. Marketplaces often create many surface variants of the same commercial idea. Roll them up before judging creative demand.

The sharpest benchmark here is not a glamorous one. It is the median spend line.

A €57 Instagram sales ad is not a fully tested growth bet. It is a question asked of the algorithm. Italian retail marketplaces are asking thousands of those questions. The brands that win will be the ones that stop treating every answer as equally meaningful.