On a marketplace, a validated customer request doesn't simply "land" at random in a company's inbox: it's attributed through a precise mechanism that decides who receives what, in what order and under what conditions. Attribution is the beating heart of the two-sided model: it's the step where the supply side — captured, scored requests — meets the demand side — companies that have configured an intake profile. Understanding this mechanism, independent of any single category, is understanding why two companies listed in the same sector don't receive the same requests, nor at the same time.
This dossier describes how attribution actually works on leads-qualifie.ch: what it means, precisely, to "attribute" a lead; which eligibility criteria first filter the possible recipients; how the companies still in the running are then ranked against one another — this is where the notion of a priority auction comes in; how the choice between exclusive and shared attribution changes distribution; and finally how every attribution decision is traced and verifiable. The aim isn't to talk returns or cost, but to make legible a mechanism often perceived as a black box.
What it means to "attribute" a lead on a marketplace
Attributing a lead means tying a specific customer request — a purchase intent that's been expressed, verified and scored — to one or more receiving companies cleared to handle it. It's neither a mass broadcast nor a send to the first available company: it's a decision, taken at a given moment, that pairs a request with recipients according to rules known in advance. On a genuine marketplace, that decision is made by the platform itself, based on the criteria each company has entered in its intake profile and the characteristics of the request — its category, geographic zone, freshness and quality score. The company doesn't go "looking" for requests in an open catalogue: the requests come to it, when it matches the criteria.
This difference is structural. In a resold-file model, the same list of contacts goes out to every buyer with no arbitration, each exploiting the file on their own side. In a marketplace model, attribution is individualised request by request: with each newly validated request, the platform assesses which companies are eligible, in what order they should be served, and how many of them at most may receive it. It's this granularity — one decision per request, not a bulk distribution of stock — that makes it possible to apply freshness, cap and rotation rules impossible to hold with a plain database export. Attribution is therefore less a transfer than a calculated match.
The eligibility criteria that filter the possible recipients
Before any ranking, attribution begins with a binary filter: a company is either eligible or it isn't for a given request. The first criterion is category match — a heat-pump installation request can only be attributed to companies active in that category, never to an insurance broker. The second is the geographic zone: each company declares a coverage area, and a request outside that area is set aside from the start, even if the company otherwise performs well. The third is intake capacity: a company that has set a desired monthly volume and already reached it naturally drops out of the queue until the next cycle, which avoids saturating a recipient at the expense of handling quality.
Other, finer filters are added depending on configuration: a preference for exclusive requests only, availability time slots, sub-segments of a category, or account status with respect to the platform's rules. What matters is that these criteria are declarative and symmetrical: they apply the same way to every company in a given category, with no special treatment. After this filtering, the platform no longer holds the full set of registrants but a narrow subset of companies genuinely relevant to this specific request. It's only within this eligible subset that the question of ranking — and of the auction — arises. A poor setting of these criteria translates, very concretely, into an absence of received requests: the company is simply never eligible.
Ranking the eligible companies and the role of the auction
Once the eligible subset is formed, it often has to be separated out: when a shared request may go to three companies but five are eligible, or when an exclusive request can go to only one while several match, the platform ranks the candidates. This ranking combines several signals: the freshness of the company's configuration and its past responsiveness, its track record in handling previous requests, its compliance with the marketplace's rules, and a priority signal the company can itself adjust. It's this last signal that people often call the "auction": not an opaque bidding war won by the highest offer, but a priority setting that indicates how strongly a company wishes to be served first in its category and zone.
An ambiguity must be cleared up here. On a well-run marketplace, the auction does not cancel the quality criteria: a company cannot "buy" a rank if its handling record is poor or if it breaks the rules, because the reliability score weights the ranking just as much as the declared priority. The operator's role is precisely to prevent the auction from becoming the sole factor — otherwise the marketplace would come down to always serving the same companies, at the expense of the average quality perceived by end customers. Final attribution is therefore the product of an arbitrated, multi-factor ranking, where declared priority weighs but does not decide alone. This dossier deliberately mentions no amount: the mechanism is understood independently of any pricing schedule.
Exclusive or shared attribution: how the cap is applied
The attribution mode determines how many companies receive the same request. In exclusive attribution, the request is assigned to a single company: no one else receives it, which removes any competition over calling the customer back but assumes the company is responsive, since it's the only point of contact. In shared attribution, the same request is attributed to a small number of companies — that number being capped in advance and known to all — each informed that it's not the only one contacted. The marketplace applies this cap strictly: once the maximum number of recipients is reached, the request leaves the queue; it is never redistributed beyond the announced cap.
This cap is one of the structural guarantees that set a marketplace apart from a resold file. In a file, nothing stops the seller from handing the same contact details to an unlimited number of buyers without saying so. On a marketplace, the number of recipients is a variable controlled by the operator, traced for each request and enforceable in a dispute. The choice between exclusive and shared is part of each company's configuration and interacts with the ranking described above: since an exclusive request can go to only one company, the separation matters more there than for a shared request where several slots are open. In both cases attribution stays bounded: never beyond what was announced, never outside the eligible subset.
The traceability of every attribution
A serious attribution doesn't merely happen: it leaves a verifiable trace. For each request, the platform keeps the timestamp of capture and of attribution, the category and zone retained, the quality score at the moment of distribution, the exclusive or shared mode, the actual number of recipients and their identity. This traceability answers three distinct needs: letting a company understand why it received — or didn't receive — a given request; letting the operator arbitrate disputes on facts rather than impressions; and documenting compliance with the cap and consent rules should a check occur. Without this memory, attribution would revert to the black box it isn't supposed to be.
Traceability is also what makes the system self-correcting. Because every attribution is recorded, the operator can measure, category by category, whether fresh requests really go out first, whether the sharing cap is respected, whether a source produces unreachable contacts, or whether the ranking unduly favours certain companies. These measurements in turn feed the adjustment of scoring criteria and the auditing of sources, in the loop described by the other pillar dossiers. For a receiving company, this traceability is a practical right: it can ask why a request escaped it and get an answer grounded in recorded elements — freshness, cap reached, out of zone, ranking position — rather than in an after-the-fact justification. It's this verifiability that closes the loop between eligibility, ranking, cap and trust.