A lead source — a partner site, a comparison platform, a form, a local network — is not a fixed object whose quality could be judged once and for all. It's a living flow that changes week by week: the traffic feeding it changes, the keywords that bring visitors move around, the season alters what customers are looking for, and the source's operator regularly tweaks its own settings. Judging a source on a single batch of requests is like photographing a river: the image is sharp, but it says nothing about tomorrow's flow. A request can look flawless on a Monday and still come from a source that, the following week, starts sending unreachable contacts one after another.
That's why a serious marketplace doesn't measure a source's quality at a single point in time, but its trajectory over time. The question isn't "is this request good?" but "does this source produce, week after week, requests that live up to their promise?". This dossier explains how leads-qualifie.ch measures the quality of a source over time, whatever the category: why quality only reads over the long run, which indicators describe a flow, how traceability ties each request to its origin, how the evolution is tracked while telling noise from a genuine drift, and what the measurement concretely triggers on the source's side as well as on the scoring system's side.
Why a source's quality is measured over time, not at a single moment
A lead source is not a fixed object whose quality could be settled once and for all. It's a living flow that shifts from week to week: the traffic feeding it changes, the keywords that bring visitors move around, the season alters what customers are looking for, and the source's operator regularly tweaks its own settings. Judging a source on a single batch of requests is like photographing a river: the image is sharp, but it says nothing about tomorrow's flow. A request can look flawless on a Monday and still come from a source that, the following week, starts sending unreachable contacts one after another.
That's why a serious marketplace doesn't measure a source's quality at a single point in time, but its trajectory over time. The question isn't "is this request good?" but "does this source produce, week after week, requests that live up to their promise?". This shift in perspective changes everything: it lets you tell a passing dip — a spike of poor quality quickly corrected — from a deep-seated decline that warrants rethinking the source's place in the system. leads-qualifie.ch reasons this way across its entire catalogue, whatever the category: quality is an observed trend, not a snapshot.
The indicators that describe a source's quality
Measuring a source first means knowing what to measure. Several indicators, cross-referenced against one another, describe the quality of a flow. Reachability measures the share of contacts you actually manage to reach — a number that rings out or an address that bounces drags this indicator down. The valid-consent rate checks that each customer genuinely agreed to be contacted again, in a traceable way. The duplicate rate flags requests already passed on elsewhere or already present in the system. Field completeness measures whether the expected information — need, area, availability — is actually filled in. Freshness, finally, measures the delay between the customer expressing the need and the request being passed on.
None of these indicators taken in isolation is enough: a source can show excellent reachability but a high duplicate rate, or very complete records but needs poorly matched to the stated category. What counts is the cross-reference, and above all its evolution. To these automatic measures are added signals that are slower to surface but very telling: the conversion-into-appointment rate observed by the receiving companies, and the complaints they file. These field returns, aggregated by source, close the loop: they confront a request's declared quality with what it actually produced once in a professional's hands.
Traceability: tying each request to its source
You can only measure a source if you know, for each request, where it comes from. Traceability is therefore the technical bedrock of any quality measurement over time. On the marketplace, each incoming request is tagged at its origin with a stable source identifier, kept throughout its journey: validation, scoring, distribution, then company feedback. Without this anchor, the indicators described above would be nothing but global averages, unable to point to the source responsible for a decline — you'd know quality was slipping, without knowing where to act.
This traceability makes it possible to reason in cohorts: you group all the requests from a single source over a given period, and follow that group over time rather than blending every origin into one figure. It also makes the measurement verifiable: a source that contests its rating can be confronted with the precise history of the requests it passed on, with their reachability and their attached complaints. This traceable link between a request and its origin is what sets a continuous assessment apart from a mere impression: the marketplace doesn't say "this source seems weaker", it shows, figures and history in hand, how its flow has evolved since it became active.
Tracking the evolution: cohorts, trends and drift
Once requests are tied to their sources, measurement becomes a reading of trends rather than a sum of isolated cases. Each indicator is observed over rolling windows — the last few weeks compared with the previous ones — to separate noise from signal. A source whose reachability wobbles slightly from one week to the next stays stable; a source whose reachability keeps receding across several consecutive readings describes a drift, even if no single batch looks alarming on its own. It's this reading over time that makes it possible to act early, before a slow decline becomes a problem visible to the receiving companies.
Tracking over time also knows how to account for context. Some variations are seasonal and perfectly normal: customer intent changes with the time of year, and a drop in volume is not a drop in quality. The challenge is to separate what stems from an expected cycle from what stems from a genuine drift of the source. A source is therefore compared to its own past trajectory and to comparable sources in the same category, rather than to an abstract norm. A recent source is, moreover, read with caution: a few good batches aren't enough to establish a reputation — a long enough trajectory is needed for the measurement to be reliable and not overreact to the first deviation.
What the measurement triggers: scoring, downgrading, feedback to the source
Measuring is pointless if the measurement triggers no action. On the marketplace, a source's observed quality over time feeds directly into the scoring system that sorts requests before they're distributed. A source whose trajectory deteriorates sees its flow progressively downgraded: its requests are examined more demandingly, distributed less widely, or even held while the origin of the problem is understood. Conversely, a source that holds its indicators over time gains priority — the measurement rewards consistency, not one-off volume alone.
The measurement also triggers a dialogue with the source itself. When a drift is detected, the operator doesn't simply cut the flow: it relays the information to the source's operator, who can then identify the cause — a shift in traffic, a poorly configured form, a badly targeted campaign — and correct it. This feedback loop is what sets a continuous assessment apart from a blunt sanction: the goal isn't to eliminate sources, but to maintain a consistent level of quality across the whole catalogue over time. The other dossiers detail the neighbouring mechanisms: the scoring of an individual request, the legal framework of consent, and the method for comparing providers against one another.