Two buyers write about the same car on the same day. The first announces a 7,000 euro down payment. The second asks whether he can come by on Tuesday at 11. The first looks like a file, the second is a sale. A qualification grid that ranks them the other way round loses money every week.

Qualifying an automotive lead means measuring, inside the conversation itself, how likely a contact is to end in a transaction, then ordering the seller’s time accordingly. It does not rest on what a buyer declares, but on what a buyer commits to. And it happens fast: according to the MIT and InsideSales study (2007), contacting a lead within five minutes rather than after thirty makes it roughly twenty one times more likely to be qualified. Twenty one times, not twenty one percent.

A qualified lead is not a lead that replies

An independent used car dealer running thirty listings gets dozens of messages a week. Most go nowhere. The sorting happens in the evening, between two calls, on a phone, with no grid and no memory from one conversation to the next. The seller follows up on the people he remembers, not on the ones worth calling.

Volume is not the problem. Order is. A listing inbox displays contacts by arrival time, never by intent. Nothing visually separates the buyer comparing three models for fun from the one who has already sold his car and needs a replacement by the weekend.

Stated signals lie, committed signals do not

What gets said

Budget, financing method, purchase timeline, model wanted: the canonical quartet of qualification grids. Its flaw is that it costs nothing. Writing “interested, I have 7,000 euros to put down” commits nothing and proves nothing. The same message often goes to three sellers the same evening.

Precise technical questions, engine size, combined fuel consumption, service history, price against the market index, say something different. They do not signal buying intent, they signal an informed buyer. That is maturity, not commitment, and it deserves an expert answer rather than an immediate call.

What gets committed

A commitment is recognisable by what it costs the person making it. A confirmed appointment with a date and a time sits at the top, because it fixes the buyer’s own calendar. Just below it, a proposed slot, still open but already dated. Then a request for a phone call, an underrated signal: the buyer is asking to leave the inbox, and therefore to enter a real negotiation.

Next comes a factual question about the vehicle itself: inspection report, fees, extra photos, warranty. It is worth more than a declaration of interest and far less than a date. And far below, at the bottom, sits any mention of a down payment, financing or a trade-in. A prospect who sends the registration document to get a trade-in valued has actually done something: he produced a document. The one who merely announces a down payment has written a sentence.

Three axes, one score

Osmosia scores every conversation out of one hundred points across three axes.

The first measures the action taken. It is the dominant axis, where the confirmed appointment, the proposed slot and the callback request concentrate. A plain “is it still available?” barely registers.

The second measures the quality of the prospect: clarity, tone, market knowledge, documents sent, length of the exchange, response to follow-ups. An explicit walk away, “too expensive”, “I found something else”, collapses that axis at once.

The third measures the trajectory of the conversation. An exchange moving towards a purchase rises. A polite but static exchange stalls. A prospect who stops answering follow-ups, or who drifts away from the advertised vehicle, falls.

One rule follows from that architecture, and it matters more than the rest: no contact crosses 60 out of 100 without a concrete action. No appointment, no deposit, no hot score. A lead only becomes hot by doing something.

What the score triggers

Below 30, the contact is curiosity. It gets a professional answer and nothing more. Between 30 and 50, it enters a follow-up cycle whose timing the seller sets himself. Above 70, the contact is close to conversion and is pushed to the seller by SMS and e-mail, with the full exchange attached.

The score is not an academic grade. It is a trigger. It decides when a human takes over, and for whom. The qualification threshold is a setting, not a doctrine: a dealer moving 150 cars a year does not set it like one selling 20.

One clarification that is not cosmetic: the score prioritises, it does not decide. Automated processing of prospect data, and scoring in particular, falls under the GDPR and carries information and rights obligations towards the people concerned. The French data protection authority, the CNIL, publishes dedicated guidance on artificial intelligence and the GDPR. On this point, consult a qualified legal advisor rather than copy a checklist from a blog.

A score is worthless without speed

The best scoring in the world, applied to messages read the next morning, produces nothing. Harvard Business Review, in “The Short Life of Online Sales Leads” (2011), documented an average response time of 42 hours across 2,241 US companies, and 23 % of companies that never replied at all. Those who answered within the hour were roughly seven times more likely to have a qualifying conversation.

Qualification and responsiveness are therefore the same subject. Osmosia replies in under fifteen minutes, seven days a week, from 7 am to 11 pm, and scores the conversation as it unfolds. The seller does not receive a list of leads, he receives the ones that crossed his threshold.

The economics are measurable. The quarterly barometer published by Mobilians and Avere-France (Q4 2025) puts the average time to resell a used electric vehicle at 161 days. Every day in stock is capital tied up, and recent used stock is thinning out: vehicles under five years old fell by 7 % in 2025 according to AAA Data. A lead on a scarce car left waiting until tomorrow does not cost a sale in theory, it costs a rotation.

Sales organisations point the same way. Salesforce, in its State of Sales 2026 report (4,050 professionals surveyed in late 2025), states that 87 % of sales organisations already use some form of artificial intelligence, and that fully deployed agents cut research time per prospect by around 34 %. That survey largely polls Salesforce customers: read it as a direction, not as a market average.

Manual sorting is not an organisational flaw. It is a task a seller should never have to perform, because it consumes exactly the time he should be spending on the phone with the three people who proposed a date.