Automated lead qualification: lead qualification software that screens every inbound lead against your criteria

Most qualification frameworks fail because the criteria are too abstract to answer. Asking whether a lead has budget produces a yes from almost everyone, because people do not describe themselves as broke to a business they want to hire. Criteria only work when they are questions of fact that a person can answer accurately without knowing what answer you want.

The reliable pattern is to convert each fuzzy criterion into a factual proxy. Instead of asking about budget, ask what the amount in dispute is, or how many units are in the portfolio, or what they are currently paying for the thing you would replace. Instead of asking about urgency, ask when the deadline is, when the letter is dated, or when the lease ends. Instead of asking whether they are a decision maker, ask who else would need to approve. Instead of asking about fit, ask the two or three facts that actually rule a matter out for you, which are usually jurisdiction, date, and matter type.

Then decide, in advance, what each combination means. A hard disqualifier is a fact that always makes the answer no: a state you are not licensed in, a matter type you do not handle, a deadline that has already passed. Those should end the conversation politely at the point of contact, and getting that right is worth more than the scoring. Soft signals are facts that shift priority rather than eligibility, and they should route rather than reject. Keep the hard list short. A screening system that rejects too eagerly is worse than none, because you never learn about the leads it turned away.

Feature pages in this category are close to interchangeable. What genuinely separates products is the unit the invoice is calculated on, because every unit except a flat firm fee grows with something you are actively trying to increase.

Per credit is the model to understand first, because it is common in legal and professional CRM tooling and it is not obvious from the sticker. Lawmatics publishes QualifyAI at $150 for 50 credits, $250 for 100, $400 for 200, $600 for 350, and $750 for 500 a month, with one credit consumed per lead scored and overage running from $4.00 down to $1.50 a credit. Divide it out and the tiers work in the expected direction on the face of it: $3.00 a credit at the entry tier down to $1.50 at 500. The detail that decides the real cost is that credits do not roll over. That means you cannot buy for your average month. You have to buy for your busiest month or accept overage in it, and every quiet month you pay for capacity you never used. A campaign that doubles your inquiries for four weeks is a bill increase, which is an odd incentive for a tool whose job is to handle inquiries.

Per seat and per agent are the other two. Calendly is priced per seat at $10 or $16 a month, so screening cost scales with headcount rather than volume. Goodcall charges per agent at $79, $129, and $249 a month for 100, 250, and 500 unique callers, which makes a six person team on the Growth plan $774 a month for the same configuration one person would pay $129 for. Per call is a fourth: Smith.ai publishes $300 for 30 calls up to $2,100 for 300, and caps standard scripted intake at five questions with additional questions billed per call, which is exactly the constraint that matters if your qualification needs eight.

Flat per firm removes all four variables at once. It is the only structure where running more marketing, hiring another person, and having a good month do not appear on the invoice. When comparing, do the arithmetic on your own numbers rather than the sticker: take your busiest month of inquiries in the last year, your current headcount, and the number of screening questions you actually need, and price each vendor on those three inputs.

Automated screening handles the part of qualification that is factual and repeatable, which is most of it. It should not handle the part that requires judgment about a person, and the boundary is worth drawing explicitly in the configuration rather than leaving to the model.

Three situations should always escalate. The first is anything the criteria do not cover: an unusual matter type, a fact pattern that does not map, a person describing something the script has no branch for. The correct behavior there is to capture what it can, tell the caller a person will follow up, and hand you the transcript, not to improvise a decision. The second is any indication of distress or urgency involving safety, where a screening script is the wrong instrument entirely and the only right answer is a fast route to a human. The third is a borderline case near a threshold you set, which should be routed for review rather than rejected, because thresholds are guesses until you have enough data to check them.

It is also worth being clear about what automated screening does not do. It does not tell you whether a matter is winnable, whether a client will be difficult, or whether the work is worth doing. Those are professional judgments made on facts, and the value of the screening is that the facts arrive complete and comparable so the judgment can be made in two minutes instead of a call. Software that claims to make the judgment itself is overselling, and a vendor that concedes this boundary in the demo is usually the more honest one.