What is lead scoring and why does it need an explanation attached

What actually gets assessed

The score is built from features of the message itself, not from data about the person. Does the text describe a task or only name a topic. Is a deadline mentioned. Is there a budget or a volume. Is the author asking for a service or offering one. Is this a first question or a reply inside somebody else discussion.

What is deliberately absent matters too: there are no guesses about income, job title or ability to pay. Guesses like that rest on stereotypes and are wrong more often than they help.

A separate feature is fit with your stated direction. A message can be an excellent request and still not be yours, in which case the form scores high and the substance scores low, and the explanation shows exactly that.

Why the explanation matters more than the number

A score with no explanation cannot be checked. The salesperson either takes it on faith or stops looking altogether, and both outcomes are equally bad. The explanation moves scoring out of the realm of magic: you can see what lowered it, and you can see when the system was wrong.

The practical benefit is that explanations are how you tune the project direction. If messages outside your niche keep scoring high, the description is phrased too broadly, and that is fixed with text rather than with complaints about the algorithm.

The second benefit is training. A new salesperson reading explanations alongside scores learns within a week what counts as a genuine request in your market.

Where scoring gets it wrong

Short messages are the leading source of errors. "Anyone building websites?" is a real request, but it carries almost no features, so it lands mid scale. The remedy is not a better algorithm but the habit of looking at the middle of the range as well.

Second come niches with their own jargon. Where buyers write in trade language, the features of a request look different, and the first days of work produce a noticeable share of misses in both directions.

Third is irony and recounted experience. "We had exactly that built recently" matches every formal feature of a request and is not one.

Using the scale in practice

A reasonable routine looks like this. Anything above 70 is handled first and within the hour. The band from 40 to 70 gets reviewed as a batch once a day: that is where short messages and unusual phrasing sit. Below 40 deserves a diagonal skim once a week, so you can see whether anything important is falling there.

Avoid setting a hard cut off in the first month. While the direction is still being tuned, a cut off hides exactly the examples you need in order to tune it.

And finally: scoring does not replace qualification. It says how closely a message resembles a request, not how well the client suits you. The second is discovered in conversation and no scale measures it.

Where the idea comes from

Scoring comes from situations where enquiries outnumber the time available to read them. At 5 a day the order does not matter: someone will look at all of them. At several dozen, working from the top of the list means the best ones sit unread until evening, and by evening half of them have already agreed terms with somebody else.

In open chats the pressure is higher than with a form on a website. A form is filled in by someone who already found you; a question in a chat is addressed to everyone at once, and several contractors answer in sequence. Here the order of processing converts directly into the share of conversations you win.

Hence the correct framing: scoring exists to arrange the queue, not to throw work away.

How to check that scoring works for you

Take the last 20 enquiries and mark which reached a conversation. Compare with the scores. If conversations are spread evenly across the scale, scoring is doing nothing for you, and the cause is almost always the project description.

Second test: look at what was set aside. Once a week, skim the bottom of the scale. One genuine request found there is worth more than the time saved by skipping it.

Third: compare scores with your salesperson judgement, not with deal outcomes. Outcomes depend on negotiation and price, while scoring assesses only the message. Mixing the two means demanding of the scale something it does not measure.