Your reps are working leads right now that will never buy. Not because they are lazy, but because nobody told them which leads deserve the next hour. That is the entire job of lead scoring: a repeatable way to rank every prospect by how likely they are to turn into revenue, so your team spends its finite time on the accounts that actually move pipeline.
The trouble is that most lead scoring lives in a spreadsheet nobody trusts. Marketing tags an MQL, sales quietly ignores it, and everyone re-litigates the definitions at the next pipeline review. A model earns trust when it is explicit, two-dimensional, and refreshed against real behavior instead of gut feel.
What follows is a practical framework you can build this quarter without a data science team. We will separate fit from intent, assign points, subtract for red flags, set clean handoff thresholds, and keep the whole thing from quietly rotting. Sendpilot's ICP Scoring and inbound campaigns are the worked example throughout, because they map onto the two axes almost exactly.
Why a Blended Score Fails
The classic mistake is blending fit and intent into one number. A prospect earns +10 for reacting to your case study, +10 for holding a VP title, +10 for commenting on your post, and lands at 30. The problem is that two completely different people share that 30: a perfect-fit buyer who reacted once and moved on, and a tire-kicker who happens to comment on everything you publish.
Once the two are averaged, sales can no longer tell someone who should buy from someone who just likes to engage. You route enthusiastic non-buyers to your closers while quiet, high-fit accounts go cold. The fix is not to abandon numbers; it is to keep the two questions on separate axes.
Sendpilot is a useful way to picture the split. Its ICP Scoring reads the whole LinkedIn profile — title, seniority, industry, company size, skills, endorsements, network, experience and activity — and returns a single 0–100 fit score. That number is fit and only fit. Intent comes from somewhere else: you point an inbound campaign at one of your posts and set the Action Words a comment has to carry, and Sendpilot captures everyone who engages with that post as a lead, logged against the post itself. The comment carrying the Action Word is the trigger — it is what makes Sendpilot reply in public and, for the people you are already connected to, send the DM. Read side by side, a 90-fit lead who has never engaged and a 40-fit lead who comments on everything look as different as they are.
The Two Axes of Lead Scoring: Fit vs. Intent
Every strong model tracks two scores per lead. Fit answers "should we sell to this person at all?" Intent answers "are they showing signs they are ready to talk?" They come from different data, move at different speeds, and mean little until you read them together.
| Dimension | Fit Score (ICP Match) | Intent Score (Engagement) |
|---|---|---|
| Question it answers | Should we sell to them? | Are they ready to buy? |
| Built from | Title, seniority, industry, company size, skills, endorsements, network, experience, activity | Which post they engaged with, whether their comment carried your Action Word, whether they connected after the reply, replies to your DMs |
| Changes | Slowly, over months | Quickly, day to day |
| A high score alone means | Worth pursuing, but maybe not now | Interested, but maybe a bad fit |
| In Sendpilot | ICP Scoring: one 0–100 fit score per lead | Inbound campaigns: everyone who engages with the post is captured and logged; a comment carrying your Action Word is what starts the reply |
The signal you actually want is the corner where both are high: a lead who matches your ideal customer profile and is behaving like a buyer. That is the account your best rep messages today — not the one who reacts to every post but works somewhere you can never close.
Build the Point System — and Do Not Skip Negative Scoring
Keep the math boring on purpose. Assign points, cap each axis at 100, and write the rules down so anyone on the team can audit a score. Here is a starting rubric to tune against your own data.
Fit points (ICP match)
- Job title matches a decision-maker or champion persona: +20
- Company headcount inside your target band, say 50 to 500 employees: +15
- Industry or vertical sits on your ICP list: +15
- Skills and endorsements line up with the problem you solve: +10
- Based in a region you actively sell into: +5
If you run this in Sendpilot, the person and company filters in the ICP builder define the profile, and every lead comes back with its fit score already attached. The rubric is still worth writing down: it is how you check that the profile is asking the right questions when you tune it.
Intent points (engagement)
- Replied to your DM, or booked a call from it: +25
- Commented on your post with the Action Word to ask for a lead magnet: +20
- Sent you a connection request after the connect-first reply: +15
- Commented on your case study or product walkthrough: +15
- Engaged with two or more of your posts in the last 30 days: +10
- Reacted to a post about the problem you solve: +5
Apart from the booked call, every one of these is something Sendpilot observes for you. An inbound campaign captures everyone who engages with the post and records which post it was, and it treats a comment carrying your Action Word as the trigger: that comment gets an automatic public reply, and if the commenter is already one of your connections, the DM follows. If they are not connected yet, the reply they get is the connect-first template asking them to send a request — so the most useful thing a stranger can do next is connect, and once they do, the DM goes out. That is why the connection request earns points of its own. A reaction still tells you something, but it starts nothing, which is why it sits at the bottom of the list. DM replies land in Unibox, and the intent side of the ledger fills itself in as you publish.
Negative scoring (the part most teams forget)
Points should come off as readily as they go on. Without subtraction, every busy commenter eventually crosses your threshold and your MQL list fills with noise.
- Title reads student, intern, or "open to work": -20
- Seniority two levels below the buyer you actually sell to: -10
- Headcount far outside your serviceable band: -15
- Known competitor or an existing customer: route them out entirely
- No activity in 30 to 60 days: let the intent score decay back toward zero
Set Clean Thresholds for MQL and SQL Handoff
Thresholds are where scoring either creates alignment or starts another turf war. Do not lean on a single cutoff. Grade fit into tiers — call them A, B, and C — then layer intent on top to decide the next action. The combination, not either number alone, drives the handoff.
- A-fit and high intent: this is a sales-qualified lead. Route it to a rep the same day, while the intent is warm. In Sendpilot, this is the moment to transfer the lead into an outreach campaign for a follow-up message or voice note.
- A-fit and low intent: a great account that is not ready. Keep it in high-touch nurture or an ABM play rather than burning a cold call.
- B-fit and high intent: a marketing-qualified lead. Nurture it, confirm the fit gaps, and hand off once it clears your SQL bar.
- C-fit, any intent: do not pass it to sales. Enthusiasm never overrides a poor fit, and this is exactly where negative scoring earns its keep.
Write these rules into your CRM so the MQL-to-SQL handoff is automatic and unarguable. When a lead is rejected, make the reason visible — "fit below B" — so marketing and sales debate a rule, not a feeling. If Sendpilot is where the leads are captured, the HubSpot integration, plus webhooks, the API and the MCP server on Growth and up, let them land in the system where those rules live — that is also how you wire in Zapier or n8n if your routing runs there; on Launch, the scored list lives in Sendpilot and you work it from there.
Keep Your Lead Scoring Model From Going Stale
A scoring model is never finished, and the fastest way to lose sales' trust is to let it drift. Two forces cause decay: intent data ages, and your definition of a good customer shifts as you close more deals. Both need a maintenance habit.
- Decay intent over time. A comment on yesterday's post is a live signal; the same comment from six weeks ago is a memory. Age intent points down automatically so old behavior stops masquerading as current interest.
- Recalibrate against closed-won. Once a quarter, pull your last batch of won and lost deals and check whether high scores actually predicted revenue. If your best customers keep entering as B-fit, the rubric is wrong, not the customer.
- Feed it behavior, not just profile data. The richest, freshest intent lives in social engagement. Someone commenting on your LinkedIn post about a specific pain point is telling you more than their job title ever could.
This is where tooling closes a gap manual scoring cannot. Sendpilot's ICP Scoring rescores leads as new profile data comes in, so fit stays current instead of freezing at the moment you first found someone, and each inbound campaign logs which post a commenter or reactor engaged with, so intent refreshes every time you publish. Scoring spends from the same monthly credit balance as the lead database, the lead extractor and enrichment — one shared pool per LinkedIn account, not a separate scoring quota — so plan your rescoring alongside the rest. Instead of scoring a prospect once and forgetting them, you get a fit-and-intent picture that updates as they act — which is the whole point of fighting drift.
The Bottom Line
Good lead scoring is not a clever formula; it is a shared agreement about who deserves your team's time, written down so it survives the next pipeline review. Split fit from intent, put points on both, subtract for the disqualifiers, and let the two scores together decide the handoff. Then keep it honest by decaying old signals and re-checking the model against the deals you actually win.
If you want fit and intent to stay in sync without a spreadsheet babysitter, that is what Sendpilot's ICP Scoring and inbound campaigns are built for: a 0–100 fit score read from the whole LinkedIn profile, and post-engagement capture that replies to the comment, asks anyone you are not connected to yet to connect first, and records the intent behind it. Both are on every plan, priced per LinkedIn account — never per user. Point them at your own pipeline and let your reps spend their next hour on the leads that were always going to close. Start a free trial, or see /pricing for what each plan includes.
