Automated lead scoring from intent signals means two numbers, kept apart and then combined. The first is fit: does this person match the buyer you sell to (title, industry, seniority, company size, location)? The second is intent: did they just do something that makes now a good time (changed jobs, commented on a post in your space, accepted a connection request)? A good system adds intent in proportion to fit, lets every signal fade with time, and re-ranks the list without anyone touching it.
That last part is where most teams fall down. The scoring idea is simple. Keeping it current by hand is not.
The manual way, and what it costs
Here is the manual version.
- A fit score in a spreadsheet. Export the list from Sales Navigator or Apollo, add columns for title match, industry match and company size, and give each a weight. A formula adds them up.
- Alerts for intent. Save the leads in Sales Navigator and, in LinkedIn's words, "let Sales Navigator send you alerts when they make a move, like changing jobs." You can also save searches "so that you're notified when there are new results."
- Re-ranking by hand. Each morning someone reads the alerts, finds the person in the sheet, bumps their score, and moves them up the call or message list.
It works for a few hundred leads and one client. Then it breaks in three places:
- Signals never expire. A job change from five months ago still carries its bonus because nobody goes back to remove it.
- Fit and intent get added flat. A student who comments on your post gets the same bump as a VP of Sales who does. The sheet can't tell the difference unless someone thinks about it every time.
- The sheet is always behind. Alerts arrive as a feed. The score only changes when someone has time to copy it over, which on a busy week is never.
Step 1: Score fit first, with weights you can defend
Fit answers "should we ever talk to this person?" Score it from fields you can check on a profile, and weight the ones that matter most for the sale.
Here are the weights we use in CampaignStack. They are a fair starting point for any B2B list:
| Criterion | Weight |
|---|---|
| Job title | 30% |
| Industry | 20% |
| Seniority | 20% |
| Company size | 15% |
| Location | 10% |
| Keywords | 5% |
One detail matters more than the numbers. Only count the criteria you actually set. If your ideal customer profile says nothing about location, a lead shouldn't lose 10 points for living anywhere. Divide by the weights you used, not by 100%.
Step 2: Add intent in proportion to fit
Spreadsheets usually skip this step. Don't add a signal's points flat. Scale them by fit.
The rule we use: total score = fit score + (signal points x fit score / 100).
So a signal is worth its full points on a perfect fit, half on a 50 fit, and nothing on a zero fit. A great signal on the wrong person stays a wrong person.
A worked example
This is arithmetic from the rule above, not a result from a campaign.
Your profile sets title, industry, seniority and company size. Those weights add up to 85%.
- Lead A matches title, seniority and company size: 30 + 20 + 15 = 65 out of 85, so a fit of 76.
- Lead B matches industry only: 20 out of 85, so a fit of 24.
Both change jobs this week. A job change is worth 30 points by default.
- Lead A: 76 + (30 x 76 / 100) = 76 + 23 = 99.
- Lead B: 24 + (30 x 24 / 100) = 24 + 7 = 31.
Same signal, very different result. In a flat sheet, both would have gained 30 and Lead B would sit at 54, close enough to the top that someone would message them.
Step 3: Pick signals that mean "now"
Not every signal means the same thing. Here are the defaults we start from in CampaignStack. Each one can be changed per campaign:
| Signal | Default points | Counts for |
|---|---|---|
| Changed jobs | +30 | 90 days |
| Changed title or seniority | +20 | 90 days |
| Updated their headline / location | +10 | 60 / 90 days |
| Joined a LinkedIn group you watch | +15 | 90 days |
| Attended an event you watch | +15 | 90 days |
| Commented on one of your posts | +50 | 30 days |
| Reposted / reacted to one of your posts | +40 / +25 | 30 days |
| Commented on / reacted to a competitor's post | +30 / +15 | 30 days |
| Accepted your connection request | +25 | 30 days |
| Ignored your connection request | -20 | never expires |
| Declined your connection request | -30 | never expires |
| Ignored your message | -15 | 90 days |
Two opinions baked into that table. A comment is worth more than a like, because writing costs the person something. And negative signals count too: someone who declined your request should drop, not sit at the top of the list forever.
Step 4: Let signals fade
An expiry date is the bare minimum. Engagement on a post should fade much faster than a job change, because interest in a post is gone in days.
We fade post engagement by the age of the post:
| Post age | Signal counts at |
|---|---|
| Under 1 day | 100% |
| 1 to 3 days | 90% |
| 3 to 7 days | 70% |
| 7 to 14 days | 40% |
| 14 to 21 days | 20% |
| 21 to 30 days | 10% |
| Over 30 days | 0% |
It also gives fresh engagement a short boost: 2x if it was seen in the last 24 hours, 1.5x within 3 days, 1.2x within a week. And it reads what a comment says. A comment that asks a question or describes a problem counts 3x. "Great post" counts half. Spam and self-promotion count zero.
You can build all of this in a spreadsheet. You will spend more time keeping it true than using it.
Set your ideal customer profile once. Every lead gets a fit score, and signals move it up or down every few minutes.
Score your leads from signalsHow CampaignStack runs it for you
CampaignStack does Steps 1 to 4 without the spreadsheet.
- You describe the buyer once, as an ideal customer profile on a campaign. Every lead you hold is scored against it.
- Signals come from LinkedIn. Profile changes show up when a lead's profile is read again. Engagement comes from the posts, groups and events you ask it to watch.
- The score updates on its own. Every 2 minutes, CampaignStack re-scores the leads whose profile or signals changed, with the same fit rule, fade and expiry.
- You can see why. The Score column opens a breakdown: which criteria matched, which signals counted, and how much each one added.
- The score drives the work. A workflow can filter on the score or on how recently someone changed jobs. A signal agent watching posts also checks fit before it acts: a strong fit can get a message or a connection request, a weak one a like or nothing.
- A person still approves the message. New workflows start with review before anything is sent.
What to watch out for
- Signals are only as fresh as the profile. A job change counts once the new profile is read. If a lead hasn't been re-read in months, the change sits there unseen. Watch the leads that matter most.
- Don't chase engagement from bad fits. It feels like intent. With the fit rule, it barely moves the score, and that is on purpose.
- Tune for the campaign. A recruiting-tech client cares about job changes more than a security client does. Edit the points per campaign instead of living with one global table.
- Third-party intent data is a different thing. Topic-research data from outside vendors is not part of this. Everything here comes from what is visible on LinkedIn and from your own outreach.
- Imported spreadsheets stay plain. A lead that exists only as a row from your CSV, never matched to a LinkedIn profile, shows no signals and no score.
Takeaway
Score fit first, and only on the criteria you set. Add intent in proportion to fit, and make every signal fade. Then take the re-ranking off a human's morning: it is the part that always slips.