# HubSpot lead scoring: the one-hour check before you trust the queue

https://salesbond.app/blog/hubspot-lead-scoring-sales-manager-check/
Published: 2026-10-15 · Updated: 2026-10-26

To check HubSpot lead scoring, export the contacts that reached your reps with the score band they had at that moment and what happened to them next, then compare the share that became deals in each band. If your top band converts at the same rate as the bottom one, the score is sorting your reps’ queue by noise. The check takes about an hour and a spreadsheet, and we think it belongs in the calendar once a quarter, because a score nobody has compared with outcomes is a guess with a number on it.

Building the score is well documented in HubSpot’s guide to [building lead scores](https://knowledge.hubspot.com/scoring/build-lead-scores). The part nobody writes down starts once the score has fed the routing workflow for a quarter. Is a high score a buyer, or a busy inbox?

## The score nobody checks

A rep opens the Monday view sorted by score. Top row: a contact at 84, label A1. Opened eleven emails, clicked the pricing page twice, downloaded a guide. The rep calls. Voicemail. Calls again Tuesday. Voicemail. Three rows down sits a 31, label C3, who replied to one email with “can we talk Thursday?” and is now a deal, because the rep ignored the sort order for once.

Nobody did anything wrong. The score counts what it was told to count, and it was told that email opens matter. The contact who opened nothing and replied once is a buyer. There is no column for “replied with a question” unless someone added one.

The failure is rarely the model. The score gets built once, wired into a routing workflow, and turns into furniture. Reps sort by it because its there. Managers trust it because the reps sort by it. Nobody has put the score next to the outcome and looked.

## What HubSpot lead scoring actually stores

The number your reps sort by is a sum of points someone chose, not a probability. HubSpot writes it into a score property on the record. By default it tops out at 100, and the points come from groups, each with its own cap that feeds the total ( [HubSpot: build lead scores](https://knowledge.hubspot.com/scoring/build-lead-scores)). An 80 is not 80 percent of anything. It is a mix of meetings, opens and clicks, in that account’s weights.

Thresholds live in a second property. For a combined score, HubSpot labels the record from A (high fit) to C (low fit) and from 1 (high engagement) to 3 (low engagement), so a low-fit but very engaged contact reads C1 ( [HubSpot: build lead scores](https://knowledge.hubspot.com/scoring/build-lead-scores)). For an engagement-only or fit-only score the labels are High, Medium and Low ( [HubSpot: understand the lead scoring tool](https://knowledge.hubspot.com/scoring/understand-the-lead-scoring-tool)). The cut-offs behind them were typed in by whoever built the score, so “High” means whatever went into that box.

Two more things worth having in your head. A score can go negative when more points are subtracted than added ( [HubSpot: understand the lead scoring tool](https://knowledge.hubspot.com/scoring/understand-the-lead-scoring-tool)), a contact below zero in the queue is a signal, not a bug. And every record keeps a score history, which is the answer to the community thread asking “what raised this lead’s score”: open the record and read it ( [HubSpot: build lead scores](https://knowledge.hubspot.com/scoring/build-lead-scores)). You will need that history in the comparison below.

There can be several scores on the same object, one per team or region ( [HubSpot: understand the lead scoring tool](https://knowledge.hubspot.com/scoring/understand-the-lead-scoring-tool)). We have watched a team argue about a score for twenty minutes before noticing the reps’ view was sorted by a different one.

## Three questions for a Monday check

A manager needs three answers: whether the top band becomes deals more often than the bottom band, how old the points behind the score are, and how long a lead waits between crossing the threshold and the first touch from a rep.

The first one matters most. When we called A1 first, were we calling the right people? If A1 and C3 produce deals at about the same rate, routing is a coin toss with extra steps. A score only ranks leads; [seven lead status values that tell a rep what to do next](https://salesbond.app/blog/hubspot-lead-status-vs-lifecycle-stage/) do the rest.

The second is the age of the points. HubSpot lets you turn on score decay for engagement and combined scores, at intervals of 1, 3, 6 or 12 months, with a month counted as 30 days. Decay runs per event and applies to historical events too ( [HubSpot: understand the lead scoring tool](https://knowledge.hubspot.com/scoring/understand-the-lead-scoring-tool)). With decay off, a webinar from last spring weighs the same as yesterday’s pricing page visit, and the top of the queue slowly fills with people who were interested once. Ask whether decay is on, and at what interval. If the answer is a shrug, the queue is older than it looks.

The third is what happens after the threshold. HubSpot’s own routing example enrols contacts in a workflow when the threshold property reaches High, defined there as 70 to 100, assigns a rep by round robin, sends the rep an internal notification and turns re-enrolment off so a contact is routed once ( [HubSpot: build lead scores](https://knowledge.hubspot.com/scoring/build-lead-scores)). The manager’s question is the gap behind it: from the minute a contact crosses the line to the first logged call or email, how long does it wait, and does anyone notice when it waits three days? With a [sequence](https://salesbond.app/blog/hubspot-sequences-small-sales-team/), the wait is the sequence delay. With a human, it is whatever was already on their plate. Writing that expectation down is the point of a [lead handoff agreement](https://salesbond.app/blog/lead-handoff-agreement-pipedrive/), and it applies on HubSpot word for word.

Short version, for the planning meeting:

- Top band against bottom band: the share that became deals.
- Age of the points: decay on or off, and at what interval.
- Threshold to first touch: who gets the lead, and how long it waits.

## Reporting on lead scoring: the one-hour comparison

Start with the window. You want contacts that crossed the threshold long enough ago to have either become a deal or clearly not: end the window at least one full sales cycle before today. If your cycle is six weeks, stop it six weeks ago.

Then export. One row per contact, with the threshold label, the create date, the lifecycle stage, the contact owner and the number of associated deals. The label you want is not the one in the property today. Today’s score has already seen what happened next: decay has taken points off, and a contact who booked a demo has gained some, so it partly measures the outcome you are trying to predict. Compare on the band the contact was in when it crossed the threshold and went to a rep. For the window you are checking, that comes from each record’s score history, and it is the slow part of the hour. From now on it is worth freezing that label at the moment of routing, in a property of its own, so next quarter’s export already carries it. If nobody set thresholds, cut the score at routing into three bands at whatever values your routing workflow uses; the point is to compare the bands the reps actually saw.

In the spreadsheet, add one column: became a deal, yes or no. “Has at least one associated deal” is good enough. Lifecycle stage works too, if your team moves it reliably. Then sum it up in three rows, one per band: contacts, how many became a deal, and the share.

![Spreadsheet mock: one row per contact with score and A1, B2, C3 label, plus a three-row summary by band with empty cells](https://salesbond.app/wp-content/uploads/2026/10/hubspot-lead-scoring-sales-manager-check-2.png)

Now read it. The top band is clearly above the bottom one: the score sorts the queue in the right direction, so leave it alone until next quarter. The bands are roughly flat: the queue is sorted by noise, and either the criteria need rebuilding around the events that preceded real deals, or the reps should stop sorting by it until they are. The bands are inverted: something is scoring the wrong behaviour, and the usual suspects are in the next section.

One honest limit. This comparison shows what happened on your data, not why. A rep who always works A1 first gives A1 more attention, and attention makes deals; part of the gap is the rep, not the score. That is fine. What you are testing is whether the order is defensible, and a top band that converts like the bottom one is not, however it got that way.

HubSpot’s help page suggests a custom report to track how scores affect deal creation or close rates ( [HubSpot: build lead scores](https://knowledge.hubspot.com/scoring/build-lead-scores)), and once the check is routine a report saves the export. The spreadsheet earns its hour the first time because you keep the rows: you can open the ten A1 contacts that never became deals and read what they did, and that is where the criteria get fixed.

## When the number is wrong on purpose

Most bad scores are not broken. They are doing exactly what someone configured, in a situation nobody pictured.

Form spam is the classic. When an engagement rule uses “contains any of” across several forms, the points are additive: a contact who submits both the Contact Us form and the Newsletter form gets the points twice ( [HubSpot: understand the lead scoring tool](https://knowledge.hubspot.com/scoring/understand-the-lead-scoring-tool)). Someone who fills every form on the site in an afternoon climbs the queue, and a rep gets a “hot lead” who was collecting PDFs.

Calls are the quiet one. Scoring on call events only counts calls initiated through HubSpot’s own calling ( [HubSpot: build lead scores](https://knowledge.hubspot.com/scoring/build-lead-scores)). A rep who dials from a mobile and logs nothing leaves the contact looking untouched, so a lead that has been spoken to twice can still sit in the the “unworked” segment that feeds the routing workflow. The fix is a logging habit, not a scoring change.

Dormant leads are the decay question again: with it off, a contact who did everything right in March still outranks the one who did one thing right yesterday.

The AI-built scores, HubSpot’s take on predictive lead scoring, deserve their own line. On Marketing Hub Enterprise, HubSpot can create a contact engagement or fit score with AI: you pick a lifecycle stage change, such as Subscriber to Sales Qualified Lead, and a timeframe, such as the last 90 days, and HubSpot evaluates your contacts to recommend criteria and points ( [HubSpot: build contact lead scores with AI](https://knowledge.hubspot.com/scoring/build-lead-scores-with-ai)). Read that carefully. The model learns from who your team marked as qualified. If reps mark SQL generously, the model learns generosity, and the one-hour comparison is the only thing standing between that and your queue.

## After the lead becomes a deal

Once a contact turns into a deal, its lead score stops being the question. HubSpot keeps the two apart: lead scores are built for contacts and companies, and scoring deals is a Sales Hub feature ( [HubSpot: understand the lead scoring tool](https://knowledge.hubspot.com/scoring/understand-the-lead-scoring-tool)). The deal side has its own blind spots, which we covered in [HubSpot deal score: what it measures and what it cannot](https://salesbond.app/blog/hubspot-deal-score-likelihood-to-close/). Its first need is a deal record filled in enough to be judged at all, and that is what [required properties per stage](https://salesbond.app/blog/hubspot-stage-required-properties/) are for.

## An hour a quarter

We build SalesBond, a sales management layer on top of the CRM a team already uses, HubSpot included. It reads deals, not leads, so it has nothing to add to the export above. Once the lead is a deal, Deal Health checks it against more than 20 rules and shows the manager which deals have gone quiet and why. [Check your CRM with SalesBond](https://salesbond.app/) when the A1 leads have become deals and you want to know which of those are already drifting.

The check itself needs none of that. Put it in the calendar for the first Monday of the quarter, export the window, count the bands, and bring one number to the planning meeting: the share of top-band contacts that became deals, next to the share of bottom-band ones. If the first number is not clearly bigger, say so out loud, and the conversation about the queue starts from evidence instead of from the sort order. If you also run Salesforce, here is [the same check for Salesforce lead scoring](https://salesbond.app/blog/salesforce-lead-scoring-manager-check/).