A lead score sales could trust, built from our own CRM data

The challenge

The platform helped Australian businesses with large workforces pay staff correctly under awards, alongside scheduling and timesheets. The business charged by monthly active users (MAUs), so a business with more staff on the platform was worth more.

As lead volumes grew, the business wanted a lead score. I wanted to be clear on what it would be used for first, because a score only helps if it changes what someone does next. Here, there were two jobs for it.

The first was helping sales prioritise. Sales only dealt with businesses that had made an enquiry, so the question was which enquiries were the most valuable and deserved the most effort, and who should follow them up. The sales team had senior and junior salespeople, and the senior salespeople won more often, so we wanted to test whether giving them the highest-scoring leads would win more of the most valuable deals.

The second was deciding who to retarget. We knew about many more businesses than had enquired. HubSpot showed us which companies visited the website, and others had subscribed to our newsletter, downloaded a piece of collateral, been added as a target account, or were deals we’d worked on or lost before. We wanted to stay in front of the best of them, without paying to retarget everyone.

My approach

Start with the data, not the points. Before scoring anything, I went through every field in the CRM and asked whether it could tell us how likely a lead was to convert, how valuable they’d be as a customer, or whether they matched who we wanted to target. For each field that might, I built a small report to check whether it was linked to a higher or lower conversion rate, or a higher or lower customer value. For example, did businesses with more staff convert at a higher rate, and did people who visited certain pages on the website become customers more often? Only the fields that held up went into the score, as positive or negative points.

This came after our move to HubSpot, once the data had been cleaned up and we had processes in place to keep it accurate. That timing was deliberate. A lead score is only as good as the data behind it, and the scores had to be accurate enough for sales to rely on.

Three questions. I built the score around three questions.

1. 1Are they the right fit? This was the gate. A lead had to be a real, active Australian business with an ABN and about 20 or more MAUs, the same criteria the business used to qualify every enquiry. Leads that met the criteria scored higher when they went beyond them, for example with more MAUs, or in an industry where past CRM data showed we won more often.
2. 2Are they interested? This scored how much they’d engaged with us: website visits and the pages they viewed, email opens, form completions, and engagement on social media. More visits and more ways of engaging earned more points.
3. 3Are they interested now? This scored how recently they’d visited, engaged, or filled in a form. It carried the most weight for contacts who hadn’t filled in a form yet, where the question was whether they were looking at us right now. Anyone who filled in a form was called within about 15 minutes, so for them recency was always high.

What I built

Points that went down as well as up. Every factor could add or take away points. A bounced email took points off a contact’s interest score, and visits to the careers page, a sign of a job seeker rather than a buyer, took points off too. On fit, a personal email address lost points, and so did industries that don’t usually employ staff under awards. Engagement points also decayed: if someone hadn’t come back to the website within a set number of days, points came off automatically, so an old burst of interest didn’t keep a lead at the top.

A lead score in HubSpot. I built the score in HubSpot, listing the fields under each of the three questions, each with a positive and a negative score. This was before HubSpot had its current lead scoring tool, which now does a lot of the same things, with separate fit and engagement scores, score limits, and decay built in.

A test of routing by score. We tested sending the highest-scoring leads to our most experienced salespeople. In the end, we routed leads by calendar availability, and the score showed each salesperson which of their enquiries to prioritise and which were likely to be the most valuable.

Retargeting audiences. The score let me decide who went into our retargeting audiences, and who didn’t. That covered companies HubSpot identified from website visits, newsletter subscribers, people who’d downloaded collateral, target accounts, and past deals we’d worked on or lost. Those that were a strong fit and had engaged recently went in, so our ads reached the businesses most likely to buy, even if they hadn’t made an enquiry yet.

Score reporting. I built reports showing how scores were distributed across our leads and which leads scored highest, so the sales team could see where to focus their time.

Results

• Sales knew where to put their effort. Every enquiry showed how likely it was to convert and how valuable it could be, built from our own CRM data, so sales could prioritise and put the most effort into the leads worth the most.
• A score that stayed current. Negative scoring and decay meant a lead’s score reflected what they were doing now, not what they did months ago.
• Retargeting aimed at the right businesses. Ads went to contacts who were a strong fit and interested now, not to everyone who had ever visited the site.
• A clearer view of lead quality. Score reports showed how many of our leads were a strong fit and highly engaged, and where sales should spend their time.

SiriusDecisions (now part of Forrester) found that trust in a lead score, and alignment between marketing and sales, correlate directly with how accurate the scoring model is. That’s why I tested every field against real outcomes before it earned points.

Source: SiriusDecisions (Forrester), “What Is Lead Scoring, Anyway?”

What I’d do next

Score our campaigns and channels. Comparing the average lead score from each campaign and channel would show early on whether they were bringing in the right kind of leads, well before those leads had time to convert.

Use the score for nurturing. The same contacts we retargeted, strong fit and recently engaged but no form yet, could also go straight into a nurture email workflow.

Let AI check the field analysis. HubSpot’s lead scoring tool can now suggest fit and engagement criteria based on which past contacts converted. I’d use it to check my own reports and find fields I’d missed, and still test anything it suggests before giving it points.