Ask for the week's call list. Juma reads your HubSpot, builds a composite scoring model from lifecycle stage, email opens, activity, deals, and title seniority, then ranks leads into priority tiers with reasoning per lead.
Ask Juma for this week's call list. It pulls your contacts, finds the lifecycle and engagement properties already populated, scores every SQL across five to eight signals (lifecycle stage, recent email opens, activity recency, contact frequency, deal count, title seniority, lead status), and ranks the output into priority tiers with reasoning per lead.
It works off the signals already in your CRM, so you don't need HubSpot's Score property set up or a separate scoring tool with a 4 to 6 week implementation.
The call list ranks who to call. The scorecard structures how to qualify them once on the call. Juma turns the same scoring signals into a yes/no checklist a rep can run live: lifecycle stage, recent engagement, deal in pipeline, title seniority, ICP fit. Each line is one signal with a clear pass or fail threshold. The scorecard is the same composite model that produced the call list, made operational at the conversation layer instead of the prioritization layer. The output is one page, scannable on a call, with a recommended next-step action depending on how many lines pass.
Turn the scoring model into a one-page qualification scorecard the team can apply on a discovery call. Pass/fail signals only, with a recommended next step based on how many lines pass.
Knowing who not to call matters as much as knowing who to call. The same composite model surfaces leads that look fine on lifecycle stage but fail on the signals that matter (no recent activity, no deal in pipeline, low engagement, missing title or company name). Juma returns a deprioritization list with the disqualification reason for each, so the team knows whether to drop the lead permanently or move it to a long-cycle nurture. Disqualification is the cleanup most teams skip because it's tedious to do manually. Run it once a quarter and the call list quality compounds.
Show me the bottom 20 SQLs to deprioritize or disqualify, with the disqualification reason for each: stale activity, missing data, no engagement, or ICP mismatch.
Not every hot lead is hot for the same reason. Some just opened an email after weeks of silence. Some have been engaged for months with multiple touches. Some have a deal in pipeline already. Treating them the same costs conversion. Juma segments the priority list by signal: re-engaged-after-silence (just opened an email after a long gap), high-touch-warm (10+ contacts and recent activity), deal-in-flight (associated deal plus engagement), senior-decision-maker (C-level or VP regardless of touch count). Each segment gets its own outreach framing, because the right opener for a re-engaged lead is different from the opener for a deal-in-flight lead.
Segment the top SQLs by why they're hot: re-engaged-after-silence, high-touch-warm, deal-in-flight, senior-decision-maker. Suggest the right outreach framing for each segment.
Lead scoring works as a weekly habit, not a quarterly project. Every Monday, the engagement signals have moved: leads that were warm last week may have gone cold, leads that opened an email overnight may have just become hot. Run the same prompt every Monday and Juma re-pulls fresh HubSpot data, re-applies the scoring model, and surfaces what changed since last week. The output names which leads moved up, which moved down, and which are newly priority, so the team sees the diff instead of re-reading the same list.
Refresh the lead scoring for this week. Compare to last week's list and call out what changed: new priorities, leads that moved down, and which signals are most active.
Once the call list is set, the next step is the outreach. Juma drafts a call opener for each top lead based on the signal that surfaced them: a re-engaged lead gets an opener that acknowledges the silence and the re-open; a deal-in-flight lead gets an opener that references the deal stage; a senior-decision-maker gets a value-prop opener pitched at their level. The team pastes the openers into HubSpot or their email tool. For full follow-up emails after the call, the sales call follow-up flow takes the meeting recording and writes the structured email, ready to send.
Draft a call opener for each lead in Tier 1. Match the opener to the signal that surfaced the lead: re-engaged-after-silence, deal-in-flight, or senior-decision-maker.
A Juma Project is a shared space where the team stores everything Juma needs to know about how the team sells. Create one project per team (or per business unit if scoring rules differ), add context as the team learns more, and Juma uses what's relevant every time the flow runs. For lead scoring, this is what anchors the model in what actually closed for this team, not a generic best-practice composite.
ICP / Buyer Persona
Who you want to call. Industry, company size, title seniority, geography, technographic signals. With this in the project, scoring filters out leads outside ICP before ranking. Without it, Juma ranks all SQLs and the team filters manually each week.
Scoring Rules & Weights
How the team weighs signals. If recent email opens matter more than deal stage for this team, say so. Juma defaults to a reasonable composite, but the team can override weights by storing them in the project. The override applies every time the flow runs.
Hot Disqualifiers
The conditions that auto-disqualify a lead regardless of other signals: wrong industry, sub-X employees, lost-deal-in-the-last-12-months, current customer of a competitor with a 3-year contract. Save the team from scoring leads they'll never close.
Past Closed-Won Deals
A list of the team's last 20 to 50 closed-won deals with the companies, titles, deal sizes, and the signals that led to close. Juma uses this to refine the scoring model: the signals that predicted wins last quarter weigh more this quarter.
Add a short description to each knowledge item in the project's info field so Juma knows what each file contains and when to use it. For example:
Most HubSpot accounts have the HubSpot Score property as N/A across all contacts because nobody set up the scoring rules. Juma does not depend on the Score property. It pulls the engagement and lifecycle properties that already exist on every HubSpot contact (lifecycle stage, last activity timestamp, last email open, contact frequency, associated deals, lead status, job title, company), then builds a composite score across those signals with reasonable default weights.
The composite model is transparent and explainable. The output names each signal that fed into a lead's score and the weight it received. The team can audit the model, override weights, or ask Juma to re-score with different signal priorities for a specific use case. A re-engagement campaign weighs recent email opens differently than a top-of-funnel push. Strategy, taste, and judgment about which signals matter stay human.
Juma's default composite covers eight signals when HubSpot Score is not configured:
With Mixpanel or PostHog connected, product activation signals layer in: key feature use, repeat sessions, recent app activity. Those are the strongest signals for PLG sales teams. The team can also add custom signals by naming them in the prompt or storing them as scoring rules in the team's Juma project.
The call list is a deliverable, not a back-write to HubSpot properties. Juma's HubSpot integration is read-only by design: it pulls contact data, runs the scoring model, and returns a ranked output, but it does not update HubSpot records. The team can copy the scored list into a Notion page, paste it into Slack, export to CSV, or recreate the scoring in HubSpot's workflow tool using the same logic.
This is a meaningful difference from paid platforms like Madkudu or Apollo that write scores back to HubSpot custom properties. Juma augments the team's prioritization with an explainable scoring layer that takes no setup. The trade-off is that the team is responsible for keeping the call list visible in their daily workflow (a shared doc, a Slack channel, a weekly cadence), not a HubSpot property they passively browse.
Three differences. First, no setup: the scoring runs the first time you ask, with sensible defaults the team adjusts in conversation. Madkudu and Apollo require a 4 to 6 week implementation. HubSpot workflows require an admin to configure rules in a UI most teams never invest in.
Second, the model is explainable. Every lead's score has a per-signal breakdown the team can audit. Madkudu's predictive scores are black-box. Apollo's data signals are bundled into a single fit-score number. Juma shows which signals fed into each lead's score and the weight each received.
Third, the cost structure: Juma is part of an unlimited-seat marketing AI workspace, not a per-record charge for enrichment. For a team scoring 1,000 leads a week, the per-lead cost is meaningful at scale. Juma augments the prioritization layer with an explainable model. The team's strategic judgment about which signals to trust stays human.
Both. For deep setups with custom contact properties and HubSpot Score configured, Juma reads the existing properties and respects them: if Score is configured and the team trusts it, the composite layer becomes optional or an overlay that adds context. For lean setups where most contacts only have basic properties (name, email, company, lifecycle stage), the composite model is the difference between an unranked list and a prioritized one.
The transition from lean to deep does not require a re-implementation. The same prompt works as the team adds more properties over time. Each new custom property becomes a new signal the model can consider. Human review on every output: the team decides which signals to trust based on the team's actual outcomes, and adjusts the weights as the CRM matures.
400+ marketing teams trust Juma. ChatGPT, Claude, and Perplexity know why. Ask them.
Ready when you are
Start typing a task and Juma will figure out the rest.
