Lead Scoring That a Two-Person Sales Team Can Actually Maintain

Every sales team hits the same wall: the pipeline holds more names than anyone can call in a week. The obvious fix is lead scoring. The obvious trap is building a model that quietly assumes someone will maintain it. When your entire commercial operation is two people who also run demos, chase contracts, and handle the odd support escalation, the scoring system has to survive on almost no upkeep. Here is a version that does - three signals you already collect, three tiers instead of a point total, and a fifteen-minute weekly habit that keeps the whole thing honest.

Why Most Lead Scoring Models Die in Month Three

Small teams usually inherit their scoring framework from a blog post written for a company with a marketing ops department behind it. And the failure pattern repeats with grim reliability. Someone builds forty point-assignment rules over a productive weekend, everyone nods at the logic, and by month three nobody remembers why a webinar registration is worth twice a pricing-page visit. The score keeps updating. People just stop looking at it.

The real constraint was never analytical sophistication. It is how many minutes per week two people can actually spend feeding a model. A score nobody trusts is worse than no score at all, because it manufactures false confidence about which deals to walk away from. So reframe the goal. You are not building a perfect ranking. You need a defensible answer to one question: who do I call first this morning?

Start With the Three Signals You Already Have

Skip the data collection project. Three dimensions carry most of the predictive weight, and all three can be read off records you already keep.

Fit asks whether this company looks like the customers who renewed and paid on time. Intent asks whether they recently did something that cost them effort - submitted a form, replied to an email, booked a call, opened a ticket. Reachability asks the unglamorous question: do you have a named human, a working address, and a legitimate reason to contact them again?

• Fit: industry, company size band, contract history, whether past deals closed or stalled
• Intent: last activity date, open deals, recent tickets, inbound replies
• Reachability: named contact, verified email, current owner, lead source

Tip: if a signal cannot be filled in without asking a colleague, it does not belong in a two-person model.

The Simplest Model That Works: Three Tiers, Not a Hundred Points

Throw out the numeric score. Use A, B, and C tiers, each defined in plain language a new hire can absorb in under a minute.

1. A - right fit plus recent intent. Call today.
2. B - right fit, no recent signal. Put it in a sequence and set a reminder.
3. C - wrong fit or unreachable. Park it. Do not delete it; you will want these later.

Numbers invite endless tuning debates. Is that lead a 71 or a 78, and does the difference change one single thing you do on Tuesday? Tiers force a decision about behavior instead of an argument about arithmetic. Write the definitions somewhere the whole team can see them rather than leaving them in one person’s head, and make sure each tier maps to exactly one next action. If you want a worked example of the same idea applied end to end, this walkthrough on how to rank prospects in practice covers the tier definitions in more detail. Then the score drives the calendar directly, with no interpretation step in between.

Make the Score a Byproduct of Work You Already Do

Maintenance drops to near zero when scoring fields update as a side effect of normal activity instead of as a separate data-entry ritual. Dragging a card across a Kanban board, logging a call, closing a support ticket, sending an invoice - every one of those is a score-relevant event somebody was going to do anyway. Deadline reminders and clear task assignment keep B-tier leads from silently aging into irrelevance while you work the A list, and a little process automation is what turns those side effects into reliable field updates.

This is also where AI earns its keep honestly. It can surface patterns across past won and lost deals that two people have no hours to dig through by hand, and it can draft the follow-up so a tier assignment actually turns into a sent email. Systems like EpicCRM keep records, tasks, tickets, and AI-assisted scoring in one place for exactly this reason.

Tip: if changing a lead’s tier takes more than one click, that tier will be wrong within a month.

The Weekly Review That Keeps the Model Honest

Fifteen minutes, same slot every week, both people present. Four questions:

1. Which A leads did we fail to contact?
2. Which C leads bought anyway?
3. Which fields were blank at the moment we needed them?
4. What single rule should change?

Question two matters most. A C-tier lead that converts is the richest feedback you will ever get, because it proves your fit definition is wrong in a specific, correctable way. Resist the urge to fix everything at once. Change at most one rule per review, so next month’s results can be traced to something. Saved views, filtering, and export turn this meeting into a reading exercise instead of a painful reconstruction of what happened - the same reason a short set of reports read every Monday morning beats a dashboard nobody opens.

What to Skip Until You Are Bigger

Predictive models trained on your own history need more closed deals than a two-person team produces in a year. Feed a model thirty outcomes and it will confidently learn your noise. Negative scoring, decay curves, multi-touch attribution - all of it adds maintenance without changing tomorrow’s call list, which is the only output that matters right now. Splitting lead scores from account scores becomes useful once territories or handoffs exist, and not a moment earlier.

The honest framing: the ceiling on a simple model sits high enough that most small teams never hit it. What genuinely scales from day one is unglamorous. Clean records, consistent pipeline stages, one owner per lead. Those three habits beat any scoring algorithm layered on top of messy data, and they are the same foundation that carries over when the operation grows into a real sales team with more than two people in it.

Frequently Asked Questions

How often should a two-person team rescore its leads?

Continuously and never, depending on which layer you mean. The underlying signals should refresh on their own as activity happens - a logged call, a closed ticket, a new contract quietly updates a lead’s standing without anyone sitting down to rescore anything. The rules themselves deserve one deliberate look per week during your fifteen-minute review, where you change a single thing. Rewriting the tier definitions wholesale should be rare, triggered only when your ideal customer profile or your offer genuinely shifts. Rescoring on a fixed monthly calendar is usually busywork dressed up as rigor.

Putting It to Work Next Monday

Three signals, three tiers, one action per tier, one weekly review. That is the whole system. Your first version should take an afternoon to define and should feel almost embarrassingly simple written down. That feeling is a good sign, not a warning. Complexity is easy to add later and nearly impossible to remove once people have built habits around it.

Judge the model by one criterion: does the team actually follow it? A clever framework that gets ignored produces exactly zero calls. A crude one that both people check every morning reshapes the whole week. Lead scoring is a habit supported by a system, not a formula living in a spreadsheet. Build the habit first, and let the tooling carry the parts your memory would otherwise drop.