Insurance Lead Quality Scoring: How Agencies Prioritize Follow-Up

Lead quality scoring gives every new lead a number, and that number decides three things: how fast someone calls, who makes the call, and what happens if the lead doesn’t pick up. Agencies that do it well stop working leads in the order they arrive. They work them in the order they’re most likely to quote and bind. This guide covers what to score, how to set score bands, how to wire the bands into routing and cadence, and how to check that the model is working.

TL;DR

• Score every lead on two things: fit (can we place this person?) and intent (are they ready now?). Fit alone sends producers after leads that never answer. Intent alone sends them after leads they can’t write.
• Use three working bands on a 0 to 100 scale. Each band gets an owner, a response time and a next action. A score with no assigned action is just a number.
• Source and consent quality belong in the score. A lead with no consent record is a liability, not a hot lead.
• Compliance sits above the score. A high score never overrides DNC, calling hours or consent rules.
• Review the model every quarter. Scores drift as lead sources, carriers and seasons change.

Table of Contents

• What lead scoring does on a call floor
• Fit and intent: the two halves of the score
• A working model: score bands and the action behind each
• Turning scores into response time, routing and cadence
• Compliance sits above the score
• Setting it up in your CRM and dialer
• Measuring whether the model works
• Using call scoring as a second signal
• Where RevRing.com fits
• FAQ
• Sources

What lead scoring does on a call floor

Strip the theory away and a score answers two operational questions: how quickly should we respond, and who owns the next step. That is how GravityCerts frames insurance lead scoring, and it’s the right test for any model you build. If your score doesn’t change a dial order or an owner, it isn’t doing anything.

The problem it solves is familiar. A batch of purchased leads, web forms and live transfers lands in the queue. Your best closer is on a call. Two junior agents are free. Without scoring, whoever is free grabs whatever is on top. With scoring, the ready-to-buy lead with a clean profile goes to your best available producer in seconds, and the tire-kicker gets a text and a later callback.

Three outcomes follow when it works:

1. Speed goes to the leads that justify it. Fast first contact is expensive in producer time. Scoring spends that time where it pays.
2. Ownership is clear. No lead sits in a shared pool because everyone assumed someone else had it.
3. Weak leads still get handled. They go into a defined sequence instead of being ignored or burning dials.

Fit and intent: the two halves of the score

The strongest models combine fit and intent. Fit measures whether the prospect matches what your agency can write and place. Intent measures urgency and engagement. Treating a high-fit, low-intent lead the same as a high-intent, medium-fit lead produces bad follow-up priorities.

Fit signals

Fit is about whether you can quote and bind this person with the carriers you hold. Typical signals for a P&C or health-focused floor:

• State and license match: is the lead in a state where you have a licensed agent available?
• Product match: did they ask for a line you actually sell?
• Carrier appetite: does the risk profile fall inside what your carriers will write?
• Complete data: valid phone, correct ZIP, usable details for a quote.
• Household or risk details that fit your target book, such as homeowner status or multiple vehicles.

Intent signals

Intent is what the lead did and how fresh it is:

• Source type: a live transfer or inbound call is higher intent than a shared web form.
• Recency: minutes old beats hours old, and hours old beats days.
• Actions taken: requested a quote, specified coverage types, submitted during business hours, replied to a text.
• Engagement after capture: picked up a callback, opened an email, clicked a link.

Use your own conversion data to set the weights. Published point tables from vendors are starting points. Your carriers, states and lead sources are different from anyone else’s.

Source and consent quality

If you buy leads, add a third group. Score the vendor, not just the person. Does every lead arrive with a consent record, a timestamp and a source URL? Is the lead exclusive, or sold to several agencies? How old is it when it hits your system? A shared lead that was sold to five agencies five minutes ago should score lower than an exclusive lead from a source you’ve tracked for months, whatever the form fields say.

This is also where the Lead Marketplace and similar sourcing decisions tie into scoring. Track close rates by source, and let those rates feed your source weights.

A working model: score bands and the action behind each

Keep the math simple enough that a sales manager can explain it on a whiteboard. A 0 to 100 scale with three bands works for most agencies. The GravityCerts guide suggests this split: above 70 gets immediate producer outreach, 40 to 69 enters a defined follow-up sequence or intake review queue, and below 40 gets an acknowledgment and a helpful next step. For each tier, define ownership, response time and next action.

Below is an example setup. The numbers are illustrative, not benchmarks. Tune them against your own results.

An example scoring sheet

Here is how a sheet might look for an agency selling personal lines. Adjust every number to your book.

Notice the last two rows. Some signals shouldn’t be points at all. They’re gates. A lead you can’t legally contact or can’t write should not be able to “earn” its way into the hot band through other points.

Turning scores into response time, routing and cadence

A score only matters once it controls the queue. Three levers do the work.

Response time

Set a first-contact target per band, and measure it. Hot leads get the shortest target, and many floors aim for the first few minutes. Our speed-to-lead playbook for insurance agencies walks through how to hold that standard on a live floor. The cost of delay is real, but treat any published conversion multiplier as directional. Most come from vendor studies on different data than yours. Your own contact rate by minutes-to-first-dial is the number to trust.

Routing

Match the lead to the person:

• Hot: route to your top producers who are available right now. If none are free, fall to the next producer rather than letting the lead wait.
• Warm: standard round robin among available agents, filtered by license state and product.
• Cold: automation first. Producer time is committed only after a reply or a callback.

Skill, geography and license filters matter as much as score. A hot lead routed to an agent who isn’t licensed in that state is a wasted hot lead. If you’re building routing rules, our write-up on routing, precedence and fallback covers the order of operations, and the lead management and routing page shows how those rules are set up in RevRing.com.

Cadence

The first call rarely connects. Define the attempt pattern by band before the lead arrives, so agents aren’t improvising. A common structure is a call, a text, a short voicemail, then spaced retries over the first week, with hot leads getting denser early attempts. Fix the script for each touch so the voicemail and the text sound like the same person. Our follow-up script for calls, voicemails and texts gives you wording to start from.

For cold leads, automated email and text sequences carry the load. If you set up automated messages, keep them readable and specific; this piece on automated follow-up emails covers keeping them human.

Compliance sits above the score

A score ranks leads. It doesn’t grant permission to contact them. Build these checks as gates that run before any dial or text, whatever the band.

• Calling hours. Federal rules restrict telephone solicitation to residential subscribers before 8 a.m. or after 9 p.m. at the called party’s location, not where your floor sits. Some states set tighter windows, so your rules need to be state-aware. A hot lead in Pacific time at 7:30 a.m. Eastern still waits.
• Consent. Keep the consent record, timestamp and source for every lead, and know what consent covers texting and autodialed calls to cell phones.
• Do Not Call and opt-outs. Scrub before first outreach, and honor opt-outs immediately across calls and texts.

These are general insurance workflow practices, not legal advice, and no software setting guarantees compliance. Have your compliance counsel confirm the rules for your states and lead sources. For a deeper checklist, see our pieces on TCPA compliance for outbound calling and telesales compliance.

One scoring tip follows from this. Make “complete consent record” a field on the lead, and let the queue exclude leads missing it. Then your fastest callers never touch a lead that shouldn’t be dialed.

Setting it up in your CRM and dialer

You can run this in a spreadsheet for a week, but it belongs in your CRM and dialer, where the score can move a lead without a human. A practical setup order:

1. Define fields. Source, lead age, state, product, consent status, and a score field.
2. Write the rules. Points for each signal, plus the gate rules for consent, state and DNC.
3. Map bands to queues. Hot goes to a priority queue with a top-producer list. Warm goes to the standard queue. Cold goes to automation.
4. Set the dial mode by band. Hot leads suit a preview or power approach where an agent is ready to talk the moment the lead connects. Larger warm volumes can run through progressive or predictive dialing once consent and DNC checks are in place.
5. Trigger automations. A reply to a text raises the score and moves the lead up a band. Three missed attempts lowers priority and shifts it to nurture.
6. Log everything. Every touch, outcome and score change should be on the lead record, so you can audit later.

If your CRM can’t hold a score field and trigger actions from it, that’s the first thing to fix. Agencies on a different CRM can keep it and connect it to the dialer; the score logic works the same either way.

Measuring whether the model works

A model that looks sensible can still rank leads wrong. Check it with data.

• Conversion by band. Hot should close at a higher rate than warm, and warm higher than cold. If bands overlap, the weights are off.
• Contact rate and time to first dial by band. A hot band with slow contact means a routing or staffing problem, not a scoring one.
• Lead to appointment or quote time, plus cost per acquisition by source.
• Holdout test. A randomized holdout comparison over 30 to 90 days shows whether scored routing beats unscored routing on your own leads.

Read funnel numbers stage by stage. A low contact rate points to speed and follow-up consistency. A low quote rate points to the opening and discovery. A low close rate points to presentation and objection handling. Scoring helps most with the first of those. It can’t fix the other two.

Watch for false positives: leads scored hot that don’t convert. If more than 30% of your hot leads fail, the threshold is likely too low. Also check false negatives, the leads scored cold that later bound. Those tell you which signals you’re underweighting.

How often to review

Treat the model as a living tool. DataOps Group recommends a standing quarterly review of conversion correlation, score distribution and false positive and negative rates, then adjusting point values and thresholds. They also warn that a model accurate early in the year can be misleading a year later, and suggest backtesting a revised model against the previous 12 months of data before going live. For an insurance floor, add triggers for events: a new lead vendor, a carrier appetite change, open enrollment season, or a new state license.

Using call scoring as a second signal

Lead scores tell you who to call. Call scores tell you how the call went. Pair them. If hot leads convert poorly for one agent but well for another, the lead score isn’t the problem. AI call scoring and coaching can flag talk patterns, missed discovery questions and compliance misses across every call instead of the handful a manager can review by hand. For the scorecard design, see our guide to AI QA scorecards, and for playbooks that tie it together, our insurance sales playbooks.

Our take on scoring

Most agencies we talk to already know which leads are good. The trouble is that knowledge lives in a manager’s head, and the queue doesn’t reflect it. We think scoring works when it is boring: a handful of signals, three bands, and an owner and a clock on each band.

We also think the score has to live where the calls happen. When CRM, dialer, texting, routing and compliance checks share one record, the score can move a lead between queues the moment something changes, and the consent and DNC gates run before the dial instead of after. That is how we built RevRing.com, and it’s why we push agencies to put scoring inside the workflow rather than in a spreadsheet beside it.

Start small. Score on five signals, review the bands in a quarter, and add signals only when the data says to.

Marc

Where RevRing.com fits

RevRing.com puts CRM, dialing, two-way texting, lead routing, AI call scoring and compliance tooling on one platform. Agencies can run on RevRing CRM or connect the CRM they already use, such as HubSpot, Pipedrive or GoHighLevel, at no extra cost. DNC and litigator checks and calling-hours rules by time zone run before calls, and routing supports round robin, skill, geo and ping-post rules. Those are tools to support your process; your compliance responsibilities stay with your agency.

Published per-seat pricing runs from $49.99 per month for Basic Dialer to $189.99 per month for Pro, with Scale at $139.99 per month. AI call scoring is $0.20 per scored call. Details are on the RevRing.com pricing page.

If you want to see score bands drive a live queue, look at the RevRing.com dialer and the lead management and routing tools, then schedule a live demo.

FAQ

What is a good lead score threshold for hot leads?

There’s no universal number. A common starting point on a 0 to 100 scale is above 70 for immediate producer outreach. Then check your hot band’s conversion rate and false positive rate each quarter and move the line until hot leads clearly outperform the rest.

Should purchased leads and web leads use the same scoring model?

Use one model with source as a weighted input. Live transfers and exclusive leads earn more points than shared leads. Add consent completeness and lead age as gates or heavy weights, since those vary most by vendor.

Can a high score override Do Not Call or calling-hour rules?

No. Treat compliance checks as gates that run before any call or text, whatever the score. Confirm the rules for each state you sell in with your compliance counsel.

How many signals should a new model start with?

Five to eight. Source, lead age, state and license match, product match, and engagement are enough to start. Add more only when your conversion data shows a signal separates winners from losers.

Sources

• A Guide to Insurance Lead Scoring That Works, GravityCerts
• Lead Scoring Pitfalls to Avoid, DataOps Group
• AI Lead Scoring for Agencies: Real 6-Month Results, Dream Code Labs
• 47 CFR 64.1200, Delivery restrictions, eCFR
• Automated Follow-Up Emails: Set Them Up and Keep Them Human, Bass Media