B2B Lead Scoring Criteria: ICP Weights & 75-Point Handoff

B2B Lead Scoring Criteria: ICP Weights & 75-Point Handoff

Last updated: 2026-06-27

TL;DR: - B2B lead scoring criteria split into two buckets: explicit (firmographic fit — company size, industry, job title) and implicit (behavioral signals — demo requests, pricing page visits). You need both; either alone produces a noisy queue. - The 75-point MQL threshold and 90–100+ SQL handoff are industry reference points, not universal rules. Your real threshold is roughly 10–15 points below the average score of your last 30 closed-won deals at first sales contact. - Negative scoring — penalizing personal email domains, competitor companies, student titles, and career-page visits — is the single most overlooked improvement teams make to a broken model. - Build ICP weights from closed-won data, not aspirations. The only reliable template for your ideal customer is your actual customers. - A Marketing Director scores differently depending on where they sit: economic buyer at a 50-person SaaS, or one of twelve influencers at a 3,000-person enterprise. Map the title to the buying role, not the seniority band.

Most B2B lead queues have the same problem. They're full — and most of it isn't ready. Revenue operations managers inherit a CRM with 40,000 contacts, half of which scored above MQL on volume signals alone: email opens, a whitepaper download, a single blog visit. Sales ignores the list. Marketing blames sales. The cycle repeats.

A scoring model built around your actual ICP breaks that cycle. Not because it's magic, but because it forces both teams to agree, in writing, on what "ready" means before anyone makes a call.

This article gives you the full framework: explicit weights, behavioral weights, the logic behind the 75–100-point handoff range, a worked example scoring two real-shaped leads, and a negative-scoring table most teams skip entirely. Take what's here, open a spreadsheet, and calibrate it against your own pipeline data today.

Who this is for: RevOps managers building a model from scratch, demand-gen directors recalibrating one that's drifted, and any marketing-sales alignment team that keeps having the "is this lead ready?" argument in every Monday pipeline review.

What Is B2B Lead Scoring?

B2B lead scoring assigns numeric point values to lead attributes and behaviors to rank contacts by their likelihood of converting. Each lead accumulates a total score; that score determines their funnel stage — cold nurture, MQL (marketing qualified), or SQL (sales qualified) — and drives the next action a human takes.

B2B lead scoring is a prioritization system, not a filter. Its purpose isn't to disqualify leads outright — it's to create a shared, data-grounded language between marketing and sales about what "ready" means. Without that shared definition, both teams operate on instinct and argue about it.

Scoring criteria divide into two types. Explicit signals capture who the company is: firmographics (company size, industry vertical, revenue band), demographics (job title, seniority, department), and fit signals (technology stack, geography). Implicit signals capture what the contact does: which pages they visit, what they download, whether they request a demo or return to the pricing page three times in a week.

A model that weights only one type fails in predictable ways. A perfect-fit company with zero engagement isn't about to buy. A highly engaged contact from a three-person bootstrapped startup rarely converts into meaningful revenue. You need both dimensions working together.

What Makes a Useful ICP for Lead Scoring?

A B2B lead qualification framework built on ICP data is only as strong as the ICP itself. Without a concrete, data-grounded ICP, scoring is guesswork dressed in numbers.

Build it from closed-won deals — not aspirations. Pull your last 100 won deals and find the clusters: what company size, which industry verticals, what decision-maker title, what deal velocity, what source channel. If you have fewer than 100 deals, use 20 and plan to recalibrate once you reach 50. That data is your ground truth. Aspirational markets you haven't sold into yet have no place in a scoring model.

A working ICP for a B2B SaaS product typically covers:

• Company size — employee count or revenue band; the range that predicts your deal cycle and ACV
• Industry vertical — mapped into Tier 1 (core ICP, consistent wins), Tier 2 (adjacent, longer cycle, lower ACV), and out-of-market
• Decision-maker title — mapped to buying role, not just seniority
• Technology stack — tools that signal readiness or integrate with your product
• Geography — matches your actual sales coverage

One thing most teams miss: ICP criteria should be asymmetrically weighted. Being slightly off on company size costs you less than being out-of-market on industry. Build that asymmetry into point allocation from day one. The company that's 600 employees instead of your sweet spot of 200–1,000 is still worth pursuing; the company in an industry you've never won in is not.

B2B Lead Scoring Criteria Examples: The Full Attribute Model

A calibrated starting point for a mid-market B2B SaaS product is below. The explicit model caps at 100 points; behavioral scores stack on top. Adjust the weights against your own closed-won data — these are the right shape, not necessarily your exact numbers.

Explicit Scoring (Firmographic + Demographic Fit)

Attribute	Criteria	Points
Company size	200–1,000 employees	+20
	1,001–5,000 employees	+15
	50–199 employees	+10
	< 50 or > 5,000	+5
Industry fit	Tier 1 — core ICP vertical	+20
	Tier 2 — adjacent, partial fit	+10
	Tier 3 — out-of-market	0
Job title / role	VP, Director, C-suite (economic buyer)	+20
	Manager, Senior IC (influencer/champion)	+12
	IC / unspecified	+5
Revenue band	$10M–$100M ARR	+15
	$1M–$10M ARR	+10
	$100M+ ARR	+8
	< $1M ARR	+3
Geography	Primary sales territory	+10
	Secondary territory	+5
	Outside coverage	0
Tech stack match	2+ integrated tools in stack	+10
	1 integrated tool	+5
	No overlap	0

Implicit Scoring (Behavioral Signals)

Behavior	Trigger	Points
Demo / trial request	Form submitted	+25
Pricing page	Visited ≥ 2 times	+15
	Visited once	+8
Webinar	Attended live	+12
	Registered, no-show	+5
High-intent content	Case study or ROI calculator	+10
Mid-intent content	Whitepaper or guide download	+8
Email engagement	Clicked CTA link	+5
	Opened only	+2
Site depth	≥ 3 pages in one session	+3

Combined score = explicit + behavioral. A contact who hits 80 on explicit fit and submits a demo request reaches 105 — well past the SQL line. A contact who maxes behavioral signals but scores 30 on fit needs a different conversation, not a sales call.

How to Score a Marketing Director: B2B Lead Validation in Practice

A Marketing Director is one of the most misscored titles in B2B lead validation — and the scoring error is almost always in the same direction: teams assign the full economic-buyer points and hand them to sales prematurely.

Marketing Director as a b2b lead validation criteria requires mapping the title to the actual buying role, not the seniority tier. The same title means different things depending on company size and purchase type.

At a 40-person SaaS startup, the Marketing Director often owns the full martech budget, signs the contract, and runs the evaluation. Score them as an economic buyer: +20. At a 2,000-person enterprise, the same title is frequently a champion or influencer — they can recommend and advocate, but procurement, Finance, and a VP above them sign off. Score them as an influencer: +12.

The practical fix: add a qualifier to your title-scoring rules. If company size < 200 employees AND title = "Marketing Director," apply buyer-tier points. If company size ≥ 500 AND title = "Marketing Director," apply influencer-tier points and flag the record for multi-threaded outreach — meaning your BDR should also be looking for the VP of Marketing or CMO in the same account.

This matters more than most scoring guides acknowledge. In enterprise B2B, Gartner's research on buying group dynamics consistently shows that the average purchase decision involves 6 to 10 stakeholders. Sending one email sequence to a mid-level Marketing Director at a 1,500-person company and marking the lead as "worked" is not a qualification process. It's a polite guess.

MQL-to-SQL Handoff: Why the 75–100-Point Range?

The 75-point MQL threshold and 90–100+ SQL handoff aren't arbitrary — they reflect a specific logic about what proportion of your ICP criteria a lead should satisfy before entering sales sequences.

If your explicit scoring model caps at 100 points across six weighted dimensions, a contact needs to satisfy roughly three-quarters of those dimensions before they're worth a marketing nurture push. Getting to 75 without gaming the model requires real firmographic fit across at least four of your six criteria — not one lucky high-value signal like a demo form that was filled out by an intern.

The SQL threshold adds intent on top of fit. A contact with an 80-point firmographic score who has never visited the pricing page or returned to the site isn't a hand-raise. Pushing them to sales too early is the fastest way to burn the relationship and destroy AE trust in the MQL list. Once sales stops trusting the list, you'll spend months rebuilding credibility. Better to set the threshold right once.

Standard Threshold Reference

Score Range	Stage	Action
0–49	Cold	Automated nurture only
50–74	Warm	Targeted content; monitor for behavioral uptick
75–89	MQL	Sales alert; BDR outreach within 48 hours
90–109	Hot MQL	Priority outreach within 24 hours
110+	SQL	Immediate AE assignment

These are a starting hypothesis, not final numbers. HubSpot's annual State of Marketing data shows that teams auditing lead scoring thresholds on a regular cycle consistently achieve higher MQL-to-SQL conversion rates than those who set thresholds once and leave them static. Across B2B verticals, average MQL-to-SQL conversion rates run roughly 13–20% (HubSpot, 2025 State of Marketing). Below 13%? Your MQL threshold is almost certainly set too low. Above 25%? You might be holding back qualified leads too long.

Validate your thresholds every quarter. Run the calculation: take your last 90 days of SQLs, look at their score at the point they were handed off, and find the average. That's your empirical SQL threshold. Set MQL 20–25 points below that.

Negative Lead Scoring: Attributes That Should Actively Reduce a Score

Most teams score positive signals and ignore everything else. That's the gap. Negative scoring applies penalty points to attributes and behaviors that predict poor fit or wasted sales time. It's not punitive — it's signal recovery.

Without negative scoring, a researcher at a competitor company who downloads your whitepaper using a Gmail address will accumulate 15–20 points on passive signals alone and may surface in your MQL queue. Sales calls them, realizes the mistake in 90 seconds, and loses a little more faith in your lead list.

Negative Explicit Attributes

Attribute	Disqualifying Signal	Change
Email domain	Free provider (gmail, yahoo, hotmail)	−10
Company type	Competitor or competitor-adjacent	−30 or disqualify
Job title	Student, intern, academic researcher	−20
Company size	Fewer than 5 employees	−15
Industry	Confirmed out-of-market vertical	−10
Geography	Outside all sales territories	−15

Negative Behavioral Attributes

Behavior	Signal	Change
Unsubscribe	Opted out of marketing	−25 or archive
Career page visits	Multiple /jobs or /careers visits	−15
Form qualifier	"Student project" or "just browsing" noted	−10
Long inactivity	No engagement in 90+ days	−10 (time decay)
Shallow bounces	Single page, < 10 seconds, repeated	−2 per instance

The career-page penalty deserves special mention. Job seekers exploring whether to apply visit the site with genuine curiosity — they'll click pricing, read case studies, even watch demo videos. Without the career-page penalty, they accumulate behavioral points and look exactly like an evaluation-stage buyer. They're not. The penalty is small but consistent, and over a full quarter it meaningfully cleans the queue.

Worked Example: Scoring Two Real-Shaped Leads

Theory is clean. Real leads are messier. Here's how two realistic contacts shake out against the model above.

Lead A — Strong fit, high intent: Jenna holds a VP of Marketing title at a 650-person logistics SaaS. Her company sits in your Tier 1 industry vertical, bills roughly $40M ARR, uses two tools that integrate with your product, and is based in your primary sales territory. She visited the pricing page twice last week and downloaded a ROI calculator.

• Explicit: company size +20, industry +20, VP title +20, revenue +15, geography +10, tech stack +10 = 95 points
• Behavioral: pricing page ×2 +15, ROI calculator +10 = 25 points
• Total: 120 points → SQL, immediate AE assignment

Lead B — Looks engaged, poor fit: Marcus is a "Marketing Intern" at a 9-person consulting firm outside your territory. He used a Gmail address. He downloaded a whitepaper, opened three emails, and visited the careers page twice.

• Explicit: company size +5, industry 0 (Tier 3), intern title +5 (then −20 penalty), revenue +3, geography 0, tech stack 0 = −7 points
• Behavioral: whitepaper +8, email opens ×3 +6, career page ×2 −15 = −1 point
• Negative explicit: Gmail −10, intern −20 (already counted above), out of territory −15
• Total: roughly −23 points → stays in cold, no outreach triggered

Without negative scoring, Marcus accumulates 19 behavioral points and looks like a warm lead.

How to Build a B2B Lead Scoring Model in 6 Steps

Start here whether you're building from scratch or rebuilding one that's drifted out of calibration.

1. Pull your closed-won data. Export your last 100 won deals. Look at company size, industry, primary contact title, deal velocity, and source channel. Fewer than 100 deals? Use 20. Plan to recalibrate at 50. This data is your ground truth — not markets you hope to enter.

2. Define ICP tiers. Group closed-won clusters into Tier 1 (best fit, fastest velocity), Tier 2 (solid fit, longer cycle), and out-of-market. Do not add verticals you've never sold into. Aspirational ICP is a scoring model built on fiction.

3. Assign explicit weights asymmetrically. Adjust the heaviest attributes to reflect the dimensions that most consistently separated won deals from lost ones. If industry was the strongest predictor in your data, give it more weight than company size. If title-to-role mapping was the clearest signal, weight that dimension above revenue band.

4. Map your behavioral signals with sales. Identify the 3–5 actions that most reliably preceded conversion in your own pipeline. Demo request plus pricing page return is the strongest combination for most B2B SaaS products. Weight those highest. Blog opens and single-session visits can score, but keep them near the floor of the behavioral range.

5. Set thresholds empirically. Look at closed-won deals: what was the average score at first sales engagement? Set your SQL threshold 10 points below that average. Set MQL 20–25 points below SQL. Ground the numbers in your actual pipeline, not borrowed benchmarks.

6. Install negative scoring and time decay. Add the penalty rules above. Then add time decay: reduce score by 10% every 30 days of inactivity. HubSpot, Salesforce, and Marketo all support this natively. Without decay, a lead that went cold six months ago still carries a score that no longer reflects their intent — and your queue becomes a historical artifact, not a prioritization tool.

If your b2b lead generation pipeline pulls contacts from LinkedIn, Google Maps, or enrichment tools, look for platforms that export structured firmographic fields — industry, employee count, decision-maker title — that map directly into step 3. That eliminates manual field-matching, which is where data quality degrades before scoring even begins.

Common Lead Scoring Mistakes That Distort Your Pipeline

Well-designed models drift. These are the patterns that cause the most damage.

Overweighting passive engagement. Email opens at 10 points inflate scores for contacts who clicked by accident. Keep the hierarchy steep: demo request (+25) should outweigh a full week of email opens (+10 combined). If your behavioral weight distribution looks flat, the model will produce a flat, undifferentiated queue.

Skipping negative scoring. As the worked example above shows, ignoring disqualifying signals surfaces researchers, competitors, and job seekers alongside real buyers. Sales calls them, recognizes the pattern, and stops trusting the MQL list. Rebuilding that trust takes months.

Building the model without sales input. If sales doesn't believe in the model — or doesn't know what a 75-point MQL means in practice — they'll call leads out of order anyway. This isn't a marketing deliverable. It's a shared operating agreement. Build it in a joint session with the AEs and BDRs who act on it daily.

Setting thresholds once. Buying committees evolve. Product positioning shifts. New channels arrive. A scoring model calibrated 18 months ago reflects a different funnel than you have today. Audit thresholds quarterly against actual MQL-to-SQL rates. Rebuild the full model annually.

Flattening job title scoring. A Marketing Director and a CMO are not the same signal. A Marketing Manager and a Marketing Director at a 30-person company are not the same signal either. Map titles to buying role — economic buyer, champion, influencer, technical evaluator — and weight accordingly. Flattening this dimension produces the exact mismatch the Marketing Director scoring example above describes.

Ignoring lead source as a qualifier. Leads from paid brand search and direct demo-page traffic convert at materially different rates than leads from broad content downloads. Track conversion rate by source in your CRM, and consider source as a multiplier or weight modifier. In practice this often matters as much as a full scoring tier.

How AI Lead Generation Changes the Scoring Equation

AI lead generation tools add a pre-scoring layer that used to happen entirely inside your CRM after import. Instead of pulling 5,000 raw contacts and running scoring logic after the fact, purpose-built b2b lead generation software applies ICP filters at collection — company size, industry, title — before a record is even created.

The result: leads entering your scoring model start with a higher baseline fit score. Behavioral scoring then separates active buyers from the rest of the queue, with fewer false positives already in the system.

According to Salesforce's State of Sales research, high-performing B2B sales teams are significantly more likely to use AI-assisted lead prioritization than average performers. The advantage isn't the AI itself. It's the discipline of defining ICP criteria explicitly enough that a system can act on them — and that discipline produces a cleaner scoring model whether you run it with AI or not.

More advanced implementations use predictive scoring: the system weights attributes automatically based on historical win and loss patterns. It works. But any predictive model still needs to be validated against your actual pipeline data — a black-box score the sales team can't explain is almost as useless as no score at all.

Tools like ConvertFleet export structured LinkedIn and company data — industry, employee count, decision-maker title — in a format that maps directly into explicit scoring fields. That removes the manual enrichment step that usually corrupts data quality before scoring begins. ConvertFleet is currently in pre-launch beta; the Pro plan is free for the first 100 signups.

Frequently Asked Questions

What are the most important B2B lead scoring criteria?

The most important criteria are ICP fit — company size, industry, job title — combined with high-intent behavioral signals: pricing page visits and demo requests. Firmographic fit predicts whether a lead could buy; behavioral signals predict whether they're about to. Neither alone is sufficient. A perfect-fit company with zero engagement isn't ready for sales. A highly engaged contact from an out-of-market company won't convert.

What is a good MQL score threshold in B2B?

Most B2B teams use 75 points as the MQL floor and 90–100+ as the SQL handoff trigger. The right threshold for your business sits roughly 10–15 points below the average score of your last 30 closed-won deals at first sales contact. If your MQL-to-SQL conversion rate runs consistently below 13–15%, your MQL threshold is too low. Benchmark against your own pipeline first, then external data.

How do negative scoring attributes work in B2B lead scoring?

Negative scoring assigns penalty points to signals that indicate poor fit or low intent — personal email domains, competitor companies, student or intern job titles, career-page visits, and extended inactivity. Those penalties reduce a lead's total score, keeping researchers and job seekers out of the MQL queue. Without them, passive signals like email opens inflate scores for contacts with no real purchase intent, and sales stops trusting the list.

How should I score a Marketing Director for B2B lead validation?

Score a Marketing Director based on their likely buying role, not the title alone. At a company under 200 employees, they typically own the budget and sign contracts — score as an economic buyer (+20). At a company over 500 employees, they're usually a champion or influencer alongside a larger buying committee — score as an influencer (+12) and flag the account for multi-threaded outreach to find the VP or CMO. Map titles to your actual buying committee structure before assigning points.

What is a B2B lead qualification framework?

A B2B lead qualification framework is the structured set of rules and criteria that determine whether a lead is worth pursuing at a given funnel stage. Common named frameworks include BANT (Budget, Authority, Need, Timeline), MEDDIC, and CHAMP. The most reliable approach in practice is an ICP-anchored lead scoring model built from your own closed-won data — because it reflects how your actual customers buy, not how a generic template imagines they do.

When should I rebuild vs. recalibrate a lead scoring model?

Recalibrate thresholds quarterly: review MQL-to-SQL rates, adjust the cutoff if conversion is too high or too low, tune the heaviest weights if a scoring dimension is consistently over- or under-predicting. Rebuild the full model — weights, ICP tiers, and all — when your ICP has meaningfully shifted (new segment, product expansion, major pricing change) or when MQL-to-SQL rates have stayed below target for two consecutive quarters despite threshold adjustments. A model more than 18 months old without a rebuild is almost certainly drifted.

Conclusion

A lead scoring model is only as good as the ICP data behind it and the discipline to maintain it. Start from closed-won deals, build explicit weights that reflect the dimensions that actually predicted revenue, add behavioral scoring for intent, and install negative scoring from day one — not as an afterthought.

The 75-point MQL / 90–100+ SQL convention is a reasonable place to start. Treat those numbers as a working hypothesis. Validate them against your own pipeline within the first quarter. Teams that do this consistently — and audit quarterly rather than setting thresholds once — outperform those who borrow a framework and leave it static.

If you're rebuilding your lead data pipeline alongside your scoring model, ConvertFleet can help — structured firmographic exports from LinkedIn and other sources, built for the exact fields your scoring model needs. Pre-launch beta is live; Pro is free for the first 100 teams.

{ "@context": "https://schema.org", "@graph": [ { "@type": "BlogPosting", "@id": "https://convertfleet.online/blog/b2b-lead-scoring-criteria-examples", "headline": "B2B Lead Scoring Criteria: ICP Weights & 75-Point Handoff", "description": "A concrete B2B lead scoring framework with ICP attribute weights, MQL-to-SQL thresholds at 75–100 points, and negative signals that disqualify leads.", "url": "https://convertfleet.online/blog/b2b-lead-scoring-criteria-examples", "datePublished": "2026-06-12", "dateModified": "2026-06-27", "author": { "@type": "Organization", "name": "Convertfleet Team", "url": "https://convertfleet.online" }, "publisher": { "@type": "Organization", "name": "ConvertFleet", "url": "https://convertfleet.online", "logo": { "@type": "ImageObject", "url": "https://convertfleet.online/logo.png" } }, "image": { "@type": "ImageObject", "@id": "https://convertfleet.online/images/hero-b2b-lead-scoring-criteria-examples.png", "url": "https://convertfleet.online/images/hero-b2b-lead-scoring-criteria-examples.png", "contentUrl": "https://convertfleet.online/images/hero-b2b-lead-scoring-criteria-examples.png", "caption": "B2B lead scoring funnel diagram showing ICP attribute weights and MQL-to-SQL handoff thresholds at 75 and 100 points", "width": 1200, "height": 675 }, "keywords": "b2b lead scoring criteria examples, b2b lead qualification framework icp, b2b lead scoring typical thresholds 75 100 points handoff, lead scoring negative attributes b2b, b2b lead validation criteria marketing director", "articleSection": "Lead Generation", "inLanguage": "en-US", "mainEntityOfPage": { "@type": "WebPage", "@id": "https://convertfleet.online/blog/b2b-lead-scoring-criteria-examples" } }, { "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "name": "What are the most important B2B lead scoring criteria?", "acceptedAnswer": { "@type": "Answer", "text": "The most important criteria are ICP fit — company size, industry, job title — combined with high-intent behavioral signals: pricing page visits and demo requests. Firmographic fit predicts whether a lead could buy; behavioral signals predict whether they are about to. Neither alone is sufficient — a perfect-fit company with zero engagement is not ready for sales, and a highly engaged contact from an out-of-market company will rarely convert." } }, { "@type": "Question", "name": "What is a good MQL score threshold in B2B?", "acceptedAnswer": { "@type": "Answer", "text": "Most B2B teams use 75 points as the MQL floor and 90–100+ points as the SQL handoff trigger. The right threshold sits roughly 10–15 points below the average score of your last 30 closed-won deals at first sales contact. If your MQL-to-SQL conversion rate consistently falls below 13–15%, the MQL threshold is set too low. Validate against your own pipeline first, then external benchmarks." } }, { "@type": "Question", "name": "How do negative scoring attributes work in B2B lead scoring?", "acceptedAnswer": { "@type": "Answer", "text": "Negative scoring assigns penalty points to signals that indicate poor fit or low intent — personal email domains, competitor companies, student or intern job titles, career-page visits, and extended inactivity. Those penalties reduce a lead's total score, keeping researchers and job seekers out of the MQL queue. Without negative scoring, passive signals like email opens inflate scores for contacts with no real purchase intent." } }, { "@type": "Question", "name": "How should I score a Marketing Director for B2B lead validation?", "acceptedAnswer": { "@type": "Answer", "text": "Score a Marketing Director based on their likely buying role, not the title alone. At a company under 200 employees, they typically own the budget and sign contracts — score as an economic buyer. At a company over 500 employees, they are usually a champion or influencer within a larger buying committee — score as an influencer and flag the account for multi-threaded outreach to find the VP or CMO. Map titles to your actual buying committee structure before assigning points." } }, { "@type": "Question", "name": "What is a B2B lead qualification framework?", "acceptedAnswer": { "@type": "Answer", "text": "A B2B lead qualification framework is the structured set of rules and criteria that determine whether a lead is worth pursuing at a given funnel stage. Common named frameworks include BANT (Budget, Authority, Need, Timeline), MEDDIC, and CHAMP. The most reliable approach in practice is an ICP-anchored lead scoring model built from your own closed-won data, because it reflects how your actual customers buy rather than how a generic template imagines they do." } }, { "@type": "Question", "name": "When should I rebuild versus recalibrate a lead scoring model?", "acceptedAnswer": { "@type": "Answer", "text": "Recalibrate thresholds quarterly: review MQL-to-SQL rates and adjust the cutoff if conversion is too high or too low. Rebuild the full model — weights, ICP tiers, and all — when your ICP has meaningfully shifted due to a new segment, product expansion, or major pricing change, or when MQL-to-SQL rates have stayed below target for two consecutive quarters despite threshold adjustments. A model more than 18 months old without a rebuild is almost certainly drifted." } } ] } ] }