How to Set Qualification Criteria Before You Pay for a Lead

Who this guide is for

This guide is for buyers, providers, revenue operations, marketing operations, finance, procurement, and product owners who need a commercial qualification unit that can be tested and disputed fairly. It does not replace ICP design, event-routing playbooks, binding terms, or legal review.

Contents

1 Treat qualification as a commercial contract 2 Complete the Qualification Criteria Agreement Canvas 3 Define fit and hard exclusions 4 Specify evidence and Unknown 5 Resolve identity, completeness, and eligible events 6 Define duplicates and Existing Customer treatment 7 Bind price, billing, dispute, and resolution records 8 Test the policy before launch 9 Recalculate an illustrative acceptance sample 10 Govern changes and exceptions 11 Common mistakes 12 Limits 13 Frequently asked questions 14 Sources and methodology

1. Treat qualification as a commercial contract

Qualification serves at least three different purposes:

• an operating decision about lifecycle state;

• a measurement definition for program performance;

• a commercial unit that may trigger a charge. Those purposes can share one policy only when the fields and evidence are explicit. A marketing score designed to prioritize nurture may be unsuitable as a billable-unit rule. A salesperson's judgment that a lead is interesting may be useful but is not a reproducible commercial definition. HubSpot distinguishes fit and engagement scores and their combination. Adobe Marketo guidance recommends behavioral and demographic inputs, negative signals, and sales-marketing agreement on thresholds. HubSpot, "Understand the lead scoring tool" Adobe Experience League, "Build Person Scoring Models for Marketo Engage Programs" These sources support structured qualification. They do not prescribe a universal threshold, billing event, or price. The commercial agreement should therefore start from the SellersGPT Qualified Lead definition and name each applicable rule rather than say "qualified according to industry standards."

2. Complete the Qualification Criteria Agreement Canvas

Use one controlled record with these sections: Canvas section Required decision Minimum output Version Which policy applies and when? Rule ID, version, effective time, owner, approval Scope Which market, industry, application, offer, account, contact, and event paths are eligible? Inclusion statement and exclusions Criteria Which fit and completeness fields must pass or be known? Field, operator, value, applicability, hard/soft type Evidence What source quality and freshness can satisfy each field? Evidence class, recency, corroboration, reviewer rule Canvas section Required decision Minimum output Unknown What happens when evidence is insufficient? Remain Lead, review, or another explicit treatment Exceptions Which hard stops or authorized overrides exist? Type, scope, authority, reason, expiry Identity/dedup What counts as the same person, account, pursuit, and unit? Resolution and deduplication rules Event path Which event can start evaluation? Event types, identity threshold, bot/duplicate handling Billing Which state event creates the charge and how is price resolved? Billable event and pricing-system key Dispute/resolution What can be challenged, by when, with what evidence, and what happens while pending and at resolution? Packet, deadline, reviewer, hold, charge/no-charge, and conditional correction Operations Who reviews, responds, escalates, and stops? Owners, service levels, fallback Change control How do amendments affect existing and new cohorts? Proposal, test, approval, effective time, migration rule Do not publish or activate the policy with material blanks. Mark unresolved items Unknown and block billing readiness.

3. Define fit and hard exclusions

Fit criteria should be inherited from an approved ICP version and translated into testable fields. Examples for an industrial exporter might include:

• operating country or facility geography;

• industry and subsegment;

• application or use-case fit;

• facility, fleet, production, or project characteristic;

• compatible route to market;

• organization scale band where material;

• role or relationship needed for the event path. For each, specify the operator and evidence. Industrial company is too vague. Company operates at least one food-processing facility in the approved region, supported by a current first-party facility page or named buyer confirmation is testable. Hard exclusions sit outside scoring. Examples can include out-of-scope application, prohibited market, invalid identity, or an Existing Customer relationship under the approved SellersGPT rule. A weighted total must not override them. Salesforce's lead-scoring guidance discusses demographic, company, behavioral, and negative or spam signals, plus data quality. Salesforce, "Lead Scoring: How to Find the Best Prospects in 4 Steps" The source supports evidence diversity, not any particular exclusion or score.

4. Specify evidence and Unknown

For every criterion, record:

• acceptable source classes;

• minimum freshness;

• direct versus inferred evidence;

• required corroboration;

• Pass, Fail, and Unknown treatment;

• reviewer and escalation rule;

• evidence retention and access scope. Use three states:

• Pass: sufficient current evidence supports the criterion.

• Fail: sufficient current evidence contradicts it or proves an exclusion.

• Unknown: evidence is absent, stale, ambiguous, or below threshold. Unknown is neither zero nor a pass. The policy may require a field to be known before qualification. It may allow a non-critical field to remain Unknown. That treatment must be explicit before execution. LinkedIn's MQL overview describes lead scoring and B2B stakeholder context but does not create a universal MQL definition. LinkedIn Business Solutions, "What is a Marketing Qualified Lead (MQL)?" The organization remains responsible for its criteria and evidence.

5. Resolve identity, completeness, and eligible events

Identity Define how a person connects to an organization and how the organization maps to the intended account. A company domain, legal name, operating site, and contact role can conflict. Establish a confidence threshold and human-review route for material collisions. Completeness Required fields should depend on the path. A direct reply from a verified work address may supply identity and context unavailable in a page visit. A form may collect structured fields but can contain unverifiable claims. The policy should list required-known fields for each path. Eligible event Name the event types that may start evaluation, such as a valid cold-email-linked landing-page visit, direct reply, or form under the approved SellersGPT path. Define bot/security-scanner checks, automated replies, unsubscribe, wrong-person responses, internal tests, spam, and duplicates. An event is evidence, not automatic qualification. How to Verify a Signal Before Creating a Qualified Lead provides the operational gate.

6. Define duplicates and Existing Customer treatment

A duplicate rule needs four boundaries: 1 Person: Which identities resolve to the same individual? 2 Account: Which legal entities, domains, and sites count as the same organization for this pursuit? 3 Pursuit: Can one record qualify separately for different approved applications or offers? 4 Time: When, if ever, can the same relationship become a new eligible unit? Without those boundaries, the same contact can be charged more than once after a forwarded email, repeat form, alias, or CRM import. Existing Customer is a separate control. Under the approved SellersGPT rule, it blocks outreach and billing for the covered relationship. Define the relationship scope, source, freshness, and what happens when systems disagree. Suppression blocks outreach only within its scope. It is not, by itself, the billing control. Store both facts so a customer status cannot be inferred from a suppression flag.

7. Bind price, billing, dispute, and resolution records

The policy must state:

• the Qualified Lead state-created event that triggers billing;

• required evidence packet and hard-stop evaluation;

• price lookup key and precedence;

• currency and effective price version;

• behavior when price resolution fails;

• dispute reason categories and evidence;

• filing window where the approved product rule applies;

• reviewer, decision, and escalation;

• pending no-deduction treatment, clear or solved month for a rejected dispute, and conditional account-credit treatment only when an already-posted charge requires correction. The current public SellersGPT workflow uses a five-day dispute window only where that approved product rule and effective terms apply. Bind that timing to the displayed product version; do not advertise it as a universal legal right. The latest effective decision in the supplied billing archive says not to deduct the disputed fee while review is open and to bill a rejected dispute in the month it is cleared or solved. An upheld item that has not posted is not charged; use account credit only to correct an already-posted charge under the applicable rule. The archive marks implementation NOT VERIFIED, so pre-launch testing must verify this behavior. The price must be the exact pricing-system value effective for the case. No criterion should contain a hard-coded price. CPC, lead-action CPA, modeled MQL cost, and the USD 5.10 portfolio mean CPC remain separate research metrics.

8. Test the policy before launch

Build a balanced test set, not only obvious passes. Include:

• complete Pass;

• required Unknown;

• explicit fit Fail;

• hard exclusion with otherwise high fit;

• ambiguous person-account identity;

• duplicate event for the same unit;

• same account with a different pursuit;

• Existing Customer match;

• Suppression without Existing Customer;

• event outside the eligible path;

• rule change between event and decision;

• dispute based on evidence available at decision time;

• late correction after billing. Test classification, state, outreach, price lookup, billable event, pending no-deduction state, clear or solved month posting, no-charge outcome, conditional correction, export, and audit history. A small bounded sample can reveal whether the policy is researchable and discriminating even though it cannot prove an entire market. How to Validate an ICP with a Small, Bounded Account Sample provides the sampling discipline.

9. Recalculate an illustrative acceptance sample

Illustrative scenario - explicitly fictional. A ten-record pre-launch test uses one frozen rule. Let P be the actual price that would be resolved for each billable Qualified Lead after launch. No charges occur during this test. Result Records Treatment All required gates Pass 4 Would become Qualified Lead Required evidence Unknown 3 Remain Lead or enter review Fit criterion Fail 2 Do not qualify Existing Customer hard stop 1 Block outreach and billing Correct prospective billable count under the test is 4. Symbolic exposure would be 4P after activation, assuming all other billing conditions pass. If a system incorrectly coerces the three Unknown records to Pass, it produces 7 units and symbolic exposure 7P. The unsupported excess is: 7P - 4P = 3P The error rate against the correct result is 3 / 10 = 30% of the test set. It is also 3 / 4 = 75% excess relative to the correct four-unit denominator. State which denominator is used. If the Existing Customer record were allowed through because a high fit score offset the stop, exposure would rise by another P. That is a gate-design failure, not a pricing problem.

10. Govern changes and exceptions

A criteria change record should include:

• proposed change and reason;

• affected fields, paths, markets, and cohorts;

• evidence and sample test;

• expected classification and economic impact;

• reviewer and approver;

• new version and effective time;

• treatment of in-flight events;

• rollback condition;

• communication and customer acknowledgement where required. Apply changes prospectively unless a controlled correction process says otherwise. Do not retroactively qualify or disqualify historical events simply to improve metrics. Exceptions need scope, authority, reason, expiry, and evidence. An exception is not permission to hide an Unknown. Report frequency and review whether repeated exceptions indicate a flawed policy. HubSpot's MQL-versus-SQL discussion supports the general need for distinct handoff stages. HubSpot, "MQL vs. SQL: What they are and how they differ" It does not determine this policy or a commercial charge. FTC guidance supports evidence-based objective claims. U.S. Federal Trade Commission, "Advertising and Marketing Basics" It does not approve a qualification agreement or replace legal review.

11. Common mistakes

• Using qualified without a rule version.

• Treating a score threshold as sufficient evidence.

• Converting Unknown to Pass or zero.

• Allowing a hard exclusion to be averaged away.

• Using one completeness rule for every event path.

• Omitting person, account, pursuit, or time from deduplication.

• Equating Suppression with Existing Customer.

• Hard-coding a product price in criteria or prose.

• Disputing based on later sales outcome rather than the effective rule.

• Changing definitions mid-cohort without a version.

• Testing only obvious passing records.

12. Limits

This canvas is a product and commercial governance tool. It does not replace binding terms, privacy and communication-law review, ICP design, event routing, sales-stage definitions, accounting, or legal advice. A precise qualification policy improves consistency and auditability but does not guarantee response, meeting, opportunity, sale, revenue, or ROI. Data can remain incomplete or wrong; material uncertainty requires review.

13. Frequently asked questions

Can we use BANT as the Qualified Lead rule? Only if the parties deliberately define and evidence those fields. SellersGPT does not treat a Qualified Lead as a BANT lead by default, and Unknown budget or authority must not be invented. Should every criterion have a score? No. Hard exclusions and required-known gates should remain controls. Scores can prioritize review after the gates pass. What if buyer and provider evidence disagree? Open a review using the same record, rule version, source standards, and decision time. Preserve both submissions and the resulting decision. Can a criteria change affect old charges? Not silently. Apply the new version prospectively unless an authorized correction rule explicitly governs historical treatment. Does a passing test sample prove the policy is accurate? No. It verifies implementation on bounded cases. Continue monitoring real classifications, Unknown rates, disputes, corrections, and drift.

Sources and methodology

The Qualification Criteria Agreement Canvas, test set, symbolic exposure example, and change-control fields are SellersGPT tools. Internal links supply the current lifecycle and product boundaries. External sources support scoring, MQL, handoff, data-quality, and claims principles; they do not create a universal qualification rule or product price. 1 LE-01-INTERNAL-DEFINITION - SellersGPT Knowledge, "What Counts as a Qualified Lead?". Qualified Lead evidence gate. 2 LE-03-INTERNAL-PRICING - SellersGPT Knowledge, "How Pay-Per-Qualified-Lead Pricing Works". Price lookup, billable event, dispute resolution, and control boundaries. 3 ICP-24-INTERNAL-VALIDATION - SellersGPT Knowledge, "How to Validate an ICP with a Small, Bounded Account Sample". Sampling and validation discipline. 4 PP-18-INTERNAL-SIGNAL - SellersGPT Knowledge, "How to Verify a Signal Before Creating a Qualified Lead". Event and evidence gate. 5 QUAL-01 - HubSpot, "Understand the lead scoring tool". Accessed August 25, 2026. 6 QUAL-02 - Adobe Experience League, "Build Person Scoring Models for Marketo Engage Programs". Accessed August 25, 2026. 7 QUAL-03 - Salesforce, "Lead Scoring: How to Find the Best Prospects in 4 Steps". Accessed August 25, 2026. 8 QUAL-04 - LinkedIn Business Solutions, "What is a Marketing Qualified Lead (MQL)?". Accessed August 25, 2026. 9 QUAL-05 - HubSpot, "MQL vs. SQL: What they are and how they differ". Accessed August 25, 2026. 10 GOV-04 - U.S. Federal Trade Commission, "Advertising and Marketing Basics". Accessed August 25, 2026.