What Is a Sales Qualified Lead (SQL)? Definition, Criteria, and When Sales Should Act
The short answer
TL;DR
A sales qualified lead (SQL) is a prospect that the sales organization has investigated and decided is worth actively pursuing, based on the evidence available after qualification — not simply because the lead crossed a marketing or predictive-scoring threshold. In other words, an SQL represents a documented sales decision to commit attention now.
Predictive scoring can help sales decide where to start. Qualification discovers information that was unavailable initially; the business then applies its own evidence and policy. The result may be pursue now, investigate further, nurture, route, or decline.
Every business uses the label slightly differently, but the useful question is consistent: has the prospect earned active human attention? A website form can collect contact details and a score can rank records. An SQL decision adds a reasoned judgment about whether the next sales action is justified. For the broader process, see our guide to what lead qualification means; this article focuses on the sales decision after that work.
MQL, SAL, SQL and opportunity
The stages below are a practical vocabulary, not a universal taxonomy. Some B2B teams use a Sales Accepted Lead (SAL) stage between marketing qualification and sales qualification; others combine or rename stages.
| Stage | Question it answers | Typical decision-maker |
|---|---|---|
| MQL | Does the available marketing evidence justify increased attention? | Marketing / scoring system |
| SAL | Is this lead worth sales investigating? | Sales |
| SQL | Has investigation produced enough evidence to justify active pursuit? | Sales |
| Opportunity | Is there now a defined potential deal being actively progressed? | Sales |
The five SQL criteria
A sales qualified lead is not defined by one score or one framework. These five areas make the decision explicit and give a team a checklist it can adapt to its own market.
- Real problem or need. Is there a meaningful problem the company can solve, and is it relevant to the prospect's situation?
- Credible path to purchase. Are the decision-makers, stakeholders, budget and buying process becoming clear enough to understand how a purchase could happen?
- Commercial value. Is the expected value, margin or strategic attractiveness sufficient to justify the required sales effort where those considerations matter?
- Readiness and timing. Should scarce sales attention be applied now, or is the prospect better served by nurture while the need develops?
- Company-specific fit and policy. Can — and should — this company serve the opportunity, given its capability, geography, risk policy and other hard exclusions?
The checklist is deliberately broader than “how likely is this person to buy?” It asks whether pursuing the opportunity is sensible for this business, at this time, with the evidence available.
When should sales act on a sales qualified lead?
Sales should move into active pursuit when qualification has produced enough evidence that the problem is real and relevant, the opportunity is commercially worthwhile, the prospect has a credible buying path, timing justifies current human attention, the company can serve the prospect well, and no hard company-specific exclusion or policy says otherwise.
A strong prospect with a long timeline can move into nurture instead of occupying salesperson time. “Not now” is not the same as “bad lead.” The purpose of qualification is to make the next human action more useful, not to force every inquiry into a binary accept-or-reject bucket. Automated qualification can reduce avoidable delay between an initial inquiry and useful human engagement by collecting the evidence a salesperson would otherwise have to discover from scratch.
Predictive lead scoring should tell sales where to start — but read the evidence carefully
Predictive scoring and SQL qualification solve related but different problems. A model can estimate which records resemble past outcomes; sales still needs to determine whether the evidence supports action now. Our guide to lead scoring models covers the model choices in more depth.
Wu, Andreev and Benyoucef's 2024 systematic review, The state of lead scoring models and their impact on sales performance, examined 44 studies. It distinguishes traditional scoring built heavily from sales and marketing knowledge from predictive approaches using data mining and machine learning. Across the literature it reviewed, predictive approaches were generally more effective or efficient, but the heterogeneous studies do not establish a universal uplift or guarantee that every implementation will improve conversion.
Source: Wu, M., Andreev, P. & Benyoucef, M. (2024), Information Technology and Management, 25(1), 69–98.
A 2025 case study by González-Flores, Rubiano-Moreno and Sosa-Gómez used CRM data to compare 15 classification algorithms. After cleaning, the dataset contained approximately 16,600 records and 22 fields. Gradient Boosting reported 0.9839 accuracy and 0.9891 AUC, while the dummy classifier reached 0.8816 accuracy because the target was highly imbalanced. The target was “qualified opportunity,” not won revenue.
The result is therefore interesting evidence from one technology-sector case, not a universal promise. “Reason for State/Status” and “Lead Classification” were among the most important features. This raises a potential leakage or temporal-availability problem: if those CRM fields are populated during or after qualification, the model may partly be learning from information that would not have been available when the initial prioritization decision had to be made. The authors also discuss class imbalance, overfitting risk and industry specificity.
A model can be statistically strong at reproducing a qualification label without proving that it predicts downstream revenue or that its score should mechanically determine the SQL decision.
SQL is a decision, not a score
Conversion probability is one input into resource allocation, not the definition of sales qualification. A high-probability prospect may be outside the firm's policy or too early to pursue. A lower-probability inquiry may deserve investigation because the need is urgent, the fit is unusually strong, or the information gap is easy to close.
The same lead can be qualified differently by different firms because each firm has different capacity, economics, service boundaries and risk tolerances. That is why a transparent rules-and-evidence process is more useful than pretending that one score can make the final decision for every business.
A practical rule for teams
Use prediction to decide where to start reading. Use qualification to discover what the first record could not tell you. Then let a human-owned policy decide whether to pursue, investigate, nurture, route or decline. For framework examples, see BANT, MEDDIC and SPIN; for the AI distinction, see AI lead scoring versus AI lead qualification.
For a small service business, the first useful SQL system may be a short set of company-specific questions and explicit scoring rules rather than a sophisticated model. The important thing is that the criteria are visible, the evidence can be challenged, and the eventual decision can teach the business what to improve next. Teams evaluating a commercial tool can read about AI lead qualification software separately.
References
- Wu, M., Andreev, P. & Benyoucef, M. (2024). “The state of lead scoring models and their impact on sales performance.” Information Technology and Management, 25(1), 69–98. https://doi.org/10.1007/s10799-023-00388-w
- González-Flores, L., Rubiano-Moreno, J. & Sosa-Gómez, G. (2025). “The relevance of lead prioritization: a B2B lead scoring model based on machine learning.” Frontiers in Artificial Intelligence, 8:1554325. https://doi.org/10.3389/frai.2025.1554325