Qualification Signals

What the Chat Needs to Detect and When

Project: AI-powered lead qualification chat
Version: 1.0
Status: Final
Last updated: April 2026

The Question

What qualification signals matter?
Company size, specific needs, timeline, previous AI attempts — what does the chat
need to collect, in what order, and how does it use that information?

This question defines the chat’s internal logic. Not its tone or personality — that is covered in p2-helpfulness-vs-capture.md — but the specific information the system needs to gather in order to determine whether a visitor should be escalated to sales, nurtured, or simply helped without a commercial follow-up.

The Cost of Getting Qualification Wrong

There are two types of error, and they are not equally costly.

False positive — escalating a visitor who is not ready or not the right fit. This wastes sales team time and generates unproductive conversations. Recoverable, but annoying at scale.

False negative — failing to recognise a hot lead and letting them leave without escalating. This costs a real opportunity. For P1 (Evaluating CTO) and P3 (Referred Decision-Maker), this is the most expensive mistake the system can make.

The system should be calibrated to minimise false negatives on high-intent personas, even at the cost of the occasional false positive. Missing a CTO who was ready to buy is worse than sending a slightly premature lead to sales.

The Four Qualification Dimensions

Not all signals carry the same weight. There are four dimensions, ordered by importance. The ordering matters because it defines the sequence in which the chat should build its picture of the visitor.

Dimension 1 — Problem Fit

Does this visitor have a problem that the company can solve?

This is the most important signal. Without a concrete problem there is no lead, regardless of authority, company size, or timeline. A visitor who describes a specific initiative is implicitly signalling that there is work to be done.

Problem fit is the first filter. A visitor who clears it is worth continuing to qualify. A visitor who does not clear it after several exchanges should be treated as a researcher or early explorer, not a lead.

Explicit problem signals

Implicit problem signals

Dimension 2 — Authority Fit

Can this visitor make or meaningfully influence the buying decision?

The company engages on contracts of meaningful size. A conversation with someone who has no authority or influence over the buying decision may have nurture value but should not consume sales team time. The chat needs to build a picture of the visitor’s role without asking directly — “are you the decision-maker?” is a sales script, not a conversation.

Explicit authority signals

Role inference from context

The chat should infer authority from role and language rather than asking directly.

Dimension 3 — Company Fit

Is this a good fit for the company?

Company fit determines whether the engagement makes commercial sense. A solo freelancer and a 500-person scale-up have fundamentally different needs, budgets, and collaboration models.

Target company profiles by persona

Signals

Weak fit signals

The chat should note these but not immediately disqualify. Context matters.

• Very early stage with no funding context mentioned (may still be P2)
• Solo founder with no team yet (low commercial fit, but potential future lead)
• Large enterprise with very rigid procurement processes (may still convert, just slower)

Dimension 4 — Timing Fit

Is there real urgency, or is this exploratory?

Without a timeline, there is no near-term conversion. A visitor with strong problem, authority, and company fit but no urgency is a warm lead for nurture — not for immediate escalation. A visitor with urgency signals should be escalated quickly even if the picture is not fully complete.

Explicit timing signals

Implicit timing signals

The Scoring Model

The four dimensions combine into a three-level qualification model. The chat uses this model to determine what action to take — not what to say, which is governed by P2.

Level definitions

Detection sequence

The system evaluates dimensions in order. This is intentional — problem fit is the most efficient filter and should be resolved first.

Signal Extraction — Explicit vs. Implicit

Visitors do not announce their qualification status. The system must extract signals from natural language, not from direct questions. This has direct implications for how the LLM is prompted and how the conversation is structured.

Examples of implicit signal extraction

What the system must not do

• Ask all four dimensions in sequence as explicit questions. This feels like a form and breaks the conversation model defined in P2.
• Require a visitor to self-identify their role or authority before answering their question.
• Treat absence of a signal as disqualification. Absence means the picture is incomplete, not that the visitor is not a lead.
• Ask about budget directly. Budget is inferred from company size, stage, and initiative scope — never asked.

Edge Cases

Visitor has problem fit and timing fit but unclear authority

Treat as warm-to-hot. Capture email and offer to connect them with the right person at the company. The sales team can determine authority in the first conversation.

Visitor has strong authority but no clear problem

This is the P3 pattern — referred decision-maker who is confirming fit, not describing a problem. Treat as hot based on referral context and authority alone. Ask one open question to surface the initiative rather than disqualifying.

Visitor is a P2 (Exploring Founder) with no timing fit

Expected. P2 rarely has timing fit in the first conversation. Treat as warm, capture email, nurture. Do not push for a call — it will feel premature and damage trust.

Visitor gives contradictory signals

For example: describes a large-scale initiative but mentions being a solo developer. Hold the qualification. Ask one clarifying question before deciding. Do not assume.

Implications for the PRD

• The LLM must be prompted to extract the four qualification dimensions from natural language, incrementally across the conversation.
• The system needs a qualification state object that tracks which dimensions have been confirmed, partially confirmed, or not yet detected.
• The escalation trigger (hot lead) must be programmatic — not left to the LLM to decide ad hoc.
• The chat must never ask about budget directly. Budget fit is inferred, not elicited.
• The sales team should be consulted to validate the hot lead threshold — specifically, whether problem + authority + one more dimension is the right bar, or whether the threshold should be higher or lower based on the company’s current pipeline needs.

Sales Team Findings

The following questions were raised during the design phase and validated by the sales team.

What is the minimum signal set that makes a lead worth a sales team conversation?

Problem fit + Authority fit is the hard minimum. A sales call without both is a waste — no problem means no engagement, and no authority means no decision. Company fit and timing are useful but not blocking: a competent rep can qualify company size and timeline in the first five minutes of a call.

The “hot lead” definition in the scoring model (problem + authority + one more dimension) is one notch above the minimum, which is the right bar for immediate escalation. But if problem and authority are both confirmed and the picture is otherwise incomplete, the lead should still be handed off. The rep handles the rest.

Implication for the chat: The escalation threshold in the scoring model is correct. Do not lower it to problem-only, but do not hold leads back if authority is confirmed alongside a clear problem.

Are there industries or company types that consistently underperform even when signals are strong?

Three patterns stand out:

1. Pre-funding early stage — founders with a genuine problem but no secured budget. Strong qualification signals, no near-term deal. Treat as P2 and nurture; do not escalate to sales.
2. Companies recovering from a bad outsourcing experience — need and budget are present but trust is not. Longer close cycles than average. Worth escalating, but the sales team should set expectations accordingly.
3. Enterprise with rigid procurement — signals are strong but the procurement process adds 60–120 days regardless of fit. Worth escalating, but not a fast conversion.

Industries are not strong predictors on their own. Fintech and healthcare involve more compliance questions but convert well when fit is confirmed.

What is the most common reason a qualified lead does not convert after the first sales call?

Three causes, in order of frequency:

1. Budget not yet secured — the contact has authority to evaluate but not to commit. They are exploring before asking their CFO. The fix is a budget-timeline question on the call, not a budget-amount question.
2. Internal politics — the contact underestimated how many stakeholders are involved. The decision is more complex than the initial conversation suggested.
3. Competitor won on price or vertical specificity — usually a competitor with more case studies in the visitor’s exact industry.

The scoring model does not account for the first case — a visitor can satisfy all four dimensions and still lack internal budget approval. This is a sales call issue, not a chat qualification issue. The chat logic does not need to change.

Should company size be a hard filter or a soft signal?

Soft signal. A 10-person team is a good lead when:

• They are well-funded (post-Seed or Series A with clear runway)
• The founder or CTO is directly involved in the conversation
• The problem is specific and urgent

The target ranges in the Company Fit dimension (50–500 for P1, 10–80 for P2) are probability guides, not gates. The chat should not disqualify based on company size alone. Small companies with strong problem and authority fit should be flagged as P2-profile and routed through the nurture path — email capture first, no immediate push for a call — rather than disqualified outright.

How should the system handle visitors who identify as freelancers or independent consultants evaluating the company for a client?

Treat them as a warm lead with multiplier potential. A consultant recommending vendors to a client is a credible influencer and sometimes a more efficient path than going direct.

Handling sequence:

1. Clarify with one question, not a form: “Are you evaluating this for a specific client’s project?”
2. If yes — qualify on the client’s context, not the consultant’s. Company size, industry, and initiative scope all refer to the client.
3. Offer materials they can share: relevant case studies, a one-pager, or a short deck.
4. Propose a three-way intro call (consultant + client contact) rather than a consultant-only call.

The qualification state passed to the sales team should flag the consultant pattern explicitly so the rep does not pitch the consultant as the buyer.

This document records the strategic decision on qualification signals for the chat system. It defines the four dimensions of fit, the scoring model, and the escalation logic. It feeds directly into the conversation state management and escalation architecture defined in the PRD.