It is Monday at 8:45 a.m. An AE opens her calendar and sees 11 demos booked across the week. By Friday, 5 of them will not show. Of the 6 who do show, 2 are end-users who cannot sign anything, 1 is a competitor doing recon, and 1 is a tire-kicker with no budget. That leaves 2 real opportunities. The AE spent 11 hours of selling time to surface 2 deals.
The qualification step in front of that calendar is the highest-leverage process in B2B SaaS sales. Most teams still run it on a 1980s framework. The five fields below are the 2026 update, and they fit inside a 90-second AI voice qualification call.
BANT was built for IBM mainframe sales and it shows
BANT (Budget, Authority, Need, Timeline) is the original qualification framework, codified at IBM in the 1960s when a single deal took 18 months and required a CFO signature. It assumes a known buyer with a known budget line evaluating a known category.
That is not 2026 B2B SaaS. Modern buying happens in distributed committees. Budget often gets allocated mid-evaluation. Categories are new and the prospect may not even know they have a need until the demo. MEDDIC and MEDDPICC (Metrics, Economic Buyer, Decision Criteria, Decision Process, Identify Pain, Champion, plus Paper Process and Competition in MEDDPICC) replaced BANT for enterprise. GPCT (Goals, Plans, Challenges, Timeline) replaced it for inbound SMB.
Both are too long for a voice qualification call. MEDDPICC has 8 fields. A caller will not stay on the line for 8 fields. The five below are the cut-down version that actually fits.
Field 1: Company size predicts everything downstream
Ask for employee count or revenue tier early, in the form of a casual qualifying question, not a form field. "Roughly how many people at the company?" or "Is this a 50-person team or 500?"
Company size predicts: deal size band, procurement cycle length, security review requirements, contract length, expected ACV, the title of the eventual economic buyer, and whether legal will redline the MSA. One number tells you which playbook the AE should run.
A 12-person company will close in 14 days on a credit card. A 1,200-person company will run a 90-day procurement cycle with InfoSec, legal, and a CFO sign-off. The same demo cannot serve both. Routing the wrong demo to the wrong AE is one of the largest unforced errors in inbound SaaS.
Field 2: Role authority is a four-tier scale, not a yes-or-no
Asking "are you the decision-maker?" produces useless data. Almost no one says no, even when they aren't. Ego, defensiveness, and a desire to seem important all push the answer toward yes.
The right capture is a four-tier scale, inferred from how the caller describes their role:
- Decision-maker. Has signing authority. Says "I'm evaluating," "I need to choose a vendor," "I'll be the one to sign."
- Recommender. Will pick the shortlist and influence the decision. Often the head of the function. Says "I'm looking for my team," "I'll bring options to my CTO."
- Evaluator. Doing research, will write a memo or build a comparison matrix. Often a manager or senior IC. Says "we're exploring," "I'm comparing options."
- End-user. Will use the product day-to-day. Has zero signing authority but high product influence. Says "my manager asked me to look," "I'd be using this myself."
An AI voice agent can infer this from open-ended questions like "what's your role in evaluating tools like this?" without ever asking the blunt yes-or-no version. Capture the tier. Route accordingly.
Field 3: Current vendor is the strongest predictor of switch likelihood
Most qualification frameworks skip this. They shouldn't. The single best predictor of whether a prospect will buy in the next 90 days is what they are using right now.
Three states matter:
- Incumbent vendor with active pain. "We use X but it's not working." Highest switch likelihood. The prospect has already done the internal work of admitting the problem.
- Incumbent vendor with mild dissatisfaction. "We use X. It's fine." Lowest switch likelihood despite the meeting being booked. Inertia wins.
- DIY or nothing. "We built our own" or "we haven't started." Variable. DIY usually means a build-vs-buy debate is happening internally. Nothing usually means the trigger event is recent.
Capture the incumbent name verbatim. The competitive matrix that follows changes by incumbent. A prospect coming from competitor A needs a different demo than one coming from competitor B. AEs who walk in blind to incumbent waste the first 15 minutes of the call.
Field 4: Budget signal is a three-state field, asked sideways
Asking "what's your budget?" in qualification is the fastest way to kill the deal. The prospect has not seen the demo yet. They have no anchor. The number they give will be either inflated to seem serious or deflated to negotiate hard, and either way it poisons the AE call.
The right capture is three-state, asked through a proxy:
- Allocated. "Yes, we have budget for this." The proxy question: "Has this been approved as a project, or is it still in research mode?"
- Exploratory. "We're scoping it." Project is real, budget is being assembled, often quarter-of decision.
- No budget. "Just looking." Add to nurture. Do not put on the AE calendar.
A skilled SDR or a well-scripted AI voice agent never says the word "budget" in qualification. They ask about project status. Project status is a much better proxy than a dollar figure.
Field 5: Urgency is a trigger event, not a date
"What's your timeline?" is the BANT question that ages worst. Prospects say "next quarter" because that's the polite answer. It means nothing.
The right capture is the trigger event. Why did the prospect dial in this week, specifically? The four common ones:
- Compliance deadline. SOC 2 audit, HIPAA, GDPR, state-level regulation kick-in date. Hard date, high urgency.
- Vendor renewal. Incumbent contract auto-renews on a known date. Reverse-engineer the buying cycle from there.
- Recent incident. Outage, breach, missed SLA, escalation. Highest urgency, shortest cycle.
- Leadership mandate. New CTO, new VP, new initiative. Medium urgency, often becomes a 60-90 day cycle.
If the answer is "no specific reason, just thought I'd reach out," the lead is real but not urgent. Different cadence. The trigger event is the field most teams forget to capture and most AI voice vendors forget to script.
Capturing five fields in 90 seconds without sounding like an interrogation
The order matters. We covered question-order failures in questions your AI voice agent should never ask first. The qualification version follows the same principle: identification, context, then specifics.
A working sequence:
- "Thanks for reaching out, what's the company you're calling from?" (Captures company name. Tools enrich employee count from there.)
- "What brought you to look at this today?" (Open-ended. Captures trigger event and often the incumbent vendor in the same answer.)
- "And what's your role in evaluating tools like this?" (Captures authority tier without asking the blunt question.)
- "Has this been approved as a project, or are you still scoping?" (Captures budget state without saying budget.)
- "What's the timeline you're working against?" (Captures urgency. The earlier trigger answer informs how seriously to weight the response.)
Five fields, five questions, 90 seconds. The five fields produce a routing score. The voice agent does not score. The voice agent captures. The CRM scores.
Routing logic by qualification score
Three buckets, each with a different downstream action:
- High-score. Decision-maker or recommender, allocated budget, named trigger event, incumbent with active pain. Book onto AE calendar with priority slot, brief AE on incumbent before the call.
- Mid-score. Recommender or evaluator, exploratory budget, weak trigger. SDR follow-up cadence over 2-3 weeks. Do not waste an AE slot.
- Low-score. End-user, no budget, no trigger. Marketing nurture. Email sequence, retargeting, no human touch until score changes.
The routing rule is the point of the qualification. Without it, the five fields are just data hygiene. With it, the AE calendar fills with deals that close.
Do not qualify out too aggressively
The mistake on the other side of bad qualification is over-qualification. Many large deals start with a low-authority caller. An end-user from a 5,000-person company is a real signal even though they cannot sign. The right play is to capture, route to nurture, and have a sales-ops or AE-led account research process find the right person.
Qualifying out the end-user from the 5,000-person company because they answered "I'd be the one using it" is how teams miss seven-figure deals. The score should drive routing, not deletion. Every captured lead stays in the system. The score decides who picks up the phone next.
Five vendor-evaluation questions for AI voice qualification
If you are looking at AI voice agents to run inbound qualification, the five questions to ask the vendor:
- Can the agent capture all five fields (company size, authority tier, current vendor, budget state, trigger event) and write them to specific CRM fields, or does it just produce a transcript?
- Does the agent infer authority tier from open-ended questions, or does it ask the blunt "are you the decision-maker?" version?
- Can the qualification flow be customized per industry without a custom integration project? Legal, real estate, and SaaS sales need different scripts.
- How is the routing score calculated, and can sales ops change the weights without vendor support tickets?
- What happens to low-score leads? Do they enter a nurture cadence or do they get dropped?
A vendor who cannot answer all five at the demo is selling a transcription service, not a qualification system.
The contrarian close
Most qualification frameworks fail because they treat qualification as a gate. The implicit logic: if the lead does not pass, do not waste AE time. The result is teams that filter out 70% of inbound and book the remaining 30% onto AE calendars where 40% of those don't show.
The better mental model is qualification as a router. Every lead is captured. Every lead gets a score. The score decides which channel handles them next: AE, SDR, marketing nurture. Nothing gets dropped. The AE calendar gets the highest-score 10%, not the entire 30% who passed an arbitrary threshold. Show rates climb because the AEs are no longer the catch-all.
The five fields above are the inputs. The router is the system. The voice agent is the data-collection layer. None of the three works without the other two.
Related reading: questions your AI voice agent should never ask first, SDR call automation with AI voice, legal intake schema by practice area for a parallel schema concept, and AI voice CRM integration models for how the captured fields actually land in Salesforce or HubSpot.
To see the five-field capture working on your inbound line, book a call with the team or start free trial. Pricing is on the pricing page.