1. Pricing decision The offer is a B2B AI support-quality assistant that reviews support tickets and surfaces coaching topics for agents, currently running as a six-customer pilot with no direct customer contact built into the product. The buyer is a customer-support leader at a software company with 40 to 150 agents and a small internal quality team, and the decision in front of them is whether to convert from pilot access to a paid subscription. Confidence here is limited by three gaps: there is no data on what these buyers currently spend on QA tooling or headcount, no signal on willingness to pay, and no read on how the four pilot teams outside the two reporting ones are experiencing the product. Any price set now is a starting hypothesis, not a validated number. 2. Value evidence map Customer outcomes: two of six pilot teams report saving roughly six reviewer hours per week; one team identified repeated escalation mistakes earlier than their prior process caught them. Economic or operational effects: the time savings are described in hours, not dollars, and there is no reviewer wage or fully-loaded cost figure supplied, so this evidence should stay in hours rather than being converted into a savings estimate. Proof strength: weak to moderate. The reporting sample is two of six pilot accounts, the reports are self-reported rather than independently measured, and there is no baseline comparison showing what those teams would have found without the tool. Unresolved questions: whether the hours saved get reallocated to higher-value work or simply absorbed, whether earlier escalation detection has any measurable effect on customer retention or satisfaction, and why four of six pilot teams have not reported comparable results. 3. Alternatives and value drivers The realistic status quo is manual sampling or spreadsheet-based scorecards, both of which are labor-intensive and typically cover a small percentage of tickets. The more capable alternative is a broader contact-center platform, which costs more and requires a longer implementation. The offer's plausible value drivers are broader ticket coverage than manual sampling allows and faster turnaround on coaching insight. These drivers weaken if a buyer's quality team already reviews close to 100% of tickets manually, if the AI's coaching suggestions are seen as generic rather than actionable, or if the buyer distrusts an AI-generated read on agent performance without human validation. 4. Pricing architectures Option A — Ticket-volume subscription: the charging metric is monthly ticket volume reviewed, packaged into two or three volume bands. This fits companies whose ticket load scales independently of headcount and avoids penalizing buyers for adding agents. Risk: buyers with lumpy volume may feel mispriced in slow months; evidence needed is the actual ticket-volume distribution across the pilot base. Option B — Flat team subscription by company size band: a single flat fee tied to the 40–75 or 76–150 agent band, not per seat. This is the simplest to sell and onboard within the eight-week window and directly respects the constraint against per-seat pricing. Risk: it undercharges larger, heavier-usage teams and overcharges lighter ones within the same band; evidence needed is usage variance within bands. Option C — Feature-tiered package: a lower base tier covering ticket review and coaching suggestions, with the escalation-risk detection capability as a distinct paid tier. This tests whether buyers value time savings or risk visibility more, directly addressing the stated learning goal. Risk: splitting the escalation feature out may undersell it if buyers see it as core, not optional; evidence needed is which capability pilot teams actually reference when asked what they'd miss most. 5. Packaging and guardrails Core to every package: ticket review and coaching-topic generation, since this is the capability with the most evidence behind it. Positioned as optional or higher-tier: escalation-risk flagging, given it currently has only one team's worth of support. Guardrails: cap or clearly meter ticket volume so heavy users don't erode margin unnoticed, and avoid any billing mechanic tied to agent count that would discourage a buyer from expanding coverage, since that directly conflicts with the stated goal. 6. Validation plan Buyer interviews with quality leads across all six pilot accounts, not just the two that reported results, to learn whether the primary draw is reviewer time saved or earlier risk detection. Sample: all six pilot accounts. Decision rule: if a clear majority cite one value driver over the other, lead the pricing narrative and packaging with that driver. Ethical limitation: avoid leading questions that presuppose either driver is the "right" answer. Proposal test comparing Option A (volume-based) against Option B (flat team fee) presented to prospects at similar company sizes. Sample: at least eight to ten prospective accounts split across the two framings. Decision rule: adopt whichever architecture produces a higher rate of moving to a paid conversation, not just interest. Ethical limitation: disclose accurate current-stage evidence if asked; do not overstate proof of ROI during the sales conversation. Lightweight package-comparison test with the four non-reporting pilot teams to understand why they haven't surfaced comparable time-savings claims. Sample: four teams. Decision rule: if the gap traces to a usage or workflow issue rather than a value issue, address that before finalizing packaging. 7. Recommended next move Start with Option B, the flat team subscription by size band. It is the fastest to implement within the eight-week window, it directly avoids the per-seat pricing the buyer group has flagged as undesirable, and it keeps onboarding light. This is a testable starting point, not a permanent structure — a shift to Option A becomes warranted if ticket-volume data shows wide variance within size bands, and a shift toward Option C's tiering becomes warranted if the buyer interviews show a clear split between time-savings and risk-visibility buyers. Before committing further, the team should learn actual ticket-volume distribution across the full pilot base and get a clean read from all six accounts, not just the two that have reported results so far. Decision table Dimension Current status Evidence strength Weak to moderate, self-reported, 2 of 6 pilot accounts Recommended starting architecture Flat team subscription by agent-count band Core package Ticket review + coaching suggestions Optional/higher tier Escalation-risk detection Key constraint respected No per-seat pricing Biggest open question Time savings vs. risk visibility as primary value driver Next validation step Full pilot-base interviews + ticket-volume distribution Do not claim yet A specific dollar value of time saved Any effect on customer retention, churn, or satisfaction That results from two pilot teams generalize to the full customer base Any conversion rate or willingness-to-pay figure That the escalation-detection capability reliably catches issues at scale