Most Voicify A-or-B dental AI deals that die in procurement do not die from the size of the concession the rep needed. They die from the latency of the answer. A 12 percent discount request that sits in the AE's inbox for six business days, routed informally through a manager who is in the field and a director who is on PTO, gives the competing vendor six days to close while the rep waits for permission to compete. The deal-desk approval matrix exists to collapse that latency to a published SLA per tier, name an approver per tier, and write down — once — exactly which concessions the rep does not have to ask for and which ones never reach the desk at all. It is the governance instrument that keeps the rep's authority defensible to procurement on day one and defensible to the CFO on day three hundred sixty-five.
TL;DR
Five tiers. Five SLAs. Five red-line terms that never reach the matrix. The Voicify A-or-B deal-desk approval matrix publishes a five-tier authority table — AE self-approves at tier 1, sales manager at tier 2 within 4 business hours, director within 1 business day, VP plus deal desk within 2 business days, CRO plus general counsel co-approve tier 5 with no fixed SLA. Five red-line terms (uncapped SLA penalties, clinical-judgment indemnification, MFN pricing, source-code escrow, perpetual derivative rights) live only on the tier-5 row labeled "no precedent set" so a submission triggers a walk-away conversation, not a discount conversation. Auto-escalation enforces SLA. The forecast call scorecard reads deal-desk latency as a deal-health column. The matrix sits downstream of the pricing battlecard and upstream of the mutual action plan, and it is the governance instrument the quarterly refresh meeting audits when win-loss debriefs surface latency patterns.
The Latency Cost Procurement Sees First
The deal-desk approval matrix is not primarily about discount control. It is about latency control. Procurement teams at single-location dental practices, DSO ops directors at multi-location groups, and finance leads at DSO holdcos all run timelines they do not share with the rep, and the rep's discount request lands in their inbox as one of forty open items. When the competing vendor's rep answers the same procurement question in 4 business hours and the Voicify rep answers in six business days, the deal does not slot-flip on price — it slot-flips on perceived organizational seriousness. The matrix collapses the latency window by publishing it in advance, naming the approver in advance, and routing the request through a single intake form that timestamps every step.
This matters most at tiers 2 and 3 — the bands where 70 percent of A-or-B deal concessions actually live. Tier 1 self-approval is the rep's existing authority. Tier 5 CRO escalation is rare. The middle three tiers are where deals cool while the rep waits, and where a published SLA does the most work for the least process overhead.
The Five Tiers in Default Order
Five tiers, ordered by authority and response-time SLA. The default below assumes a mid-market dental AI ICP with deal sizes from $24K to $240K annual. DSO deals and single-location practices use the same tier structure with different threshold values published per segment in a one-row addendum.
| Tier | Authority scope | Approver | SLA |
|---|---|---|---|
| 1 | Up to 5% off year-one list; 12-month term; no security exception; standard BAA | AE self-approves | Instant |
| 2 | 5–12% discount paired with 24-month commit; payment within one quarter; BAA without redline | Sales manager | 4 business hours |
| 3 | 12–20% discount paired with 36-month commit + rate-lock; payment across fiscal years; one non-load-bearing security control exception | Director of sales | 1 business day |
| 4 | >20% discount; multi-location phased rollout; indemnification carve-out short of red lines; MSA redlines from pre-cleared menu | VP sales + deal desk | 2 business days |
| 5 | Anything touching the five red-line terms or any one-off not on the deal-desk menu | CRO + general counsel co-approve | No fixed SLA — walk-away conversation |
The matrix publishes thresholds, not aspirations. Bands move only at the quarterly refresh meeting based on win-loss data, not in-deal. A rep who needs a one-off carve out for a deal in flight submits at tier 5 — the answer is often no, and the discipline of getting that answer in writing protects the next ten deals from precedent creep.
How the SLA Per Tier Actually Gets Enforced
Three mechanisms enforce the SLA. Without all three, the published matrix decays the same way an ungoverned battlecard library decays — see the battlecard governance SOP for the structural parallel.
- Auto-escalation on missed SLA. Tier 2 escalates to tier 3 automatically at the 4-hour mark. Tier 3 escalates to tier 4 at the 1-day mark. Tier 4 surfaces to the CRO weekly digest at the 2-day mark with no automatic escalation (the CRO chooses whether to expedite or hold). Auto-escalation is not punishment — it is recognition that the original approver is unavailable and the rep should not pay for that fact.
- Deal-desk latency as a forecast call column. The forecast call scorecard reads open deal-desk request age as a deal-health signal: an open tier-2 request past 8 business hours grades yellow, past 1 business day grades red, regardless of other dimensions, because the latency is no longer the rep's fault but the deal carries the cost. Surfacing the column weekly creates the social pressure that keeps approvers responsive.
- Quarterly approver-response audit. The quarterly refresh meeting reviews approver-response medians per tier. Any tier consistently running over SLA gets either a process change (route to a backup approver), a threshold change (raise the AE self-approval band so tier 2 sees less traffic), or a personnel conversation. A published SLA the organization cannot meet teaches reps to route around the matrix, and routing-around is how concession discipline collapses.
What Concessions Never Reach the Matrix
Five concessions are pre-decided not at any tier, and the rep does not submit them at all. These are the same five red-line terms from the pricing and procurement battlecard and they appear on the deal-desk matrix only on the tier-5 row labeled "CRO + GC co-approval required, no precedent set" — meaning a tier-5 submission triggers a structured walk-away conversation, not a discount negotiation.
- Uncapped SLA penalties. Liability scales without ceiling; one outage exposes the whole company.
- Clinical-judgment indemnification. Vendor agrees to indemnify for outcomes of clinical decisions a human provider made — uninsurable and category-defining.
- Most-favored-nation pricing. The rep gives away pricing authority on every future deal.
- Source-code escrow with broad release triggers. A vendor-failure clause that releases IP on commercially reasonable definition turns the contract into an IP transfer.
- Perpetual derivative rights to anything trained on customer data. The customer gets ownership of model improvements derived from their data — kills the platform's compounding moat.
A rep who submits red-line tier-5 requests on more than one deal per quarter is a coaching case, not a deal-desk case. The manager coaching cadence picks it up, and the discovery brief gets re-audited for whether the rep is qualifying late-stage red-line risk early enough in the cycle to head it off — usually it is not, and the fix is in discovery, not in the deal-desk routing.
The Intake Form and the One-Page Submission
Every deal-desk request submits through a single intake form with seven fields. The form is the matrix's enforcement surface — a request that does not flow through the form does not start the SLA clock, and an unsubmitted ask is the rep's authority alone. The seven fields are: deal ID, segment (single-location / DSO / DSO holdco), tier requested, specific concession ask in one sentence, justification in three sentences referencing buyer rubric or pilot scorecard data, slot tag (A-slot horizontal voice-AI / B-slot dental-pure / unknown), and close-date impact if approved versus denied. Slot tag matters because tier-3 discount math against an A-slot competitor reads differently than against B-slot — the published bands are slot-aware in the addendum even though the headline tiers are not.
The form auto-routes to the named approver for that tier, copies the rep's manager regardless of tier, and posts to a deal-desk Slack channel so latency is visible to the org. Reps who request below their authority — submitting a 3 percent discount at tier 2 instead of self-approving at tier 1 — get coached, because tier 1 self-approval is the rep's authority and treating it as optional teaches the org to escalate everything, which is how the matrix dies.
How the Matrix Connects to the Cluster
The matrix is the governance layer between the pricing artifact and the deal lifecycle. The pricing and procurement battlecard sets the four traded levers (24-month commit, 36-month commit + rate-lock, multi-location phased, public reference) and the five red-line terms — the matrix encodes them into authority and SLA. The mutual action plan threads the matrix at milestone seven where procurement and legal review run, and a MAP without a deal-desk SLA per submitted concession is a guess at the procurement window. The forecast call scorecard reads deal-desk latency as a deal-health column. The win-loss debrief tags deal-desk latency in any loss that occurred in milestone seven, and the quarterly refresh rebands tiers when the loss pattern points at the matrix rather than the rep.
The deal-desk approval matrix is the artifact that lets a rep say "I can give you that answer by 5pm tomorrow" to a procurement lead — and mean it. That sentence, said and meant, wins the kind of A-or-B deal that competing reps are trying to win by promising something they cannot deliver. The matrix is the boring infrastructure that makes the bold sentence safe to say.