Every Voicify "Competitor A or Competitor B" battlecard in your library is a stack of claims — claims about pricing, about integrations, about clinical capabilities, about what the competitor cannot do. Most dental AI sales orgs treat those claims as facts the moment they get printed on the card. Six quarters in, the battlecard is repeating yesterday's pricing structure to today's buyer, the integration that the matrix says is missing actually shipped two months ago, and the rep who said it out loud on a Zoom call is now arguing with a prospect who has the competitor's docs page open in the next tab. The fix is not better claims. The fix is an evidence library — a parallel database, one row per claim, that traces every line on every card back to a source, a strength tier, and a re-validation date. The card is the customer-facing artifact. The library is the discipline behind it.
TL;DR
Eight fields per claim. Four strength tiers. One expiration date that drives everything. The Voicify A-or-B battlecard evidence library is a parallel database — one row per public claim on every card — storing claim text, source URL, capture date, strength tier (A: primary source; B: verified customer report; C: inferred from public signals; D: anecdotal, not publishable), reviewer initials, validation date, and an expiration date that auto-flags the claim stale when it lapses. Legal review happens at the tier-and-phrasing layer, not the claim layer, which converts legal from per-claim nightmare into a one-hour quarterly ratification. The quarterly refresh meeting reads expiring rows alongside the analytics quadrant chart. The competitor move response playbook writes rows before publishing claims. The result is a battlecard library reps can speak out loud on a live call without checking a slack thread first — because the source is logged, the tier is visible, and the expiration is enforced.
Why Battlecards Drift From Truth Without a Library
A new dental AI battlecard ships with thirty claims on it, each of them true on the publication date. Three quarters later, the competitor has updated pricing twice, shipped an integration the card says is missing, deprecated a feature the card says they still offer, and renamed a clinical workflow that the comparison matrix tracks by the old name. None of these changes is dramatic individually. Cumulatively, the card is now repeating fiction. Worse, the fiction is being said out loud by reps to buyers who have the competitor's documentation open in the next tab. The rep who reads the card and trusts it is the rep whose deal slips when the buyer says, "Actually, they shipped that last month."
The evidence library is the mechanism that prevents that specific failure. Every claim has a source URL behind it. Every source URL has a capture date. Every capture date has an expiration that triggers re-validation. When the competitor changes their pricing page, the row expires, the card flags stale on the analytics dashboard, and the rep sees the flag before she opens the card for her next call. The card stops being a snapshot of the past and starts being a continuously verified statement of the present.
The Eight Fields
The library is one row per public claim, eight fields per row. The discipline is in the eight, not in the temptation to add a ninth that no one will fill in.
| Field | Format | Why it exists |
|---|---|---|
| Claim ID | card-slug.claim-number | Ties the row to a specific line on a specific card |
| Claim text | Verbatim from card | Self-contained row; no need to cross-reference |
| Source URL or doc | Link + archived screenshot | URLs rot; the screenshot is the durable artifact |
| Capture date | YYYY-MM-DD | Day the source was archived |
| Strength tier | A / B / C / D | Determines publishability and phrasing rules |
| Reviewer initials | Named person | Validation is owned, not collective |
| Validation date | YYYY-MM-DD | Distinct from capture; rows re-validate against same source |
| Expiration date | YYYY-MM-DD | Auto-flags the card stale when lapsed |
The expiration date is the single load-bearing field. Without it, the library is a museum that grows; with it, the library forces a quarterly purge cycle that keeps the card library aligned with reality. Default expirations are sixty days for pricing claims, ninety days for integration and feature parity claims, one hundred eighty days for clinical workflow claims, and three hundred sixty-five days for company-level claims like funding stage or headcount range.
The Four Strength Tiers
The strength tier is what converts the library from a list of sources into a usable policy layer. Two reviewers grading the same source must produce the same tier, which means the tier definitions have to be tight enough that a junior competitive intel analyst and the head of product marketing read the rubric and assign the same letter.
- Tier A — primary source. The competitor's own published material. Pricing page, documentation, security trust page, public security questionnaire response, press release, official partner directory. Tier A claims can be stated as facts on the battlecard and on a live call. Re-validate every sixty to ninety days against the same URL.
- Tier B — verified customer report. A named operator at a named practice has told a named rep something specific, and the rep has logged the conversation in the field intel capture system. Tier B can appear on internal battlecards and be spoken on calls, phrased as "we've heard from operators that…" rather than as flat fact. Re-validate every ninety to one hundred eighty days.
- Tier C — inferred from public signals. Hiring posts implying a feature gap, changelogs implying a deprecation, partnership announcements implying an integration shift, public LinkedIn posts from competitor employees. Tier C claims can appear on internal battlecards only with inference-visible phrasing — "public hiring posts suggest" rather than "they do not have" — and the evidence library row stores the approved phrasing template. Never speak Tier C as fact on a call.
- Tier D — anecdotal. Slack screenshot from a friend at the competitor, hallway gossip at a conference, social-media post without context. Tier D does not appear on the battlecard at all. The library still stores Tier D rows because a D today can become an A in three months when a documented source confirms it, and the library is where the competitive intel lead tracks unconfirmed candidates.
Legal Review at the Tier Layer
Legal review on competitive battlecards is the single most under-invested discipline in dental AI sales orgs, and the evidence library is what makes it tractable instead of a quarterly nightmare. Legal reviews the tier rubric and the phrasing rules once per quarter — Tier A claims pass as a class because the source is documented and re-verifiable, Tier B claims pass with the "we've heard from operators" phrasing, Tier C claims pass with the inference-visible phrasing template, Tier D claims are blocked from publication. That one-hour conversation ratifies the policy. The competitive intel lead then applies the policy across the whole library without re-engaging legal on individual claims.
Compare that to the standard nightmare: legal reviewing forty individual claims, asking for source documentation in each round, the cycle taking six weeks, the resulting cards already three weeks out of date before they publish. The evidence library converts legal from a per-claim bottleneck into a quarterly policy review. The cost is the up-front discipline of writing the tier definitions tightly enough that two reviewers agree.
Connections Into the Rest of the Cluster
The evidence library is the source-of-truth layer that the rest of the Voicify A-or-B cluster reads from. The battlecard governance SOP inherits the expiration-date field as its hard refresh trigger — when a row expires, the card it backs flags stale on the battlecard analytics dashboard. The quarterly refresh meeting reads the library report of expiring rows as a primary agenda input alongside the analytics quadrant chart, so the meeting is governed by two artifacts — what's failing and what's expiring — instead of one. The competitor move response playbook writes new rows into the library before the corresponding card update is published, so a public claim is never on a card without a documented source behind it.
The rep certification program tests reps on whether they can name the tier of a given claim on a card they certify on, which forces reps to look at the library when learning the card and creates a culture where reps understand that "we've heard from operators" is a different statement than "they do not have." The win-loss debrief feeds field intel into the library as Tier B candidate rows, which the competitive intel lead either upgrades on validation or downgrades to Tier D pending corroboration. The comparison matrix and the core battlecard become trustable artifacts because every line on them traces to a row in the library, and any line whose row has expired is visibly marked.
The evidence library is boring infrastructure. It is also the boring infrastructure that makes the difference between a battlecard library reps trust enough to speak out loud and a battlecard library reps quietly stop opening because they got burned the last time a claim was wrong on a live call. The eight fields, the four tiers, and the expiration discipline are what convert a snapshot into a living instrument — and a living instrument is the only kind worth shipping into a competitive market where the truth on the other side of the comparison is moving every week.