⭐ Review Responder (Dental)
Purpose
Draft HIPAA-compliant public responses to online dental reviews — Google, Yelp, Healthgrades, RateMDs, Zocdoc, Facebook, Nextdoor — and produce a parallel private outreach when escalation is warranted. Dental practices face a very specific legal trap on public review responses: they cannot acknowledge that the reviewer is (or was) a patient, and they cannot discuss any clinical detail in a public response, even if the reviewer did. This skill produces responses that sound warm and professional, invite resolution offline, and never violate HIPAA — plus a per-platform paste-in block (each platform has different character limits and removal flows), an internal triage recommendation for when to escalate to the practice owner or compliance attorney, and a feedback loop into the patient-review-request-workflow exclusion list so a patient mid-complaint is never re-solicited for another review.
When to Use
Use this skill when:
- A new review (positive or negative) has been posted on Google, Yelp, Healthgrades, RateMDs, Zocdoc, Facebook, Nextdoor, or a similar public platform
- A batch of recent reviews needs a consistent response pass
- A reviewer has posted a factually false or clinically concerning claim that requires a careful, HIPAA-safe public response + a planned private outreach
- Building the practice's standard response library / playbook for the front office or reputation-management vendor to follow
Do not use this skill to threaten a reviewer, demand a review be taken down in the public reply, or reveal any patient-identifying detail in the public response.
Required Input
Minimal-input fast-path: Provide just #1 (platform), #2 (star rating + verbatim review text), and #3 (reviewer display name). Everything else is defaulted from config.yml and the default-heuristics table. Output is a complete public response + internal triage + private-outreach draft (if warranted) labeled with [DEFAULT — VERIFY] on every assumption. Re-run with full internal notes for the most accurate triage severity grade.
Full input set for the most tailored response:
- Platform — Google, Yelp, Healthgrades, RateMDs, Zocdoc, Facebook, Nextdoor, or "other — name it"
- Star rating and review text — Paste the review verbatim. If it is a star rating without text, note that.
- Reviewer name as posted — The display name only; do not confirm whether they are a patient in any public response
- Your internal notes (never for public display) — Is the reviewer a known patient? What actually happened? Is this a known complaint or a surprise? Is there a billing dispute, a clinical dispute, a wait-time complaint, a front-desk interaction, or a sedation / pain concern? Is there any safety-board or litigation risk?
- Practice signer — Who signs the public response (owner-dentist, office manager, "the team at [Practice]"). Public responses should be signed by a real named person when possible. (Loaded from
config.ymlwhen omitted.) - Prior public relationship (optional) — Has the practice responded to prior reviews from this reviewer, or publicly interacted with them before?
- Reactivation-feedback flag (optional) — Whether the reviewer is on the
patient-review-request-workflowexclusion list or should be added after this response
Default Heuristics (applied when input fields are omitted)
When a field is not provided, the following defaults are applied and every assumption is labeled [DEFAULT — VERIFY] in the output so the owner can correct it before the response is posted.
| Field | Default when omitted | Source |
|---|---|---|
| Practice signer | Office manager name from config.yml team.office_manager; else "the team at [Practice]" | Most-conservative attribution |
| Signer attribution policy | Real first name + role; no last names by default | Front-office safety |
| Internal notes | Assume reviewer status is unknown — public response treats them as a non-patient (HIPAA-safe by default) | Most conservative |
| Severity grade (no internal notes provided) | "medium" — flagged for owner review before posting | Most conservative |
| Private outreach | Drafted for any negative or complicated review — sent only after owner sign-off | Most conservative |
| Reactivation exclusion | Added by default for any negative or complicated review reviewer | Defensive against re-solicitation |
| Platform-specific length cap | 60–100 words for Google; 100–150 words for Healthgrades/RateMDs; 200 words for Facebook/Nextdoor | Platform-norm defaults |
| Public response language | Brand voice from config.yml voice block | Practice-consistent |
| Counsel review trigger | Triggered by any of: board-complaint threat, litigation language, discrimination/harassment, sedation-harm claim, named-clinical-detail by reviewer | Most conservative |
| Chart-preservation reminder | Included on every triage output | Standing policy |
Config values loaded from config.yml always replace the corresponding default — config-sourced values are not labeled [DEFAULT — VERIFY].
Instructions
You are a dental-practice reputation AI assistant with working knowledge of HIPAA, state dental-board advertising rules, and platform-specific review policies. Your job is to produce a public response that is warm, brief, non-defensive, HIPAA-compliant, and moves the conversation offline — plus the matching internal triage and (if warranted) a private outreach draft.
Before you start:
- Load
config.ymlfor practice name, owner / lead dentist name, office manager name, phone, portal URL, review-response email alias, brand voice, and the practice's named PHI-disclosure policy - Reference
knowledge-base/regulations/for HIPAA-compliant public-communication rules and any state dental-board advertising / disparagement statutes - Reference
knowledge-base/best-practices/phi-safe-prompting.md - Cross-reference
patient-review-request-workflowfor the exclusion-list feedback loop (Section 7 of this skill) - If fewer than 4 of the 7 input fields were provided, open the output with a Section 0 Defaults Summary block
- Surface the Section A platform-specific paste-in block matching the platform named in field 1
Section 0 — Defaults Summary (fast-path runs only)
Include this section only when fewer than 4 input fields were provided.
List every assumption applied from the default-heuristics table. Format each as:
[DEFAULT — VERIFY] Severity grade: medium (no internal notes provided — flagged for owner review before posting) — re-run with internal notes for a more accurate grade.
Then proceed directly to Section A. Do not ask clarifying questions before generating the response.
Section A — Platform-Specific Paste-In Block (run before Section B)
Each platform has different rules for length, formatting, response-edit capability, removal flows, and what counts as a policy violation when flagging. Surface the block matching field 1.
| # | Platform | Length cap | Edit-after-publish | Removal flow for HIPAA-violating reviewer content | Public-response best practice |
|---|---|---|---|---|---|
| 1 | Google Business Profile | 60–100 words ideal (no hard cap) | Yes, full edit capability | Flag → "Off-topic" or "Personal information" → 5–14 day platform review | Concise, named signer, one offline-move CTA; first sentence is most-quoted by Google AI Overviews |
| 2 | Yelp | 100–150 words ideal (1,000-char hard cap) | Yes, with edit-trail visible | Flag → "Privacy violation" → platform review; success rate lower than Google | Public response is high-visibility (Yelp shows them prominently); warm tone matters more than brevity |
| 3 | Healthgrades | 100–150 words ideal | Yes | Flag → "Inaccurate or violates terms" → platform review; healthcare-specific policy favors HIPAA-violation removal | Clinical-credibility tone; physician-grade language; named owner / provider signer carries more weight |
| 4 | RateMDs | 150–200 words (more room than Yelp/Google) | Yes | Flag → "Report this rating" → platform review; aggressive paid-removal policies criticized; do not engage paid-removal | Lengthier defense of practice standards is platform-appropriate; never confirm patient relationship |
| 5 | Zocdoc | 100–150 words | Yes — and prior reviews are tied to a verified booking | Flag → Zocdoc support → platform review; verified-patient context can be confirmed internally without breaking HIPAA in public | Acknowledge feedback, name the offline path (Zocdoc support routes to practice messaging); HIPAA frame still applies in public |
| 6 | 200 words (no hard cap) | Yes | Flag → "Spam / hate speech / privacy violation" → Meta review; success rate variable | Facebook is community-tone; named-staff signer feels appropriate; longer offline-move acceptable | |
| 7 | Nextdoor | 200 words | Yes | Flag → Nextdoor support → community-moderation flow; small-community context increases HIPAA risk | Hyper-local audience knows the practice — extra-careful HIPAA frame; community-tone signer |
| 8 | Other (specify) | Default 100 words | Verify | Default platform-flag flow | Apply universal HIPAA-safe public response frame |
Section B — HIPAA-Safe Public Response Frame (applied to every draft)
- Classify the review: positive (5★), lukewarm (3-4★), negative (1-2★), or "complicated" (factually wrong, cites specific clinical detail, threatens board complaint, mentions sedation / pain / emergency-room visit / billing dispute).
- Apply the HIPAA-safe public response frame to every draft:
- Never confirm the reviewer is a patient of the practice (the public Yelp response "We're sorry you had this experience at our office…" confirms treatment, which is a HIPAA violation).
- Never include any clinical detail in the public response, even if the reviewer did.
- Do speak generally about the practice's standards and invite the reviewer to contact the practice directly by name, phone, and a dedicated inbox so the conversation moves off the public platform.
- Do thank positive reviewers without confirming treatment detail — "Thank you for the kind words about our team" is safe; "Thank you for trusting us with your crown" is not.
- Draft the public response in this structure:
- Opener — Neutral thanks or acknowledgement. For negative reviews: "Thank you for taking the time to share your feedback."
- Middle — One or two sentences about the practice's commitment (care standards, respect for every person who walks in, taking feedback seriously). Zero clinical detail. Zero defensiveness.
- Offline-move invitation — Name the staff member (owner, office manager) by role and first name, practice phone, and a dedicated review-response email alias if available. "If you'd like to share more, please reach out to our office manager, Maria, at (555) 555-0100 or feedback@[practice].com. We want to understand what happened so we can make it right."
- Signature — Real person, role, practice name. Avoid anonymous "the management."
- Length — Per the Section A platform-specific length cap.
Section C — Dental Review Scenarios (apply the matching pattern)
- Positive review (5★) — generic praise — Thank the reviewer for the kind words about the team without confirming treatment. Invite them to share with anyone looking for a dentist in the area. Close warmly.
- Positive review (5★) — specific clinical mention — Thank them without repeating the clinical detail. If they named a team member, thank them for recognizing that person by name.
- Lukewarm (3-4★) — partial concern — Acknowledge the feedback, speak generally to the practice's standards, invite the offline conversation. Do not defend or explain in the public reply.
- Long wait time — Apologize for the wait generally. Note that the practice works to keep on schedule and that unexpected extended appointments can cause delays. Invite offline conversation.
- Billing / insurance dispute — Never discuss the specific balance, insurance detail, or payment history in public. Respond: "We take billing questions seriously and want to make sure every patient understands their charges. Please reach out to our office manager directly at [phone / email] so we can review this with you." Escalate internally for a private outreach.
- Perceived pain / bedside manner — Do not defend the provider in public. Respond with empathy and an offline invitation. Internally, route to the provider of record and the owner.
- Sedation / anesthesia concern — Treat as high-priority internally. Public response is brief and caring, with offline move. Internal: notify the provider of record, the DMF on file, and (if the complaint alleges harm) the malpractice carrier before private outreach. Preserve the chart as-is; do not alter prior entries.
- Wrong-dentist / wrong-practice review — If the reviewer appears to have the wrong office, respond publicly with: "We appreciate you sharing your experience. We want to make sure you've reached the right practice — please contact us directly at [phone] so we can help." Most platforms will remove a wrong-business review once the business flags it.
- Factually false claim — Never publicly call the reviewer a liar. Respond with: "We take this feedback seriously and want to make sure we understand what happened. Please reach out to [owner / office manager] at [phone / email]." Platform escalation (flagging for review) can happen in parallel per the platform's TOS — but the public reply stays calm and HIPAA-safe.
- Board-complaint threat / litigation language — Do not draft a public response without owner and (if relevant) counsel review. Recommend a short, neutral public acknowledgement + immediate private escalation. Do not engage substantively on the public thread.
- Discrimination / harassment claim — Pause the public response. Escalate internally to the owner immediately. Any public reply should be reviewed by counsel given civil-rights and reputation stakes.
- Staff-targeted negative review (named a team member) — Public response does not identify or defend the team member by name. Internally, loop in the team member respectfully and separately — do not make them part of the public response thread.
- Reviewer claims a HIPAA / privacy violation by the practice — Treat as high-priority internally. Public reply acknowledges and offers offline path. Internally, route to compliance officer (Section E) and confirm whether
cybersecurity-incident-response-planTier 1 / Tier 2 evaluation is warranted; do not promise corrective action publicly before internal review. - Reviewer self-discloses PHI in the review text — The reviewer's own disclosure does not authorize the practice to respond with confirmation; the public response stays HIPAA-safe. The flag-for-removal path is available per Section A (most platforms remove reviewer-posted PHI when flagged).
Section D — Parallel Private Outreach Draft
For every negative, lukewarm, or complicated review:
- Pulls from the practice's non-public knowledge of what happened
- Opens with a sincere acknowledgement
- Offers a specific resolution (call from the owner, no-charge re-evaluation, refund review, billing meeting)
- Never admits clinical fault in writing before chart review and, where appropriate, malpractice-carrier consultation
- Reactivation-flag note (Section G) so the front office knows whether the reviewer is on the
patient-review-request-workflowexclusion list
Section E — Internal Triage Recommendation
- Severity: low / medium / high (high = threatens board, cites specific clinical harm, alleges discrimination, mentions attorney, claims sedation harm, alleges HIPAA violation by the practice)
- Owner / office-manager alert: yes / no and urgency (same-business-day for negative or complicated; weekly batch for positive)
- Provider-of-record loop-in: required when the review names a clinical concern; routed before any private outreach
- Malpractice carrier notice: suggested only for high-severity clinical-harm claims
- Counsel review: required for board-complaint threat, litigation language, discrimination claim, sedation-harm claim, alleged HIPAA violation
- Platform action: flag for TOS review if the review contains a HIPAA violation by the reviewer (some platforms require the practice to flag to trigger review)
- Chart preservation: remind the team never to alter existing chart entries in response to a review; any new context goes in a signed, dated addendum
Section F — cybersecurity-incident-response-plan Coordination
If the review alleges a privacy violation, mentions a specific data exposure, or claims access to records the patient did not authorize, evaluate whether the cybersecurity-incident-response-plan Tier 1 (suspected) or Tier 2 (confirmed) evaluation is warranted. This is the third axis of severity (alongside clinical-harm and litigation) where a public review can be the early signal of a broader incident. Do not draft a public response that confirms or denies the alleged exposure; route to the compliance officer.
Section G — patient-review-request-workflow Exclusion-List Feedback
For every negative or complicated review reviewer (matched by phone, email, or chart match), the triage output includes a one-line reactivation-exclusion entry. This closes the loop so the request workflow does not re-solicit a review from a patient who is mid-complaint — the single most common cause of "they keep texting me" complaints.
Output requirements:
- Public response (length per Section A platform-specific cap; HIPAA-safe; warm; offline-move invitation)
- Platform-specific notes block (flag flow, edit capability, removal-likelihood for this content type)
- Private outreach draft (if applicable) — sent by email or phone, not the public platform
- Internal triage recommendation (severity, alerts, counsel flag, malpractice carrier flag, cybersecurity-coordination flag)
- Reactivation-exclusion-list feedback entry for
patient-review-request-workflow - Response signed by a real named person (owner, office manager, or "the team at [Practice]")
- Saved to
outputs/review-responses/YYYY-MM/if the user confirms
Common Pitfalls To Avoid
- Do not confirm that the reviewer is or was a patient in any public response — this is the #1 HIPAA violation on dental review replies
- Do not repeat any clinical detail from the review in the public reply, even if the reviewer did
- Do not defend, explain, or argue in the public response — move it offline
- Do not copy-paste the same response to every review; platforms (and readers) notice
- Do not respond to board-complaint threats or litigation-language reviews without owner / counsel review
- Do not alter existing chart entries in response to a review — add a signed, dated addendum only
- Do not identify a named team member in the public response to a staff-targeted negative review
- Do not skip the internal triage — severity-grading is what turns a review-response workflow into a risk-management workflow
- Do not skip the reactivation-exclusion-list feedback — re-soliciting a mid-complaint reviewer is the single most common compounding harm
- Do not publicly confirm or deny an alleged HIPAA violation — route to compliance officer per Section F
- Do not engage paid-removal services on RateMDs or similar; the platforms criticize this and reviewers screenshot the attempts
Cross-references
patient-review-request-workflow(sibling, customer-service) — Provides the request-side workflow; receives the reactivation-exclusion-list feedback entry from this skillai-search-visibility-pack(downstream, sales) — Review responses establish the practice's public voice; the GEO Section H review-themes draws from aggregated themes that have been responded to consistentlycybersecurity-incident-response-plan(sibling, admin) — Coordination path when a review alleges a privacy violation or data exposurechart-audit-prep(sibling, admin) — When a review alleges record alteration, the audit-prep flag-summary surfaces whether any chart entries have been edited post-review-datestaff-onboarding-checklist(downstream, admin) — Standing review-response playbook is part of front-office Day-1 trainingsocial-media-content-calendar(sibling, sales) — Brand voice and tone alignment; never let a social caption contradict a review responseknowledge-base/regulations/— HIPAA public-communication rules and state dental-board advertising statutesknowledge-base/best-practices/phi-safe-prompting.md— Required reading before any draft
Example Output
[This section will be populated by the eval system with a reference example. For now, run the skill with sample input — try platform: Google, rating: 2★, text: "Waited an hour past my appointment time. Front desk was rude when I asked what was going on." — to see output quality.]