Back to Review & Reputation Management

Process Runbook / SOP

How the automation works day-to-day: what is automatic, what needs a human, how exceptions are handled, and who to contact.

4 pagesPDF · Operations
FS-DOC-03Operations

Process Runbook / SOP

Review and Reputation Management

[YourCompany.com] · Marketing Department · Prepared by FullSpec · [Today's Date]

This runbook is the day-to-day operating guide for your marketing team once the Review and Reputation Management automation is live. It explains what the system does at each step, what your team is responsible for, how to handle situations the system cannot resolve on its own, and who to call when something goes wrong. FullSpec builds, monitors, and maintains the automation end to end. Your team's only standing responsibility is approving draft responses to negative reviews before they are posted.

01Process overview

The Review and Reputation Management process monitors your connected review platforms continuously, classifies every incoming review by sentiment, generates a brand-voice-aligned draft response, routes negative reviews to a manager for approval via Slack, posts approved responses, and keeps your HubSpot CRM and Notion tracker up to date without any manual copying or logging. A weekly Slack report summarises volume, average rating, and response rate automatically every Monday morning. The only step that requires a human decision is the manager approval for reviews rated two stars or below, which is kept deliberate because the reputational stakes are high.

Process name
Review and Reputation Management
Trigger
A new review is detected on Google Business Profile or Trustpilot (polled every 15 minutes)
Final output
Response posted to the review platform, HubSpot contact record updated, Notion tracker entry completed, and weekly Slack report sent
Agents running
Review Classifier Agent, Response Drafting Agent, CRM and Reporting Agent
Tools involved
Google Business Profile, Trustpilot, Slack, HubSpot, Notion, Gmail
Weekly volume
60 to 120 reviews per month (approx. 15 to 30 per week)
Human checkpoint
Manager approval required for all reviews rated 2 stars or below, actioned via Slack
Process owner
[Your name] — Marketing Manager
Process Runbook / SOPPage 1 of 4
FS-DOC-03Operations

02Step-by-step: what happens and who acts

What you actually need to do: Your team has one active responsibility in this process. When a negative review (2 stars or below) arrives, the Marketing Manager receives a Slack message containing the review text, the AI-generated draft response, and an approval link. You read the draft, edit it if needed, and click Approve or Request Edit. Everything else in the process runs without any input from your team.
Step
What happens
Who acts
Type
1
The automation platform polls Google Business Profile and Trustpilot every 15 minutes. When a new review is detected, the review text, star rating, platform name, reviewer name, and timestamp are captured and passed into the workflow.
System
Automated
2
The Review Classifier Agent reads the review text and assigns a sentiment label (positive, neutral, or negative), a confidence score, and a concise one-line summary of the topic raised. This output determines which downstream path the review follows.
Review Classifier Agent
Automated
3
A new row is created in the Notion reputation tracker automatically, recording the review text, sentiment label, star rating, platform, reviewer name, timestamp, and confidence score. No manual copy-paste is required.
System (Notion)
Automated
4
The Response Drafting Agent generates a contextually appropriate reply using the sentiment classification, star rating, review content, and the brand voice guidelines stored in its configuration. Positive and neutral reviews receive an auto-approved draft; negative reviews receive a draft held for human review.
Response Drafting Agent
Automated
5
For reviews rated 2 stars or below, a Slack message is sent immediately to the designated manager channel. The message includes the full review text, the AI-generated draft response, and a direct link to approve or request an edit. For positive and neutral reviews, this step is skipped.
System (Slack)
Automated
6
The Marketing Manager reads the Slack alert, reviews the draft response, makes any edits needed, and clicks Approve. This is the single human decision point in the process. The target is to act on each alert within 2 hours of receipt to maintain response time targets.
Marketing Manager
Human
7
Once the draft is approved (automatically for positive and neutral reviews, or after manager approval for negative ones), the response is posted to the relevant review platform via the Gmail API or the platform's own posting endpoint.
System (Gmail API)
Automated
8
The CRM and Reporting Agent searches HubSpot for a contact matching the reviewer's name or email address. If a match is found, the review rating, sentiment, and response outcome are logged as a timeline activity on that contact record.
CRM and Reporting Agent (HubSpot)
Automated
9
The Notion tracker row created in Step 3 is updated to mark the review as responded, recording the final response text, the posting timestamp, and whether manager approval was required.
System (Notion)
Automated
10
Every Monday morning, the CRM and Reporting Agent compiles a weekly summary from the Notion tracker covering total review volume, average star rating, response rate, and the most common complaint themes, then posts it to the marketing Slack channel automatically.
CRM and Reporting Agent (Slack)
Automated
Process Runbook / SOPPage 2 of 4
FS-DOC-03Operations

03Handling exceptions

Situation
What the system does
What you do
Review is missing key data (no reviewer name, no text, or only a star rating with no comment)
The Review Classifier Agent logs the review to Notion with available fields and marks the sentiment field as 'Insufficient data'. The Response Drafting Agent generates a short generic acknowledgement response appropriate to the star rating.
No action required unless you prefer to personalise the response. You can edit the draft directly in Notion before it posts, or leave the generic reply to publish automatically.
The same review appears twice (duplicate detection from a platform re-sync or polling overlap)
The system checks the Notion tracker for an existing entry with the same platform, reviewer name, and timestamp before creating a new row. If a duplicate is found, the second record is discarded and no second response is drafted or posted.
No action required. If you notice a duplicate has slipped through (visible as two identical rows in Notion), flag it to FullSpec support at support@gofullspec.com and the deduplication logic will be reviewed.
Reviewer cannot be matched to a HubSpot contact (anonymous review, masked email, or reviewer not in the database)
The CRM and Reporting Agent logs a partial record in HubSpot using whatever reviewer information is available, flags it with a 'Reviewer unmatched' tag, and continues the workflow. The Notion tracker entry is completed in full regardless.
No action required for the automation. If the reviewer later contacts you directly and you can identify them, update their HubSpot record manually and remove the 'Reviewer unmatched' tag.
Manager does not act on a negative review Slack alert within 4 hours
The system sends a second Slack reminder to the manager and copies the process owner. If no approval is received within 8 hours of the original alert, the review is flagged as 'Escalation pending' in Notion and no response is posted until a human acts.
Check your Slack notifications and act on the approval link as soon as possible. If you are unavailable, ask a delegated team member to approve via the same link. Contact FullSpec support if the link has expired or is not working.
A review platform (Google Business Profile or Trustpilot) is unavailable or returns an API error during polling
The automation retries the polling request three times at 5-minute intervals. If all retries fail, the workflow pauses for that platform and sends a Slack alert to the process owner noting which platform is unavailable and the time of the first failure.
Check the platform's own status page to confirm whether the outage is on their side. If the platform is back online within a few hours, the automation will resume automatically at the next poll. If the outage lasts more than 24 hours, contact FullSpec support so the error log can be reviewed and any missed reviews recovered manually.
The Response Drafting Agent produces a response with a low confidence score or the draft is flagged as off-brand by internal review
Reviews where the drafting agent's confidence score falls below the configured threshold are automatically held and routed to the manager via Slack for approval, regardless of star rating. The Slack message includes the confidence score and a note indicating the draft may need editing.
Review the flagged draft carefully before approving. If the draft quality is consistently low for a particular review type or topic, contact FullSpec support so the agent's prompt configuration can be updated. Do not simply approve poor drafts repeatedly, as this affects the system's baseline quality.

04Who to contact and when

Fill in the rows below once your team is confirmed. Keep this table updated whenever a role changes so that escalations reach the right person without delay.

Role
Name
How to reach them
Process owner (Marketing Manager)
[Your name]
[Rep email]
Marketing Coordinator (day-to-day monitoring)
[Your name]
[Rep email]
Backup approver (if Marketing Manager is unavailable)
[Your name]
[Rep email]
FullSpec builder
FullSpec team
support@gofullspec.com
FullSpec support
FullSpec support
support@gofullspec.com
Contact FullSpec support at support@gofullspec.com for: API errors or polling failures lasting more than 2 hours, draft quality issues that recur across multiple reviews, changes to connected platforms or account credentials, adjustments to the escalation threshold (currently set at 2 stars or below), and any situation where a response has been posted incorrectly or needs to be retracted.
Process Runbook / SOPPage 3 of 4
FS-DOC-03Operations

05Ongoing maintenance

When
What to do
Who
Whenever brand voice, tone, or messaging changes
Send the updated brand voice document or a set of revised example responses to FullSpec support. The Response Drafting Agent's prompt configuration will be updated and re-tested before changes go live. Do not attempt to edit the prompt directly.
Marketing Manager + FullSpec
Whenever response templates or approved phrases are revised
Share the revised template text with FullSpec so the drafting agent's baseline examples can be refreshed. Allow up to 3 business days for the update and a brief QA pass before the new templates are active in production.
Marketing Manager + FullSpec
Monthly spot-check (first working day of each month)
Open the Notion reputation tracker and review 10 to 15 recent entries at random. Confirm that sentiment labels look correct, responses read on-brand, and HubSpot contact matches are accurate. Flag any anomalies to FullSpec support.
Marketing Coordinator
Weekly error-log review (every Monday, alongside the automated Slack report)
Review the weekly Slack report for any flagged errors, held reviews, or unmatched HubSpot contacts. If the report does not arrive by 9am Monday, check the Notion tracker manually and contact FullSpec support.
Marketing Manager
Whenever a team member joins, leaves, or changes role
Update the contact table in Section 04 of this runbook. Notify FullSpec support so Slack alert routing and approval permissions can be updated to reflect the new team structure. Do not leave a vacant approver slot.
Marketing Manager + FullSpec
Quarterly volume and performance review (every 3 months)
Compare actual review volume and response rate against the targets set at launch (95%+ response rate, under 2-hour response time for negatives). If volume has grown significantly or new review platforms have been added, contact FullSpec to discuss whether the polling frequency or platform connections need to be adjusted.
Marketing Manager + FullSpec
The most common maintenance issue for this process is an outdated brand voice configuration. When your messaging, tone, or standard phrases change and the Response Drafting Agent is not updated to match, draft quality drops noticeably and managers begin rejecting or heavily editing responses. Treat any brand or messaging refresh as a trigger to contact FullSpec at support@gofullspec.com so the agent configuration is kept current.
Process Runbook / SOPPage 4 of 4

More documents for this process

Every document generated for Review & Reputation Management.

Launch Plan
Operations · Owner
View
ROI and Business Case
Finance · Owner
View
Developer Handover Pack
Technical · Developer
View
Integration and API Spec
Technical · Developer
View
Test and QA Plan
Quality · Developer
View