AI agent for social media managers
Brand Voice Drift Audit Agent
Measure voice consistency, fix the drift and keep the guidelines current.
What it does
With five people posting, the brand sounds formal on Monday and jokey on Thursday. The agent takes a sample of recent posts across accounts and writers, and compares each one against the voice guidelines, which cover tone, vocabulary, banned phrases, emoji use and sentence length. It scores each post and finds patterns, such as one channel drifting casual or a new hire not using the preferred terms. It proposes edits for posts that miss and, where the guideline itself is unclear or out of date, proposes changes to the guideline with examples. The next period it takes a new sample and compares scores to see whether the drift narrowed. Edge case: a campaign that deliberately uses a different voice. The manager approves any change to published guidelines.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- Monthly audit starts
- Pull a sample of 40 recent posts across channels and authors
- Remove campaign posts with an approved different voice
- Score each post against the voice criteria
- Find patterns by channel, author and topic
- Is the scoring consistent when 8 posts are scored a second time?If not: clarify the criteria and score the whole sample again. Back to step 3.
- Propose edits for low-scoring posts, examples for the team and guideline changes where rules are unclear
- Manager approves changes to published guidelinesThe agent waits here for your OK.
- Next month, take a new sample and compare scores
- Did the average score and the lowest channel improve?If not: propose more training examples and tighter rules. Back to step 7.
- Voice report
How it decides
Each post is scored out of 10 on the guideline criteria. Scores under 7, or a channel average that falls, count as drift. Unclear criteria lead to a guideline proposal.
- Count a post as drift when it scores under 7 out of 10
- Re-score 8 posts to test the consistency
- Exclude campaigns with an approved voice exception
- Propose a guideline change when 3 posts fail the same unclear rule
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- Sample size (default 40)
- Scoring criteria
- Drift cutoff (default 7 of 10)
- Channels included
- Audit schedule
What keeps you in control
It always asks you first
- Manager approves changes to published guidelines
- Manager approves feedback sent to individual writers
Hard limits
- Never edit published posts itself
- Give feedback on posts, not on people
It stops when
- Done: report approved and next sample scheduled
- Stop: no current guidelines exist, ask the manager to supply them
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide
