Scan, triage, hand off fixes, re-scan — a repeatable loop sized for a $49/mo agency seat and a real calendar.
Agencies do not need a twenty-person "AI search task force" to ship AEO. They need a repeatable loop that fits a calendar: scan, triage, hand off fixes, re-scan. BrandKnown's Agency plan is $49/mo with 10 saved brands and 100 scans/month — enough headroom for about ten active clients if you are disciplined about when you re-scan.
This playbook sizes the loop for a real month, not a conference talk.
The loop in one paragraph
Save the client brand, run a ~60-second scan, sort findings into access / identity / evidence, file tickets the client's team (or your SEO pod) can ship, re-scan after merges, and put only proven website changes in the client report. Do not promise citation rankings. The readiness score is website checks, not AI ranking likelihood.
Monthly capacity math
| Resource | Agency plan fact | Practical use |
|---|---|---|
| Saved brands | 10 | One slot per active client |
| Scans / month | 100 | ~8–10 scans per client if needed; usually fewer |
| Scan time | ~60 seconds | Fine for live calls or async triage |
| First scan | Free | Use for prospects before you spend a seat |
Ten clients × one baseline scan × one mid-month check × one end-month re-scan = 30 scans. That leaves budget for hotfixes and sales demos. If you burn scans on vanity refreshes, you will hit the ceiling.
Weekly calendar (example)
Assume one AEO owner (you) and developers/content owners on the client side.
Week 1 — Baseline and triage
- Confirm each client's canonical site URL and staging rules.
- Run a BrandKnown scan on production.
- Label every finding: Access, Identity, Evidence, or Ignore.
- File tickets with owners and due dates.
- Send a short kickoff note: what you will fix this month, what you will not claim.
Week 2 — Access first
Ship crawler and WAF fixes before copy projects. A blocked PerplexityBot or ChatGPT-User makes content work invisible to live fetch. Re-scan only after production deploys.
Week 3 — Identity and evidence
Align H1 / title / Organization name, About clarity, sameAs profiles, and one answer-shaped page per client. Keep scope brutal: one definition page beats five "AI thought leadership" posts.
Week 4 — Re-scan and report
Re-scan production. Diff score and open findings against Week 1. Write the client AEO report with next-month backlog. Roll unused tickets forward honestly.
Triage table (use on every scan)
| Finding type | Examples | Owner | Ship rule |
|---|---|---|---|
| Access | robots deny, WAF 403, JS-empty shell | Dev / platform | This week |
| Identity | Name mismatch, missing Organization, weak About | SEO + content | This or next week |
| Evidence | Thin answer pages, no FAQs that are real, stale docs | Content | Scheduled, not dump-all |
| Ignore | Cosmetic Lighthouse nits, unrelated CRO | — | Out of AEO scope |
If everything is P0, nothing is. Access always outranks a new blog series.
Handoff format that clients actually complete
Paste this into tickets:
- URL affected
- Current behavior (include status code or screenshot)
- Desired behavior
- Copy-ready fix (robots line, JSON-LD snippet, or paragraph)
- How we will verify (re-scan, curl with user-agent, View Source)
BrandKnown is built around copy-ready fixes for a reason: agencies lose weeks when findings stay abstract.
Multi-client batching tips
- Same CMS cohort: Fix Next.js middleware patterns once, reuse notes across Vercel clients (Next.js / Vercel crawler access).
- Same WAF cohort: One Cloudflare allowlist playbook, ten domains.
- Template the notebook: 20 prompts per client category, reviewed monthly — not daily.
- Protect scan budget: Do not re-scan after every staging commit; re-scan production after meaningful merges.
What to tell sales and account teams
Allowed:
- "We improved crawler access and entity clarity; readiness checks passed on re-scan."
- "We track AI referrals and branded queries as directional metrics."
Not allowed:
- "We guarantee ChatGPT citations by Friday."
- "A score of 78 means you will rank in Overviews."
Honest ceilings keep renewals. Hype creates chargebacks.
Sample month for ten clients
| Day | Work |
|---|---|
| 1–2 | Scans + triage for clients 1–5 |
| 3–4 | Scans + triage for clients 6–10 |
| 5–12 | Access tickets across all accounts |
| 13–20 | Identity / evidence tickets |
| 21–24 | Production re-scans |
| 25–28 | Reports + next-month backlog |
One owner can run this if clients have developers who ship. If every fix waits on a quarterly release train, cut active AEO clients until throughput matches.
Tools in the stack (keep it small)
- BrandKnown Agency seat for scan / re-scan.
- Shared spreadsheet: findings, owners, dates.
- GA4 exploration for AI referrals.
- Search Console branded filter for lagging demand.
- Prompt notebook (optional, sampled).
No need for five overlapping "AI visibility" SaaS tools in month one.
Closing the loop
Scan → triage → access → identity/evidence → re-scan → report. Size it to ten saved brands and 100 scans. Promise website readiness improvements you can prove. Leave citation forecasting off the invoice.