The Briefs System: Claude Remembers Every Client, Every Session
You’ve run three audits for this client over the last two months. You’ve done a geogrid scan, a citation audit, fixed their GBP categories, responded to reviews, and identified a competitor gaining ground in the southeast corner of their service area.
Now you’re starting a new session. Claude doesn’t know any of that. You’re re-explaining the business, re-stating the keyword targets, re-describing the problems you’ve already diagnosed. Half your session is context recovery instead of forward progress.
Every practitioner hits this wall. The work is cumulative — rankings shift, reviews come in, citations get fixed, new competitors emerge. But each session starts from zero because there’s no persistent state connecting them.
What Changes When Briefs Are Running
Mention a business for the first time. Claude asks five setup questions — primary keyword, website, service area, client or owner, notification preferences. Then it runs the initial diagnostic: GBP audit, profile health, geogrid scan, citation check, review velocity, competitor gap.
Everything lands in a brief. One markdown file per location. Structured, scannable, always current.
Next session, you say “Let’s work on Smith’s HVAC.” Claude reads the brief, scans recent output files, and responds:
“Smith’s HVAC Buffalo — last worked April 7. ARP improved from 8.2 to 7.6. SoLV up to 58%. One unanswered 1-star review from yesterday. Citation fixes from last session are live. Next action: respond to the 1-star review, then run a follow-up geogrid to confirm ranking gains. Want to start there?”
No re-explaining. No context loss. Claude knows this client’s history, current state, and what needs to happen next.
See It Work: First Session to Ongoing Management
Session 1 — First mention of the business:
Prompt: "New client — Keystone Insurance, 200 Main Street Buffalo NY.
Target keyword: insurance agent Buffalo."
Claude:
Setting up brief for Keystone Insurance — Buffalo.
1. Primary keyword: insurance agent Buffalo ✓ (from prompt)
2. Website URL?
> keystoneinsurancebuffalo.com
3. Service area radius?
> 15 miles
4. Are you managing this as a client or as the business owner?
> Client
5. Want Slack notifications for alerts?
> Yes, #local-seo-alerts
Running initial diagnostic...
Skills fired: local-seo-audit + gbp-optimization + geogrid-analysis
+ citation-audit + review-management + localseodata-tool
Brief created: briefs/keystone-insurance/buffalo/location.brief.md
INITIAL FINDINGS:
Critical: GBP completeness 54% — missing 6 secondary categories
Critical: 3 citation NAP mismatches (Yelp, BBB, Yellow Pages)
Important: Review velocity declining — 2 reviews/month vs competitor avg 6
Important: No LocalBusiness schema on website
Monitor: ARP 6.8 — ranking but not competitive
Monitor: No AI visibility for primary keyword
Next Action: Add missing GBP categories (highest leverage, 15 minutes)Session 2 — Two weeks later:
Prompt: "Check in on Keystone Insurance"
Claude reads brief, scans recent files:
"Keystone Insurance Buffalo — last worked March 25.
Since then: ARP improved from 6.8 to 5.9 after category additions.
Citation fixes submitted but 2 still showing old address.
Review velocity unchanged at 2/month. 1 new review (5-star, responded).
No new alerts.
Next action: Follow up on pending citation corrections,
then launch review generation campaign. Want to start there?"The brief is the memory. Every session starts with full context and a clear next step.
What the Brief Contains
Identity — business name, address, phone, website, GBP URL, business type, primary category, target keywords, service area. Set once during setup, referenced every session.
Findings — categorized as Critical (ranking suppressors, suspension risks), Important (30-60 day priority gaps), and Monitor (watch items). Updated after every audit, scan, or analysis. Findings move between categories as issues get resolved or escalate.
Session Log — date and one-line summary of what was worked on. Newest first. The running record of every interaction with this client.
Tools Run — date, tool name, endpoint, one-line finding. Which data was pulled, when, and what it showed. Prevents redundant scans and tracks data freshness.
Deliverables — checkbox tracker for pending and completed work. Reports sent, posts drafted, schemas implemented, citations submitted.
Next Action — single explicit next step. Updated at the end of every session. When you return, Claude knows exactly where to pick up.
Multi-Location: Brand-Level Rollup
Agencies managing multi-location clients get a two-tier structure:
briefs/
keystone-insurance/
_brand.brief.md ← Cross-location rollup + config
reports/ ← Brand-level aggregated reports
buffalo/
location.brief.md ← Buffalo-specific state
reports/ scans/ drafts/ alerts/
rochester/
location.brief.md ← Rochester-specific state
reports/ scans/ drafts/ alerts/
syracuse/
location.brief.md ← Syracuse-specific state
reports/ scans/ drafts/ alerts/The brand brief tracks all locations in a summary table — last worked date, critical count, ARP, SoLV, next action per location. Patterns that appear across three or more locations get noted at the brand level.
Run portfolio health checks across all locations in one prompt. See which locations need attention without opening each brief individually.
The Brief Stays Lean
The brief is an index, not a dump. Detailed outputs — full audit reports, geogrid scan data, drafted GBP posts, monitoring alerts — live in subdirectories. The brief points to them.
[2026-04-07] — Weekly scan complete → ARP 7.6 (↓ from 8.2)
→ see scans/2026-04-07-geogrid.md
[2026-04-01] — Monthly report delivered → 3 findings resolved
→ see reports/2026-04-01-monthly.md
[2026-03-25] — Citation corrections submitted (3 of 6 fixed)
→ see scans/2026-03-25-citation-audit.mdYou can read the brief in 30 seconds and understand the current state. Drill into subdirectories when you need the details.
Who Uses Briefs and How
Freelancers Managing 5-15 Clients
Each client has a brief. Start every session by mentioning the client name. Claude loads context, states current priority, and you’re working in under a minute. No spreadsheet of “where was I with this client” required.
Agencies Running 50+ Locations
Brand-level rollups surface which clients need attention. Individual briefs maintain detailed per-location state. Junior staff pick up briefs and have full context without senior staff re-explaining history. Client transitions between team members are seamless.
Business Owners Managing Their Own Presence
One brief for one business. Every session builds on the last. The brief becomes the complete record of optimization work — what was done, what changed, what’s next. No ambiguity about whether something was already addressed.
Without Briefs vs. With Briefs
Without briefs:
"I need to work on my client's local SEO. Their business is Smith's HVAC
at 4521 Main Street Dallas TX. We've been working on their GBP — I think
we fixed the categories last month. Their main keyword is HVAC repair Dallas.
I'm not sure where we left off but I think there were some citation issues..."With briefs:
"Let's work on Smith's HVAC."Same output quality. Different friction. The difference compounds across every client, every session, every week.
How Briefs Connect to the Rest of LocalSEOSkills
Briefs aren’t isolated — they’re the state layer that the rest of the system writes to and reads from.
Scheduled tasks write their outputs into brief subdirectories. A weekly geogrid scan deposits results in scans/, updates the session log, and refreshes the next action. You don’t run the scan manually — the task does, and the brief reflects it.
Approval workflows hold drafts in drafts/ until a human approves them. GBP posts, review responses, client reports — all queued in the brief’s directory structure, tracked in the deliverables section.
Notifications trigger from brief-configured thresholds. A 1-star review, a ranking drop, a GBP edit — alerts fire to Slack or email based on settings in the brand brief.
Every skill that runs updates the brief. Audit findings go to Findings. Data pulls go to Tools Run. Deliverables get tracked. The brief is always current because the system maintains it automatically.
Get Started — Briefs Are Free and Open Source
Briefs are part of the LocalSEOSkills library. MIT licensed. The template files, directory structure, and automation logic are all in the repository.
First prompt to try:
"New client — [Business Name] at [Address]. Target keyword: [keyword]."Claude sets up the brief, runs the initial diagnostic, and populates findings. From that point forward, every session starts with context and ends with an updated brief.
The state layer your local SEO practice was missing.
Skill Documentation
For technical details on brief structure, templates, session maintenance protocols, and multi-location rollup configuration, see the full briefs documentation.