A benchmark playbook is LocalAmp’s structural map of how content tends to land in your specific niche. It’s the difference between “generic AI LinkedIn writer” and “writer that knows what already works for operators like you.”
This applies to both LinkedIn for businesses and LinkedIn for individuals — same playbook engine, same rules.
What we extract — and what we don’t
We extract structural patterns from operators you admire and posts you’ve liked:
- Hook archetypes (question-led / contrarian / story-led / data-led)
- Length distributions
- Paragraph rhythm
- Opinion-pattern shapes
- CTA registers (soft / hard / no-CTA)
- Formatting habits (bullets, line breaks, emoji density)
We do not extract or reuse:
- Raw post text, sentences, or phrasing
- Specific stories, case studies, or claims
- Names, places, or other identifying details
The playbook lands as abstracted JSON. The post writer reads only the abstracted patterns. Raw pasted text stays in your benchmark library for audit + reprocessing on regeneration, but it never reaches the writer.
“Inspired by patterns, not copied from posts.”
Three sources of input
- Profiles you admire. Up to 10 LinkedIn URLs you tell us are worth learning from. We fetch only public signals — headline, claimed follower count from the bio, posting frequency if mentioned. We don’t access engagement data behind login.
- Posts you like. Pasted post URLs or full pasted post bodies. The more pasted-text examples you provide, the higher the data confidence — and the more your playbook reflects what you actually want to sound like.
- Discovered candidates. Public-web search via Perplexity surfaces LinkedIn profiles in your niche. You pick which ones become benchmarks; we never auto-include results.
Data confidence: low / medium / high
A confidence badge on every playbook tells you what it’s grounded in:
- Low — profiles only, no pasted post text. The playbook can run, but it’s mostly inferred from headlines and metadata.
- Medium — profiles + at least one pasted post body. The playbook can map structural patterns from real text. Recommended baseline.
- High — profiles + multiple pasted post bodies + manually labeled patterns. Highest fidelity to your taste.
The fastest way to upgrade Low → Medium: copy 3–5 LinkedIn posts you genuinely like into the Posts you like tab.
Approving and rejecting candidates
Discovery surfaces candidates with status='candidate'. Nothing reaches the playbook until you explicitly approve them. From the Profiles tab, you can:
- Approve — move into the active playbook
- Reject — drop with a one-line reason (improves future discovery scoring)
- Archive — keep for reference but exclude from the playbook