Know what’s behind
every number.

A public snapshot. A defined sample. No invented certainty.

Where the data comes from

New profile reviews and public post collection use Bright Data. Historical reports may contain FastFetchz or public-page evidence; each report retains its recorded sources. FastFetchz remains a legacy rollback option, not a second automatic collection route. Free signed-in reviews accept up to eight post URLs. The owner-only Pro preview can request up to 8, 16 or 24 posts and accept supplied public links. A requested depth is a maximum, not a promise that the source can retrieve that many posts or every latest post. We sort dated posts chronologically and only include posts whose authorship matches the requested profile. Missing fields remain unknown and truncated text remains partial. We never use your LinkedIn login or access private messages.

What we measure

• Followers: the public follower count returned at collection time.
• Engagement rate: average reactions plus comments, divided by the current follower count, multiplied by 100. We require at least three posts with both counts and a positive follower count. This is not impression-based engagement.
• Median reactions: the middle visible reaction count, reducing the influence of a single unusually popular post.
• Posts per week: the intervals between dated posts divided by the time from the earliest to the latest, scaled to seven days. At least three dated posts are required.
• Comments per 100 reactions: total comments divided by total reactions across posts with both metrics, multiplied by 100.
• Top-post multiple: the largest reaction-plus-comment total divided by the median of those totals. We require at least three measured posts and a positive median.

Missing data stays missing

A dash means unavailable, not zero. Each average uses only posts where the relevant value is available. Exact zeroes remain zeroes. We show the number of posts with both reaction and comment counts separately.

Profile presentation grading

Our versioned profile checklist grades presentation and completeness, not reach, talent or peer performance. Each check earns its full weight when demonstrated by the returned evidence. Failed checks earn zero; unavailable or masked checks are excluded. Truncated text earns only checks demonstrated in the visible excerpt.

The score is earned points divided by assessed points, multiplied by 100. Coverage shows how much of the 100-point rubric was assessed. Every incomplete score is labeled partial; different coverage is not directly comparable. A+ starts at 90, A at 80, B at 65, C at 50, D at 35, and F is below 35.

Weights: profile picture 15, cover 10, name 10, headline 15, About 20, experience 15, education 10 and location 5. Text checks measure structure using simple length and phrase rules, not semantic writing quality. A school’s prestige, a person’s name origin, age, degree level or appearance never changes their grade.

Image checks use the returned file’s decoding and aspect ratio. A square photo is within 15% of 1:1; a cover ratio is between 3:1 and 5:1. Provider thumbnails do not establish original upload quality. Lighting, face framing and visual presentation are not part of this automatic grade. Optional AI visual advice in a private content plan is separate from the checklist and requires review; it does not verify identity or authenticity. LinkedIn’s cover image guidance recommends 1584 × 396 pixels.

Headline checks: present, at least 25 characters of context, and at most 220 characters for concision. About checks: present, at least 150 characters, blank-line paragraph breaks, and a next step detected using English phrases or a link. Description checks: at least 40 characters for education or 100 for experience in readable entries. These editorial thresholds are Profilyst’s rubric, not verified LinkedIn ranking factors. Older snapshots cannot distinguish a biography excerpt from a headline and leave that section unassessed.

LinkedIn peer bands are not calibrated. We do not publish a performance percentile or Instagram-derived benchmark. Public follower counts alone are not proof of content quality.

The limits of a snapshot

The visible sample may not contain all recent posts. Provided URLs are a selected sample, not necessarily a chronological feed. Older posts have had more time to collect interactions. We cannot observe impressions, unique reach, profile views, saves, private audience demographics, or causation. We don’t measure growth without comparable historical snapshots.

Reports stay available

Opening or sharing an existing report does not refresh its data. The collection date is always visible. Removed profiles are hidden from the directory and blocked from future audits.

History and experiments

Explicit refreshes fetch fresh profile and post values after a 24-hour cooldown. Saved versions are immutable. Follower change requires snapshots at least one day apart. Post interaction change uses matching IDs only and includes the effect of posts getting older.

Content analysis and planning

Theme labels use overlapping English keyword rules, with an unclassified group. Hook type comes from the first nonempty line. CTA detection checks the final paragraph for a question or action phrase. Format is reported only when supported by source metadata. Timing is UTC. Group comparisons show their sample sizes and do not identify causal effects.

The free report includes a basic writing outline and up to three ideas. Recommendations name evidence, actions, potential impact, effort and metrics. They do not promise higher engagement. Early versus recent comparison requires at least six measured posts over two publication dates, splits equal-sized groups, and excludes an odd middle post.

Private AI planning

Public report calculations, profile grades and text-rule observations remain deterministic. The Pro workspace is coming soon and is currently available only to the owner for testing. It can use your business, audience, goal, language, timezone and publishing capacity, and choose up to ten people to learn from. Readable profile and post evidence, your context and any supplied images can then be analyzed with OpenAI to create a private plan, using GPT-5.6 Luna by default. AI planning is available only when the provider is configured and capacity remains.

AI results include source references, suggested profile improvements, content pillars, peer observations and a four-week schedule. They are suggestions for human review, not verified claims about the LinkedIn algorithm or guaranteed outcomes. A peer's sampled performance does not establish why a post worked. We keep available evidence separate from text and images you supply, and do not rewrite the public report with private supplements. Missing evidence is disclosed rather than invented.

Free access and collection depth

Existing reports are public and do not use an allowance. New reviews require sign-in and use one of two free reviews per UTC calendar month, with up to eight posts. A review that finishes with a failure is refunded to its original month. Queued and running reviews stay reserved. After a failure, members wait ten minutes before retrying the same profile; the owner can explicitly retry sooner. The configured owner has unlimited personal reviews, with shared provider budgets still enforced. Deeper collection and peer planning are currently owner-only Pro previews. They are not included in free accounts.

AI planning has a separate limit of three new plans per account per UTC day and a shared daily spending cap. Reading a saved plan is free and does not call the model again. Downloaded prompts and skills provide context for your own writing tools; they do not connect a scheduler or publish posts. MCP and publishing integrations are future work.