BrandLoop

Glossary

Definitions of the technical and legal terms used across this site. Each definition matches how the term is used in the sourced writing here.

GEO (generative engine optimisation)
The body of work aimed at getting a page or person cited as a source in answers produced by AI engines.
Not a continuation of classic search optimisation: the measured factors differ and some run the other way. Keyword stuffing was the only method measured to reduce visibility.
Detail →
Dwell time
The time a member spends on a post in the feed; LinkedIn has stated since 2020 that it uses this as a ranking signal.
Two measures exist: on-feed dwell, starting when at least half the post is visible, and after-click dwell. The threshold in seconds has never been published.
Detail →
Long dwell
One of two objectives LinkedIn’s ranking model predicts: time on a post above a threshold that depends on post type.
The other is contribution — a like, comment or reshare. Both are named in LinkedIn’s February 2026 paper.
Detail →
Second pass ranker
The LinkedIn model that merges candidates ordered by each content source into a single personalised feed list.
Because of this structure there is no single “algorithm”. A post must first enter a candidate pool and then be ordered within it; it can be dropped at either stage for different reasons.
Detail →
Earned media
Content published about a person or organisation by an independent outlet without payment.
Roughly 84% of citations in AI answers come from this kind of content; paid content accounts for 0.3%. It is the heaviest measured factor in personal-brand visibility.
Detail →
OAI-SearchBot
OpenAI’s search crawler. Whether a page can appear as a source in ChatGPT answers depends on this bot being able to crawl it.
Frequently confused with GPTBot. GPTBot is for model training; blocking it does not directly end search visibility. OpenAI explicitly recommends allowing OAI-SearchBot.
Detail →
GPTBot
The crawler OpenAI uses to train its foundation models.
Blocking it means the content stays out of training data. It is not the bot that determines whether you appear in search answers.
Detail →
llms.txt
A plain-text file proposed for the site root to summarise a site’s content for AI models.
No measured evidence in its favour. Google’s John Mueller stated no AI system uses it, and no vendor mentions it in crawler documentation. It has not been shown harmful either.
Detail →
robots.txt
The root-level text file stating which crawlers may access which parts of a site.
A permission mechanism, not a guarantee: OpenAI states robots.txt rules may not apply to user-initiated fetches, and a CDN can block a bot independently of robots.txt.
Detail →
Structured data (schema.org)
Markup that describes what a page is about in a form machines can read.
In a controlled study of 1,885 pages already receiving citations, the effect on AI citations was indistinguishable from zero. Discovery was not tested. Treat it as hygiene, not leverage.
Detail →
Entity page
A crawlable page describing a person or organisation with a one-sentence definition, verifiable records and links to their other profiles.
The precondition for AI visibility. A login wall, JavaScript-only rendering and a noindex tag each close this threshold.
Detail →
Title field
The structured “Title” field in a LinkedIn experience entry; the hard filter in recruiter search reads it directly.
Not the same as the headline. The headline appears only in keyword search and is truncated at around sixty characters on the results card.
Detail →
Standardised skill
A skill entry chosen from LinkedIn’s own list and therefore recognised by search.
A skill typed as free text is not recognised. Pinning the three to five most directly tied to your target title decides what surfaces on the card.
Detail →
Impression
The number of times a post was displayed on someone’s screen.
Across 476,000 personal-profile posts the median was about 840. The distribution is skewed: the 94th percentile begins at 15,000 impressions.
Detail →
Engagement rate
Reactions, comments and shares on a post as a proportion of its impressions.
It can move opposite to reach: in the same measurement it falls from 2.86% in the bottom 80% of posts to 0.89% in the top 5%. Bigger reach can mean thinner attention.
Detail →
Browser automation
A tool acting on a platform through the user’s browser session instead of the official API.
Section 8.2 of the LinkedIn User Agreement prohibits it directly, and the party bound is the user, not the vendor. Consequences are usually staged: temporary limits, feature restrictions, suspension.
Detail →
r_member_social_feed
The LinkedIn permission that would let an application read comments on a member’s behalf.
Granted only to selected developers. Writing comments sits in the open category while reading them does not, which is why “automatic comment replies” cannot be delivered through the API.
Detail →
Public disclosure (alenileştirme)
Under Turkey’s Law 6698, personal data being made publicly available by the data subject’s own will.
The regulator’s settled reading is that such data may only be processed in a way connected to and limited by the purpose of disclosure. Being public does not mean processable for any purpose.
Detail →
Data controller
Under Law 6698, the natural or legal person determining the purposes and means of processing personal data.
The burden of proof sits here. In published decisions, a controller that cannot document the processing condition it relied on loses.
Detail →
VERBİS
Turkey’s public registry of data controllers, which controllers above a threshold must join.
Exemption applies to controllers with fewer than 50 employees and an annual balance sheet under 100 million TRY whose main activity is not processing special-category data.
Detail →
Canonical tag
The tag stating which address counts as the original when the same content is reachable at several URLs.
It matters on bilingual sites: Googlebot carries no cookies, so if the language decision is not made from the path, two languages collide on one address and indexing becomes ambiguous.
Sitemap
An XML file listing a site’s crawlable addresses in machine-readable form.
Submitted separately to each search console. Bing is no longer the main lever for ChatGPT but it is the index Microsoft Copilot draws on.
Detail →
Query-answer match
A heading on a page matching, word for word, the question people type into a search box.
It scores as the highest-impact factor in GEO measurements. Writing headings as questions is different from — and works opposite to — keyword repetition.
Detail →