Actavero

Methodology · Visibility

How the Social media: topics and events is calculated

Method version topics-v2 · Last updated

In short

A topic links a brand to an event or subject for a period, e.g. a sponsor and a tournament, and can compare it with up to four other tracked brands. Actavero reads the brands' own posts in that period that use one of the topic's search terms, and searches Instagram, TikTok, YouTube, Facebook, LinkedIn, X, Threads and Reddit for posts by others. Claude reads every post found and marks whether it is about the topic, which brands it mentions, and whether it comes from the event's own accounts. The event's own posts are shown apart: an event named after its sponsor mentions the sponsor in every post. The result shows, per brand, how many posts mention it, the interactions and views they got, how many people could have seen them, and its share of the conversation about the topic.

Formula

Brand mentions = the brand's own posts using a search term + posts by others that Claude marks as about the topic and mentioning the brand (the event's own accounts not included)
Interactions of a post = likes (Reddit: score) + comments + shares (X: reposts + quotes), where the platform shows them
Views = Σ views of the posts where the platform shows them (TikTok, YouTube, Instagram when given, Facebook videos, X)
Potential reach = Σ followers of the distinct authors, where the platform shows them (TikTok, Instagram, LinkedIn and the brand's own accounts)
Share of topic conversation = brand mentions ÷ all posts about the topic (any brand or none, without the event's own accounts) × 100

How it is calculated, step by step

  1. You enter the brand, the brands to compare it with, the topic's name, up to five search terms or hashtags, the event's own accounts (optional), and the first and last day.
  2. The brands' own accounts (Company & competitors) are read back to the first day; posts within the days that use a search term count as own mentions.
  3. The search terms, and each brand with the first term, are searched on eight platforms (at most six searches per platform); only posts published within the days are kept. A search that fails with a server error is tried once more.
  4. Claude (Haiku) reads each post found: about the topic or not, which brands it mentions, and whether the author is the event's own account (as are the accounts you listed). Unrelated posts are left out.
  5. Collect again later to update likes and views, or change the topic (terms, brands, days); download the posts as CSV or the result as an HTML report (Scale).

What goes into it

Topic
Brand, compared brands, name, search terms (hashtags or phrases), the event's own accounts and the period.
Posts
Public posts with their author, date, text, likes, comments, shares and views.

Data sources

  • ScrapeCreators (search of Instagram Reels, TikTok, YouTube, Facebook videos, Threads and Reddit; LinkedIn and X posts found through Google's index).
  • Claude Haiku (Anthropic) for reading whether a post is about the topic, which brands it mentions and whether it comes from the event.

How often it is updated

When you create or change the topic, or collect it again.

How to read it

  • Brand mentions by others (earned) next to the brand's own posts show whether the sponsorship or presence was picked up beyond the brand's own channels.
  • The share of topic conversation shows how visible each brand was within everything posted about the event; compare brands within the same topic, collected the same way.
  • Views are the closest public measure of reach; potential reach is an upper bound, since not every follower sees every post.

Example

BNP Paribas Fortis vs Ethias × European Open, 17–25 October: 300 posts about the tournament (without the event's own accounts). 76 mention BNP Paribas Fortis (14 own, 62 by others) and 12 mention Ethias. Shares of topic conversation: 76 ÷ 300 = 25% and 12 ÷ 300 = 4%.

Limitations

  • Searches return the platforms' most relevant posts, not every post: counts are a sample and are best compared between topics or brands collected the same way.
  • Facebook's search covers videos and Reels only; LinkedIn and X posts come from Google's index, so recent and less popular posts can be missing.
  • True reach and impressions are only visible to account owners; no platform shows them publicly.
  • Posts that show the brand only in an image or video, without naming it in the text, can be missed.
  • The further back the period, the fewer posts the searches return; collect soon after the event.

Frequently asked questions

Why are some platforms marked as partial?

Facebook's public search only finds videos, and LinkedIn and X have no public search; their posts are found through Google, which doesn't index everything.

How do I pick search terms?

Use the event's official hashtag and name, and the venue or tournament nickname if people use it. Each term is searched separately, plus each brand with the first term.

Why are the event's own accounts left out?

An event named after its sponsor (like the BNP Paribas Fortis European Open) mentions the sponsor in every post it publishes. Counting those would inflate the sponsor's mentions, so they are shown as a separate figure.

  • Social media: own channels — Each brand's own social accounts compared over the last 30 days: share of engagement, followers and posts, follower growth, posts per week, engagement rate, top posts, formats and posting days.
  • Share of conversation (online mentions) — Your share of the online mentions of all tracked brands in the last 30 days, on websites, blogs, forums and news sites in your market, with sentiment.
  • Earned media value — What the visibility of social posts about a topic would have cost as paid social: views × CPM, or interactions × CPE for posts without views, at rates that each carry their source.

Feedback on this explanation

Loading…