Share of voice dashboard for your brand and competitors

I run communications at a company and I want a share of voice command center that my comms team opens every morning. Build it on GDELT for the news data, Notion for the filed reports, and Slack for the weekly heads up. GDELT is a free public API with no authentication, so the only credentials anyone connects are Notion and Slack.

On the main screen I set up a tracked set: our brand, up to four competitors, and any campaign or category terms I care about. Each entry has a friendly display name and the search terms behind it, so I can fold a few spellings or product names into one bucket. Tracked sets persist per user so every brand lead keeps their own board, and I can save several sets and switch between them, for example one per product line.

For every name in the set, render these side by side so they can be compared at a glance: a coverage volume timeline and a share of voice split for the last 90 days using Article Volume Timeline; a sentiment trend and a sentiment histogram using Article Tone Timeline and Tone Distribution; a where in the world breakdown by language and source country using Coverage Breakdown by Language or Source Country; a broadcast pulse using TV Coverage Timeline; and a clickable list of the articles sitting behind all of it using Search News Articles.

Show share of voice as a percentage of the tracked set, not as a share of all world news. Take each name's article volume, divide by the combined volume of everyone in the set, and make it add to 100 percent. Render it both as a split for the current window and as a trend line over the window, because the trend is what tells me a competitor is gaining on us.

Clicking a spike on any chart filters the article list to that name and that date window, so someone can answer what happened on Tuesday in two clicks. Each article row shows the headline, the outlet, the source country, the language, and the date, and links out to the original story.

Keep the framing non technical throughout. Label tone as sentiment everywhere in the interface, express share of voice as a percentage, and default to English language sources with a toggle to include all 65 plus languages. Default the window to the last 90 days, which is as far back as the news data reaches, and let me narrow it to 30 or 7 days.

Add a Write this week's coverage report button that kicks off a background agent. It pulls the biggest volume and sentiment moves across the tracked set over the last 7 days compared against the 7 days before, reads the top articles behind each move with Search News Articles, then writes a narrative report page into Notion with Create a Page covering share of voice, sentiment shifts, competitor wins, and the stories worth a response. Once the page exists it posts the headline numbers plus the Notion link into a Slack channel with Send a Message, and saves the same report back into the app.

Keep a reports page inside the app listing every report that has been generated, newest first, with the full text readable in the app and a link out to the Notion page. That way the board builds a history of past weeks and I can see how the story changed over time.

Some constraints to build around. The news search covers a rolling three months, so 90 days is the natural maximum window. The article list call returns at most 250 articles per request, so narrow the date range and stitch consecutive windows when more are needed. The TV data excludes the last 24 hours of broadcasts unless that is explicitly requested, and it needs either a station filter or the combined data option. Keep requests serial at roughly one per second and back off on rate limits. Cache results, because the underlying data only refreshes every 15 minutes and caching keeps the board fast when several people open it at once.