Competitor research

Scope the standard research flow to competitor keywords or profile URLs, group the datasets in a project, then analyse share of voice, audience, and trends with skills.

When you sell a B2B SaaS project-management tool, the questions that move the roadmap are about your rivals: what users actually say about each one across the open web, which feature requests and pain points keep recurring, who their audiences are, and whose share of voice is growing. Competitor research in buzzabout is not a separate screen or wizard — it is the standard research flow scoped to competitor keywords (user-generated mentions) or competitor profile URLs (their own posts). You collect one dataset per competitor, group them in a project, and then point the assistant and skills at the result to compare share of voice, benchmark engagement, and profile the people talking.

Prerequisites

A buzzabout account with available credits (1 mention ≈ 1 credit). Audience analysis and CSV export require a Pro plan or higher.
Optional: competitor profile or post URLs on a supported platform (reddit, tiktok, youtube, x, instagram, or linkedin) for URL-mode research.
New to the research flow? Start with Brand research — competitor research reuses the same preview and run mechanics.

Boolean keywords for competitor sets

Competitor research lives or dies on the keyword query, so it is worth knowing the operator model before you start. The keyword chip editor supports AND, OR, NOT plus parentheses for grouping and "double quotes" for exact phrases. The default join between two keywords is OR (it broadens reach) — add AND to require co-occurrence, NOT to exclude. Operators are case-insensitive and shown uppercase.

asana monday "clickup" — any post mentioning any of the three rivals (implicit OR), good for a head-to-head competitor set.
asana AND (pricing OR "too expensive") — Asana posts that also raise cost, to isolate one pain theme.
"project management" NOT jira — the category conversation with one incumbent removed.

Sanity-check complex queries against the live mention-count estimate the editor shows — long boolean chains can broaden or narrow reach more than you expect.

Walkthrough

Competitors

From the home screen (Discover what people buzz about), open the Competitors tab in the suggested questions and pick a starting prompt — for example How are people comparing CrewAI vs LangChain for building agents? or Reverse-engineer what's working for a competitor — paste their profile URL. These are examples to adapt; for a project-management tool you might instead ask how people compare your two closest rivals.

Keyword query

Type a competitor comparison query into the chat input and send it — for example a phrase comparing two rival project-management tools. This creates a chat, and the assistant returns a research preview of type Keyword query. A keyword query captures user-generated mentions across the web that match your keywords. Edit the keyword chips to refine which terms are matched, using the boolean operators above.

URL query

Alternatively, paste a competitor's profile or post URL into the chat input. The assistant returns a URL query preview, which scopes to that competitor's own profile and posts and reports a profile count (profiles found) rather than a mention estimate. Add or remove competitor URLs with the Add URL… field — the source is auto-detected from the host. URL queries have no Research depth picker and no credit badge; the scope is the profiles you add.

Research depth

For keyword previews, refine the filter row (sources, time and date range, language, country) — each edit regenerates the preview and updates the estimate. Then pick a Research depth. The picker controls How many mentions to collect when you run research. and sets the credit cost: Overview (100 mentions), Landscape screening (200), Qualitative research (500), or Quantitative analysis (1,000). Depth applies to keyword runs only; URL previews are scoped by the profiles you add.

Preview mentions

Click Preview mentions to inspect a sample of matching mentions in the canvas. This is free — it does not run collection or spend credits — so use it to sanity-check that your competitor keywords are pulling relevant conversations before you commit.

Start research

When the preview looks right, click Start research. This spends credits and runs the dataset collection. The preview hides and a dataset card takes over the turn, showing Collected N mentions where N is however many your depth and matches produced. The run is asynchronous — the preview polls while pending, so give it a moment to finish.

Create project

To keep competitors together, run a separate dataset for each rival and group them in a project. Click Create project (Projects help you organize chats, datasets, and listening agents around a topic or brand.), give it a Project name such as Q2 Competitor Analysis, and click Create. Run each competitor's research chat from inside the project.

Datasets

Every dataset you collect from a project chat lands in the project's Datasets panel (Linked datasets from project chats). This is where your competitors sit together — one dataset per rival keyword set or profile URL. The project groups the datasets; it does not add a built-in side-by-side compare view, so you read across them in the panel and in chat.

What you can do after collecting

A collected competitor dataset is raw material. The payoff is analysing it — and most analysis runs in chat, either by asking the assistant directly or by applying a skill. Open the skills picker from the chat input (the book icon, Open skills), choose Browse all skills, open a skill, and click Use this skill to inject its instructions into your next message. Most skills return the same shape: a hero chart, a detail table, and three takeaways. Several depend on pre-computed enrichment (sentiment, emotion, brand, topic columns, or an audience dataset) and will say so in one line if the data is not present.

Share of voice is not a button, page, or standalone feature — it is something you ask for. Apply the Mentioned brands skill, which reads the pre-computed brand-mention field per post and renders a per-brand share-of-voice bar coloured by sentiment, plus a brand detail table. Or ask the assistant in plain language, for example compare share of voice across Asana, Monday and ClickUp in this dataset or show share of voice by platform. The assistant renders the comparison as horizontal mention-volume bars — each brand's or platform's portion of total relevant mentions. For a single competitor dataset this tells you who dominates the conversation; across the project's datasets you read the shares side by side.

Analyse who engages with each competitor

To understand the people behind a competitor's mentions, build an audience dataset from a collected dataset. In the chat where the dataset lives, ask the assistant to profile that dataset's audience. An Audience research in progress card appears; when it finishes, click the View … profiles button (it shows the live profile count) to open the audience canvas — a sortable, filterable profiles table. Open any profile for its location and language, content niche, interests, brand affinities, authenticity, creator tier, a marketing-focused summary, and a Personality (OCEAN) section with bars for Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism.

OCEAN is inferred by an LLM from each person's public posts — a directional AI estimate, not a psychometric test — and is hidden for low-signal profiles. Used as creative guidance, a high-Openness competitor audience tends to respond to novel, experimental messaging, while a high-Conscientiousness audience responds to proof and reliability framing. The Audience demographic chart and Audience psychographic chart skills summarise the same data across the whole audience dataset. Audience analysis is a Pro-plan-and-up feature and costs 3 credits per profile. See Run audience analysis for the full flow.

Benchmark mention volume and engagement

Ask the assistant to benchmark each competitor's collected mention volume and engagement, and to break the numbers down by social network — for example which network drives the most engagement for these mentions?. Skills like Topics vs. Reach – Bar Chart, Emotions vs. %ER – Bar Chart, and Intentions vs. %ER – Bar Chart turn the dataset into per-segment bars you can read across competitors. Because each rival lives in its own dataset, you run the same skill on each and compare the outputs.

Surface trends, feature requests, and pain points

Point pattern-pipeline skills at a competitor dataset to extract what to build and what to avoid: What's trending splits the dataset into recent versus baseline windows and shortlists rising themes; Feature requests ranks product-directed asks and triages each one; Analyze pain points clusters complaints weighted by reach and intensity. Positive / negative quotes pulls the strongest testimonial-grade and warning-grade quotes from a rival's conversation. These pattern runs are asynchronous and may charge credits.

To keep any of this live, attach a listening agent so the competitor's queries re-run on a schedule and the dataset keeps growing.

Gotchas

Owned (URL) vs user-generated (keyword) are different scopes. A URL query scopes to a competitor's own profile and posts and returns a profile count; a Keyword query captures public mentions across the web and returns a mention estimate. Choose deliberately based on whether you want their channel or the conversation around them.
You cannot fully exclude a competitor's own posts from a keyword run. Keyword runs collect public mentions matching the keywords, which can include the brand's own posts. There is no owned/user-generated exclusion toggle.
Preview type is locked. A keyword preview always regenerates as a keyword preview, and a URL preview always as a URL preview. You cannot switch a preview between keyword and URL mode.
Preview mentions is free; Start research spends credits. 1 mention ≈ 1 credit, at the depth you picked (100, 200, 500, or 1,000).
Volume matters. Meaningful share-of-voice and audience comparisons need a healthy sample — aim for ~100+ mentions per competitor. A thin dataset for a small rival will skew any comparison.
Research depth is plan-clamped. On the Starter plan you pick a single source and only Overview (100) and Landscape screening (200) are selectable; multi-source research, deeper depths, audience analysis, and CSV export unlock at Pro and above. Options above your plan's maximum are disabled and auto-fall-back to the highest allowed depth.
The URL-type source picker is locked. For URL previews, sources are derived from the pasted URLs and are not independently editable.
URL hosts are restricted. Only reddit, tiktok, youtube, x/twitter, instagram, and linkedin are accepted. An unknown host shows Unsupported URL — must be reddit, tiktok, youtube, x, instagram, or linkedin and does not regenerate. A duplicate shows URL already in list.
Runs are asynchronous. The preview polls while pending; a stuck run is flipped to failed by a backend watchdog after about 60 seconds, after which the dataset card exposes a Show preview toggle.
A project groups, it does not compare. Projects keep chats, datasets, and listening agents together — there is no built-in side-by-side compare view, so comparison across the Datasets panel and in chat is manual.