Competitor research: a morning brief that watches the market for you

What this is

Most owners know they should watch their competitors and do it in bursts, usually after a surprise. This playbook turns it into a routine: a small team reads the same named sources every morning and posts a brief you can read in two minutes, with a link for every claim.

Who it helps

Anyone selling against a handful of named competitors: software, e-commerce, agencies, local services with a few rivals. It also suits a research function that briefs a leadership team.

The team

• Research Lead (manager). Owns the list of sources and the standard for the brief, and answers your follow-up questions in the room.
• Monitor. Reads the sources you name: pricing pages, release notes, changelogs, job postings, press pages, and a news search restricted to the competitor's name.
• Analyst. For each change, writes one sentence on what it is and one on why it might matter to you.
• Editor. Cuts the whole thing to five bullets, each with a link, and posts it. If nothing changed, one line says so.

Sample assignment

Illustrative; put your competitors and sources in.

Objective. A weekday morning brief on Northwind, Contoso and Fabrikam, in the Research room by 8 am.

Context. We sell scheduling software to dental practices. Sources per competitor are listed in the room: pricing page, release notes, careers page, and a news search on the company name. Add a source only if I approve it.

Rules. Public sources only. No sign-ups, no trial accounts, no contact with anyone. Do not paraphrase a claim without a link to where it came from. If two sources disagree, say so rather than picking one.

Done looks like. Five bullets or fewer, each with a link and a one-line "why it matters". A one-line "no change" if nothing happened. Once a month, a page summarising the month's changes per competitor.

What happens each morning

1. The routine fires at 7 am. The Monitor loads each named page and compares it to what it saw yesterday; the record of yesterday is in the team's memory.
2. Changes go to the Analyst, who explains each one in plain terms: "Northwind added a 'Family' plan tier; likely aimed at multi-clinic groups, which overlaps with our Team plan."
3. The Editor ranks the changes, keeps the top five, attaches the links, and posts to the Research room.
4. Activity records every page read and every post made, so you can check any claim against its source.
5. You reply in the room with a question ("what does their Family plan include?") and the Analyst answers from the page it already read.

Example output

An illustrative brief:

Competitor brief, Tuesday 8 September

1. Northwind added a "Family" plan between Solo and Team. Why it matters: it targets two-to-five clinic groups, which is our core segment. [pricing page]
2. Contoso published release notes for online rebooking from SMS reminders. Why it matters: prospects have asked us for this twice this quarter. [release notes]
3. Fabrikam posted two "Customer Success, dental" roles in Manchester. Why it matters: possible UK expansion; we have no UK presence. [careers page]
4. Contoso pricing page unchanged; last change 14 August. [pricing page]
5. No news coverage of any of the three in the last 24 hours.

The competitors and changes are fictional; only the shape is real.

Requirements and availability

• Available in Korps. Routines, teams, rooms and messages, team memory, and Activity.
• A worker reads public web pages through a browser session. Korps supports a browser provider (Browserbase, a local browser beside the control plane, or the runtime's own browser); one of them needs to be configured. See research and briefing.
• A runtime (Hermes Agent comes bundled) and a model on your account. A capable but economical model is usually enough for monitoring; see choosing a model.
• The free account is open now; the app that hosts workers opens at launch.

Limitations and approval points

• Nothing is sent and nothing is bought, so the approval points are the source list (adding one needs your approval) and any contact with a third party (forbidden in the assignment).
• A worker can only read what a page shows. Sites that block automated reading, require login, or render prices only after interaction will produce gaps; the brief should say "could not read" rather than guess, and the assignment should require that.
• A brief is not analysis of your strategy. It tells you what changed. What you do about it is still yours.
• Terms of service for the sites you monitor apply to automated reading. Stay with public pages and modest frequency.