How to find the competitors you haven't named yet

Competitor Radar is a PrismCI capability that finds companies that compete with you, checks each one against evidence, scores it, and hands the ones you choose to track to monitoring, the knowledge base, a battlecard draft and enablement. It exists because of a gap most competitive-intelligence programs share: they watch the competitors someone already wrote down.

The problem: you can only monitor names you already know

A tracking list is a snapshot of what your team knew when it was written. Markets move faster than lists. A new entrant raises a round, an adjacent vendor adds your category to its roadmap, a small rival starts appearing in your late-stage deals, and for weeks nobody notices because nobody is watching a name that isn't on the list.

The usual fixes are manual. An analyst searches review sites once a quarter. A rep mentions a new name in Slack. Someone asks an AI assistant "who else competes with us?" and pastes in the answer. Each of these finds something, and each has the same weakness: nothing checks whether the name is real, relevant, or actually in your deals.

What Competitor Radar does

You describe what you sell, prefilled from your company profile. The radar then looks for companies that sell something like it, in this order:

• Your own data first: competitor fields in your CRM, call transcripts, Slack debriefs and field reports, news, AI-visibility results, and CheckThat. This is where real competition shows up, and it finds the rivals already in your deals that nobody formally tracks.
• Then the open web and review sites, merging results from several search providers and removing duplicates.

How it decides what is real

Finding a name is the easy part. The radar's value is in what it does next. Before anyone sees a candidate, it has to pass three checks:

• A product claim on its own website: the company must say, in its own words, that it sells a product.
• A domain that resolves: the website has to be a live, reachable site.
• Real overlap with your products: the claim has to match what you sell, not just share a keyword.

Companies that pass are scored on three separate things: relevance, discovery confidence, and a 0–100 threat score built from overlap, traction and verified momentum. The evidence behind each is shown next to the number. AI reads pages to extract what a company says it does; it never sets a score. The scoring is rules, so the same evidence always gives the same result and you can explain any number to a colleague.

Nothing is hidden: the Below the bar tab

A company that fails a check isn't thrown away. The Below the bar tab lists every company the radar found but did not put in front of you, with the three checks, where it came from and what the radar saw. "Couldn't check" (a search provider was over its limit, or a site blocked the request) is kept apart from "checked, and below the bar," so a failed lookup never quietly becomes a rejection. If you disagree, you can Add anyway, mark it Not a competitor, Check again, or add its website.

From candidate to action

Each candidate gets a suggested next step, and your team decides:

• Track: add the competitor and switch on monitoring, the knowledge base, a battlecard draft and enablement in one step.
• Watch: check news about it each week, and bring it back when the evidence grows.
• Investigate: park it for a closer look.
• Dismiss: remembered, so it doesn't keep coming back.

The radar also learns from you. Calibration compares its suggestions with your decisions over 90 days, so admins can tune thresholds to how your team actually judges a competitor.

A record you can trust

Every run is logged: why it ran, what it found, what changed since the run before (a threat rising or easing, a company back from Watch, one that now passes or is now below the bar), what your team did with the results, the searches and cost, and the policy in force. Run history exports to CSV. After a run, the people who manage the radar get a bell notice, an email and optionally a Slack message. By default that happens only when something is new, changed or added, and always when a run fails or only partly finishes. Messages carry company names and counts only, and customer and people's names never leave your server.

How it differs from the alternatives

Monitoring tells you what the competitors you named are doing, and PrismCI does that too. Prompt-based discovery asks a language model to suggest who else might compete, which can be a useful starting point, but it takes the model's word for it. For example, Crayon publishes an AI toolkit prompt for identifying new competitors from recent news, press and social activity; check Crayon's own documentation for how it works today. In Klue's public materials we found competitor monitoring and profiling but no automatic discovery of new competitors; verify with the vendor.

The Radar's approach is different in three ways. It begins with your own deals, not the open web. It verifies each candidate against the company's own website before showing it. And it scores by rules and shows everyone it turned away. If you are evaluating tools, ask any vendor how a suggested competitor is verified and where the rejected ones go.

Where and how teams use it

• CI lead: find the rival that keeps appearing in deals but was never on the list, and set it up in a click. The new competitor flows into ThreatScope and the Intelligence Feed.
• New CI team: start from an empty Competitors page in the Setup Wizard and get a vetted shortlist instead of a blank screen.
• Product marketing: spot emerging challengers early, with the evidence to decide who earns a battlecard.
• Sales leadership: see which unfamiliar names reps are meeting before they become a quarterly surprise.
• Strategy and executives: a weekly read on who entered the market, who moved, and who dropped off.

An illustrative example

The company names and numbers here are invented to show how the radar is used; they are not a customer story.

A team tracks eight competitors. On Monday the weekly run finds Helios Analytics, which no one had listed. It appears on six calls and in four CRM deals, its own site claims a competing product, its domain is live, and it overlaps two of the team's products. It scores 78 out of 100 for threat, with the evidence listed. A second company, Quillfox, turns up on review sites with a score of 54, and the radar suggests Watch. Nine others land in Below the bar, each with a reason.

The CI lead chooses Track for Helios. Monitoring starts, a battlecard draft appears, and the next ThreatScope refresh includes it. The whole thing took one run and two clicks, and the team can explain exactly why Helios is on the list and the nine others are not.

Run it on your terms

The radar's weekly run is a choice, not a default surprise: on for new installs, and a switch under Admin on existing ones, because each run uses AI and web-search calls. Page fetches can go through your outbound proxy, and the thresholds, the recommendation table and what Track turns on are all set by your admins. See Competitor Radar for the full feature list, or request early access to see it on your own market.