Find buyers engaging with competitors

Watching a competitor's engagement is a job somebody does by hand for about a week. You open their page, scroll the reactions on the last few posts, copy the interesting names into a sheet, look up a handful of them, and then stop, because doing it properly means checking a few hundred profiles every week and the returns are invisible until you have done it for a month.

Both halves of the work already exist as data. Who engaged with a public post is public, and what a good account looks like is written down in your CRM as the shape of everything you closed. Joining and filtering the two is the rest of the job, and it is a job an AI agent can run every week.

How AI finds leads in competitor engagement

Each weekly run pulls the reactions and comments on your competitors' posts from the last 14 days through Apify, on LinkedIn or on any network where that engagement is publicly readable. Two weeks is the window worth using: older engagement has gone cold, and a shorter window returns too little to rank.

The whole set is collected in one run, deduplicated, and kept with its context: which post, what kind of interaction, and the comment text where there is one. That context is what the opener is written from later.

Engagement on its own is a weak signal. A like is not intent, and treating it as intent is how this becomes spam. Everything after collection is filtering, hard enough that what survives is worth an email.

Filtering competitor engagement down to likely buyers

Three groups are removed first, in this order:

• The competitor's own staff. Employees engage with their own company's posts constantly.
• Your existing pipeline. Anyone already in HubSpot as an open opportunity or a customer, checked by email where there is one and by company domain otherwise.
• The wrong function. A commenting engineer at a target account is not the buyer, though they may be worth keeping as a route in.

Everyone left is scored against your ICP on the axes you actually sell against: company size, sector, geography, and how close the title sits to the buying role. The company and the person are scored separately and then combined, so a perfect-fit company with the wrong contact shows up as exactly that instead of being averaged into the middle. Anyone under the threshold is dropped and counted, not queued.

What makes competitor engagement prospecting hold up

Enrichment runs after scoring, never before. A competitor post with decent reach returns a few hundred people, and enriching all of them buys you the discovery that most were students, employees and people in the wrong country. Scoring costs nothing, so the free filter goes first. On a typical week you pay for thirty lookups instead of four hundred. People above the threshold with no verified email stay on the list, marked unreachable, since a named person at a target account is still useful to whoever owns it.

Deduplicate by profile URL rather than by name. The same person reacting to three posts in a fortnight is one prospect, and their display name will not group them. Middle initials appear and disappear, job titles get appended, accents get stripped by whatever scraped them. The profile URL is the only identifier the platform itself treats as unique.

"I saw you commented on their post about warehouse routing" is a sentence with evidence behind it. Keep the engagement context attached to each person during collection and the opener writes itself from it. Strip it out and write the opener later, and you get "I see you're interested in logistics", which reads as generated because it is.

A draft campaign a person approves

Each person above the threshold gets one opener, grounded in the post they engaged with and what they said about it, loaded into Lemlist as a draft campaign. A summary goes to your team's channel in Slack: how many people were found, how many survived scoring, how many were enriched, and a sample of the openers. Nothing sends until a person approves it.

Each run reports the engagements collected, how many were dropped at each filter and why, how many were enriched, how many are unreachable, and the campaign's draft status.