What the system actually does.

1. The reactors become rows A table bound to the person pulls the profiles of everyone who reacted to your posts. A schedule keeps fetching new reactions on its own once a new post goes out.
2. Fill in what is missing Each profile gets company, role, company size and a verified email address. What one provider cannot deliver, the next one in the waterfall tries.
3. Check against your ICP A qualification step scores every person against the stored ideal customer profile and sets a gate. Anyone who does not fit does not move on, and stops costing money.
4. The post becomes the opener The outreach refers to the exact post the person reacted to, and to what they said. That is the difference between cold outreach and joining a conversation that already happened.
5. Enrol and let them reply Qualified people enter a sequence across LinkedIn and email. Replies are detected, classified and answered, and a request for a meeting gets booked.

Connected beforehand: A connected LinkedIn profile · Posts that actually get reactions

1. Find the right posts An agent watches your topic area and flags posts getting unusual traction. Not every post is a find. What you are looking for is reach on your topic, not reach as such.
2. Author and reactors arrive separately For a given post the source returns both: the person who wrote it, and the people who reacted. The author carries the value POST_AUTHOR in the interaction type column.
3. The gate on POST_AUTHOR This is exactly where the two plays part ways. This one filters on POST_AUTHOR and leaves the reactors alone. The author took a public position, the likers only nodded.
4. Enrich and check against the ICP The person gets company, role, size and a verified email address. Then they are scored against your ideal customer profile, because reach alone does not make a buyer.
5. Their own post is the opener The outreach refers to what the person wrote, to their argument rather than their job title. Someone who takes a public position expects disagreement or a follow-on, not a pitch.
6. Enrol across two channels LinkedIn and email run in one sequence. Replies are detected and classified, and whoever replies on one channel stops being contacted on the other.

Connected beforehand: A connected LinkedIn profile · Named topics or search terms to watch

1. The signal gets named What counts as a buying signal for you is written into the workspace knowledge. That is not a setting in the code, it is an entry you can change when your market changes.
2. A workflow listens for it A workflow hangs off that exact event. It does not run on a schedule, it runs when the signal fires, and only then.
3. Check first, act second Before any action the gates run: unsubscribe, blacklist, contactability, caps. A signal is a reason, not a free pass.
4. Enrich and classify The company in question and the right contacts are fetched and scored against the ICP. A signal at a company that does not fit leads nowhere.
5. Outreach that names the reason The copy names the concrete trigger. Not 'we help companies like yours', but the reason this message arrives right now.

Connected beforehand: At least one signal source being watched · A stored ideal customer profile

1. The ICP comes from real data What your best customers have in common is derived from them: industry, size, region, technology, the role of the decision maker. The result lands in the workspace knowledge, where it can be read and changed.
2. Search for lookalikes A lookalike source finds companies matching that pattern and attaches them as rows. The source keeps running on a schedule if you want it to.
3. Verify against the ICP Resemblance is not proof. Every company found is scored against the derived profile, and only what passes moves on.
4. Find the right people Each company gets the contacts in the role that decided at your existing customers, not just any contact but the same function.
5. Two channels, one plan Email and LinkedIn run inside one sequence, not in two separate tools. Whoever replies on one channel stops being contacted on the other.
6. Scale the winner Which variant, which industry and which opener actually produce meetings is in the workspace numbers afterwards, and becomes the basis for the next round.

Connected beforehand: A list of your best existing customers

1. The list exists anyway You do not need a second process. The rows researched, enriched and scored against the ICP for outbound already are the better audience.
2. Qualify before the ad spend Gate first, export second. Anyone outside the profile costs neither a message nor a click.
3. Push it to the ad account The checked rows go into the ad account as an audience. Not interest targeting, but named companies and people.
4. Both channels on one message The ad uses the same opener as the sequence. Whoever gets the message sees the ad, and the other way round.
5. Measure it back Someone replying after seeing an ad is the same lead in the workspace. The ad's contribution becomes visible instead of disappearing into the ad account.

Connected beforehand: An ad account (LinkedIn, Meta or another)

1. The opener is short WhatsApp does not tolerate email length. The first contact is a question, not an introduction, and it passes the same gates as every other channel.
2. The reply gets classified Replies are read and sorted: interest, question, wrong person, refusal. A refusal ends the sequence immediately and puts the lead on the exclusion list.
3. Qualify in the conversation Instead of a form, an agent asks the two or three questions that decide fit for you, inside the running chat and in your tone of voice.
4. Book the meeting there and then If it fits, the booking link goes out with concrete slots. The booked meeting lands in the calendar and in the CRM with no step in between.
5. A human takes over when it matters Anything ambiguous goes into the queue rather than out the door. You approve, and the system does nothing consequential unasked.

Connected beforehand: A connected WhatsApp account · A connected booking tool (cal.com or Calendly)

1. The CRM is the source, not the target Lost and stalled records are pulled in as rows in a table. No purchase, no enrichment cost for contacts you already hold.
2. Check what has changed Is the contact still there? Has the company grown? Has the original objection gone away? A contact where nothing changed is no reason for a second message.
3. The objection becomes the opener Too expensive, no budget, bad timing, went with a competitor. Each of those needs a different approach. The copy is built from the actual reason, not from a template.
4. Approach gently Existing contacts tolerate less frequency than cold ones. The sequence is shorter, the intervals are longer, and the exclusion list applies unchanged.
5. Write the results back Replies, meetings and the new status flow back into the CRM, onto the existing record, without creating duplicates.

Connected beforehand: A connected CRM (HubSpot, Salesforce or another)

1. The workflow hangs off the event Not a nightly run that goes through everything again, but a response to that one new record.
2. Fill in what is missing Company data, role, company size, verified email. What one provider cannot deliver the next one tries, and every attempt is logged.
3. Classify, do not just populate An agent scores whether the record fits the ICP and writes down why. A field with a number and no reasoning helps nobody.
4. Write back without duplicates The results land on the existing record. No second contact with the same name, no parallel truth.
5. Hand over when it is urgent If the lead fits and is hot, a notification goes to the person who should take it, with the context rather than just the name.

Connected beforehand: A connected CRM with write access

1. Reactions become data Whoever likes and comments lands in a table with company, role and industry, instead of in a notification nobody reviews.
2. Describe the audience Which roles, which industries, which company sizes. After a few posts there is a profile that was not guessed.
3. Hold topics against resonance Which post reached whom. The topics that move your ICP often differ sharply from the ones that collect the most likes.
4. Write it back into knowledge What holds up becomes a messaging angle or a persona in the workspace, and therefore the basis for the next sequence, not just the next post.

Connected beforehand: A connected LinkedIn profile

1. Import rather than source The list becomes rows in a table. Duplicates surface there instead of being contacted twice later.
2. Close the gaps Missing company data, roles and email addresses are filled in, through a waterfall across several providers, so one blank is not the end.
3. Verify the addresses Every email is checked before it is used. A derived address is a guess, and guesses ruin deliverability.
4. Dry run first, paid run second A trial on ten rows shows what comes out and what it costs. Only then does the whole list run, under a cap you set.
5. Prioritise instead of contacting everyone The enriched rows are scored against the ICP. What does not fit is not contacted, which saves budget and protects the domain.

Connected beforehand: A list as CSV or export

1. Is it even sending? The first question, and the one answered wrongly most often. A green run does not mean anything went out. A dry run simulates, and exactly that has left campaigns quietly idle before.
2. Where are rows stuck Which enrolments stalled, which steps are overdue, which cells errored. With the reason, not just the count.
3. Are the agents doing their job What an agent decided is in the run report, along with the tools it called. A judgement without a reason is a finding, not a result.
4. Is the automation working Not whether the workflow fired, but whether anything was different afterwards. Replies, meetings, pipeline, held against the effort.
5. Repair before you extend A stalled sequence costs more than a missing feature. What is not running takes priority over anything you could additionally build.