How to Set Up Signal-Based Outbound in 7 Steps

Here is how to set up signal-based outbound: list the three events that preceded your last ten deals, write the filter a company must pass before you care, set up the LinkedIn searches and notifications that catch each event, log every hit in a sheet with a date, and send one short message that names the signal within a day. One person can run this with a free LinkedIn account in about 45 minutes a day. The exact search strings, the sheet layout, the messages and the daily routine follow.

What you need to set up signal-based outbound

• A LinkedIn account. Free is enough for this guide. Sales Navigator adds saved people searches, job-change alerts and more filters, which start to pay off once you track more than about 40 companies.
• Your last ten closed deals and, if you have them, your last ten lost ones. Step 1 is built from them.
• A spreadsheet. The layout is in Step 4; it takes five minutes to set up.
• A one-line description of who you sell to. If you do not have one, the lesson on what a buying signal is explains the difference between fit and timing, which is the whole idea here.
• About 45 minutes a day for the first two weeks, then 20 to 30 once the searches are tuned.

Step 1: Pick three signals that preceded your last ten deals

Open your CRM or inbox and, for each of your last ten customers, write down what was true about them in the month before they first replied. Not what you did, what happened to them. Common answers: they hired someone into the role you sell to, they posted about the problem you solve, they commented on a competitor's post, they changed tools, they raised money, they followed a competitor's page.

Count how often each event appears. The three most frequent are your starting signals. Three is the right number because each signal needs its own search, its own message and its own review, and more than three means none of them gets tuned.

If you have fewer than ten deals, use the signal that produced your first one. When we launched Intentsly we ran this on ourselves: we targeted founders who had posted on LinkedIn about cold outreach burnout, let our conversational AI handle the replies and the booking, and got 14 qualified demos in 14 days with no manual outreach. The reply rate was 4x our previous cold campaigns, and most of the leads came from signals caught in real time. One signal, chosen because it described our own pain, was enough to start.

The free Buying Signal Playbook lists 12 signals worth chasing and 8 to ignore, with the LinkedIn trigger for each, if you want a menu to choose from.

Step 2: Write the filter a signal must pass

A signal from the wrong company is noise, so write the fit filter before you catch anything. Four lines are enough: company size range, industry or business type, geography, and the two or three titles you sell to. Do not spend more than 20 minutes on this today; the other guides assume the same four lines and you can sharpen them after the first weekly review.

The filter is applied after the signal, not before. You catch everyone who posts about the problem, then open each profile and keep only the ones that pass. Reversing the order (build a list of ideal companies, then wait for them to do something) is slower to start and misses the companies you did not know existed.

Write an exclusion line too: competitors, agencies selling the same service, students, recruiters, and anyone you have messaged in the last 90 days.

Step 3: Set up the LinkedIn searches and alerts that catch each signal

Each signal type has one place on LinkedIn where you can see it with a free account. Set these up once and bookmark the result pages.

Problem posts. Type a phrase in the LinkedIn search bar, press Enter, click the "Posts" filter, then "Date posted" and choose "Past week" (or "Past 24 hours" once you check daily). LinkedIn's search accepts quotation marks for exact phrases and the operators AND, OR and NOT in capitals. Three strings for a payroll product, as an example:

"payroll" AND ("nightmare" OR "manual" OR "spreadsheet") AND ("finance" OR "HR")
"switching payroll" OR "payroll provider" OR "payroll software recommendations"
("anyone recommend" OR "any recommendations") AND payroll

Bookmark each results URL. The date filter is kept in the URL, so reopening the bookmark re-runs the search.

Competitor and industry-leader engagement. Follow two or three competitor company pages and the three people whose posts your buyers read. On each of their recent posts, click the reactions count under the post to open the list of people who reacted, and scroll the comments. Open the "Follow" menu on a person's profile and turn on notifications (the bell icon) so new posts reach you the same day. This is the manual version of post-engagement tracking, and the slowest part of the routine.

New hires and open roles. On a free account, open each target company's page, go to the "People" tab, search the titles you sell to and check the start date on the current position; a start date inside the last 90 days is the signal. The company's "Jobs" tab shows the roles your product supports. On Sales Navigator, save a lead search and turn on the "Changed jobs in the past 90 days" filter; the alert does the checking for you.

This step is the one Intentsly removes: it watches LinkedIn post engagement, keyword posts, competitor page followers and hashtag engagement for you, scores every person against your ideal customer profile and lets you export the list, so the 40 minutes of checking becomes a prioritised queue (LinkedIn signals). The rest of this guide works without it.

Step 4: Build the signal log

Every hit goes into one sheet, the same day you see it. The log is what turns a habit into a system: it stops double-messaging, it tells you which signal produces replies, and it gives the weekly review something to read. Nine columns:

date_seen | signal_type | person | title | company | evidence_url | passes_filter (y/n) | message_sent (date) | outcome

Rules for the log:

1. evidence_url is the post, comment or profile URL that is the signal. Without it you cannot write a message that names what you saw, and in a week you will not remember.
2. passes_filter is filled in before message_sent. If it is "n", the row stays as a record so you do not re-check the same person.
3. outcome has four allowed values: no reply, reply, meeting, not a fit. Anything else makes the weekly count useless.

Companies that matter to your product rarely decide alone. Forrester's January 2026 buyer research puts the typical business purchase at 13 internal stakeholders and nine external influencers, with procurement acting as a decision maker in 53% of buying cycles (Forrester, 21 January 2026). You do not need all of them in the log, but when one person at a company fires a signal, add a second row for the person who owns the budget; the guide to finding decision makers on LinkedIn shows how to find them in a few minutes.

Step 5: Write one message per signal

The message names the signal in the first line, says one thing you learned from it, and asks a question the person can answer in a sentence. No pitch, no link, no calendar in the first message. Three templates, one per signal type, written for LinkedIn where the connection note is limited to 300 characters (a hypothetical payroll product again, so replace the specifics):

[Problem post, connection note]
Saw your post about month-end payroll taking three days. We hit the same wall at 60 people.
Curious what you tried before the spreadsheet: a provider, or in-house?

[Competitor post engagement, connection note]
Noticed you commented on [competitor]'s post about multi-state payroll. Are you comparing
options right now, or just following the topic?

[New hire, message after connecting]
Congrats on the Head of People role at [company]. In the first 90 days most people in that
seat either keep the payroll setup or replace it. Which way are you leaning?

Send the problem-post and competitor messages within a day of the signal; after a week the post is old news to the person who wrote it. The new-hire message can wait two or three weeks, until they have had time to form an opinion. The guide to writing a LinkedIn outreach message that gets a reply covers the follow-up and what to do when they answer.

Small, specific batches outperform large ones. Hunter's State of Cold Email 2025, built on 11 million cold emails sent in 2024, found an average reply rate of 4.1%, with campaigns of 50 or fewer recipients averaging 5.8% against 2.1% for campaigns of 1,000 or more (Hunter, State of Cold Email 2025). That is email, not LinkedIn, but the direction is the same: a message to ten people who did something this week beats a sequence to a thousand who did not.

Step 6: Set the daily routine and the sending caps

The routine is the same every working day, in this order, so that signals are caught before messages go out:

1. Open the bookmarked Posts searches and the notifications from the people you follow (15 minutes).
2. Log every hit, open each profile, fill in passes_filter (10 minutes).
3. Send the messages for rows that pass and have no message_sent date, newest signal first (15 minutes).
4. Answer replies before anything else the next morning.

Cap the sending. LinkedIn restricts accounts that send too many connection requests, and the restriction costs you more than any week of volume gains. Intentsly caps each connected LinkedIn account at 30 messages and 30 connection requests a day and recommends 15; use the same numbers by hand. At 15 connection requests a day you will run out of fresh signals before you run out of quota, which is the correct way round.

Do not send on weekends unless your buyers post on weekends. Do not send the same template to two people at the same company on the same day.

Step 7: Review weekly and drop what does not work

Every Friday, count the log by signal_type: hits, passed filter, messaged, replied, meetings. Four weeks of counts is enough to make the first decisions.

Replace one signal at a time. If you change the search, the message and the signal in the same week, you will not know which change did it.

A worked example: Ridgeline Payroll's first two weeks

Ridgeline Payroll is a hypothetical 12-person company selling payroll software to companies of 50 to 300 people in the UK. The founder sells. Her last ten customers: six had posted or commented about payroll pain within a month of buying, three had hired a Head of People or finance lead, one came from a referral. Signals chosen: problem posts, competitor post engagers, new People or finance hires.

Week one. She bookmarks three Posts searches (the strings in Step 3), follows two competitor pages and three HR influencers with the bell on, and sets up the nine-column log. Daily routine takes 50 minutes the first days, mostly because the first search string returns job seekers and payroll clerks; she adds NOT ("looking for" OR "open to work") and the hits halve. By Friday the log has 41 rows: 23 problem posts, 14 competitor engagers, 4 new hires. 19 pass the filter. She sends 17 connection notes, under the 15-a-day recommendation across the week.

Week two. Nine of the 17 accept. Five reply. Problem-post messages account for four of the five replies; competitor-engager messages get one. The new-hire rows are still waiting their two weeks. Friday review: the competitor-engager message is rewritten to quote the specific comment the person left rather than the post topic. Two of the five replies become calls the following week.

Nothing in this example is a result to expect. The counts are there to show what the log and the review look like at week two, when most people give up because "it is not working yet".

Where this goes wrong

Chasing every signal from day one. Six signals means six half-tuned searches and no idea which one works. Three, tuned over four weeks, then a fourth.

Messaging the signal without checking the fit. A founder posting about payroll pain at a 4-person company is a real signal and the wrong customer. The passes_filter column exists so you make this call deliberately, once, rather than discovering it on a call.

Sending late. A post from nine days ago is not a signal any more; the person has either solved it or moved on, and a message about it reads as surveillance. If the daily routine slips for a week, do not catch up by messaging the backlog. Clear the old rows as "not sent" and start fresh.

What to do with the list

The log is your list, sorted by date_seen descending. Each morning, the rows that pass the filter and have no message_sent date are the day's outreach, newest first, up to your cap. Replies get a two-line answer and, only if they ask, a time to talk. Meetings get the outcome column updated the same day, so the Friday review is honest.

When the routine is stable and you want the 40 minutes of checking back, try Intentsly without an account: it shows the LinkedIn buyers currently matching your profile, scored and ready to export into this same log, so the daily routine starts at Step 5.

FAQ

What is signal-based outbound?

Signal-based outbound is outreach sent because the person or company did something observable that suggests they are in the market: posted about a problem, engaged with a competitor, hired for a role, changed jobs. The message names that event. Volume outbound sends the same message to a list built on fit alone, with no timing information.

How many signals should you start with?

Three. Each signal needs its own search, its own message and its own line in the weekly review, and three is the most one person can tune at once. Start with the three events that most often preceded your last ten deals, run them for four weeks, then replace the weakest or add a fourth.

Do you need Sales Navigator or paid intent data?

No. Problem posts, competitor post engagers and open roles are all visible with a free LinkedIn account using the Posts search, the reactions list and company Jobs tabs. Sales Navigator adds saved searches and job-change alerts, which matter once you follow more than about 40 companies. Paid intent data is built for larger teams and bigger lists.

How fast do you have to act on a signal?

Within a day for posts and comments, because the person's attention is on that topic now and a week later it is not. New hires are the exception: wait two to three weeks so they have formed a view of what to change. If you miss the window, do not send the message late; log the row and wait for the next signal from that person.

What is a good reply rate for signal-based outbound?

Hunter's analysis of 11 million cold emails sent in 2024 found a 4.1% average reply rate, and campaigns of 50 or fewer recipients averaged 5.8% (Hunter). On LinkedIn, AI SDR outreach across Intentsly customers had a 21% reply rate as of October 2026. By hand, anything above the small-campaign email figure means the signals and messages are working.

Can one person run signal-based outbound?

Yes, and most of the time it is one person. The by-hand setup in this guide takes about 45 minutes a day in week one and 20 to 30 minutes once the searches are tuned, at 15 or fewer connection requests a day. The limit is not time but signals: a narrow market may produce fewer than ten qualifying hits a week, which is still enough to fill a calendar.