LinkedIn prospecting: find the right people, then write.

Write down who you sell to before you search

Most weak prospecting starts in the search box. Someone types a job title, gets tens of thousands of results and starts sending. Do it the other way round: write down who buys from you, then build the search that finds exactly those people.

The reason now matters most. Two people can match every filter, and only one of them has a reason to answer you this month. That reason becomes your first line, so if you cannot name one for a segment, the segment is not ready.

Look at the last ten deals you won, not the market you wish you had, and write down what those buyers share:

• Title and seniority: the person who signs, and the person who feels the problem first
• The company: headcount, industry, region and the stage it is at
• The reason now: a new sales team, a new office, a new leader, a post about the problem
• Who you skip: competitors, recruiters, students and companies too small to buy

In GetSales you write this as one sentence, such as "heads of sales at UK SaaS companies", and AI Search turns it into filters you can read: titles, industry, company size and location.

Get your profile ready before anyone sees it

Expect the people you write to to open your profile before they accept, so treat it as the landing page of your outreach. A stranger decides in seconds if you are worth hearing from.

Line the profile up with the message. If your headline says you help sales teams hire their first SDRs, your first notes should go to people for whom that is true.

Fix four things before the first request goes out:

• A clear photo, and a headline that says who you help and with what, more than your job title
• An About section in plain words: the problem you solve, for whom, and one example
• A Featured item a buyer would open: a short case, a guide or a talk
• Some recent activity, a few posts or comments in the last month, so the profile looks used

GetSales writes from your own sender profile, your LinkedIn account and your mailboxes, so the profile your prospects open is yours.

Build the search: filters, Boolean strings, saved searches

Start with filters, since they are exact. LinkedIn's people search filters by connection degree, location, current company and industry. Sales Navigator, if you hold the licence, adds seniority, company headcount, function, years in role and filters for people who changed jobs or posted recently.

Use the keyword box for what the filters cannot say. LinkedIn reads Boolean operators there: AND, OR and NOT typed in capitals, quotes around an exact phrase and parentheses to group. It reads quotes first, then parentheses, then NOT, AND and OR, and it has no wildcard, so spell out each form of a word (LinkedIn Help, article a524335, read 24 October 2026). Put every OR group in parentheses: without them, "VP sales" OR "head of sales" AND SaaS means any VP of sales, plus heads of sales who mention SaaS. Write your strings in a plain text editor, since a word processor curls the quotes, and if you have Sales Navigator, try the same strings in its keyword field.

When the first page looks right, read 20 results. If more than a few are wrong, tighten the string before you go on. Then save the search: in Sales Navigator a saved search flags the new people who match later, so the list grows without another session of typing.

If you describe the buyers to AI Search instead, its filters follow the same logic: the values inside one filter are joined by OR, the filters by AND. It has no NOT, so for a string with an exclusion, build the search in LinkedIn and paste its link. Strings to copy, with your own words in them:

Paste a Sales Navigator search link into AI Search: the results land in your table a page at a time, and the search keeps going with the app closed.

Look past search: posts, events and job changes

Search finds people who match. The best prospects are often the ones who just did something: wrote about your problem, commented on someone else's post, joined an event in your field or started a new job. Each of those hands you the reason for your first line.

Work them by hand at first: read the post and the comments, note who said what, and write to the people whose comment gives you something to answer. A reply to their own words beats any template. The places to look:

• People who reacted to or commented on a post about the problem you solve, your own post, a peer's or an industry voice's
• Attendees of a LinkedIn event in your field, once you have joined it yourself
• Members of a group you take part in, where the conversation is already going
• People who started a new role in the last few months, since a new leader looks at vendors early
• Companies hiring for the team you help: a job post for SDRs says outbound is on the plan
• People you wrote to months ago who just posted or changed jobs: a second chance with a fresh reason

Paste a post link into AI Search and it brings in who reacted, then who commented. For people already in your tables, Signals shows a job change or a new post a workflow reads, and a play you switch on answers it.

Qualify and enrich before you write

A list that matches the filters still holds people you should not write to. Before anyone gets a request, read the profile, the last few posts and the company page, and sort each person into one of three groups: a clear fit with a reason now, a fit with no reason yet, and no fit.

Write only to the first group. Park the second until something changes, such as a post, a hire or a new role. Drop the third, and put your customers and open deals on a do-not-contact list so nobody writes to them by mistake.

While you read, note the one detail you will open with, next to the name. Find a work email for each fit as well, since some will not accept on LinkedIn and email is the second way in, and check every address before you send to it.

In a Playtable an AI column scores each person against your words and keeps its reason, the table looks up each work email and checks it, and your stoplist covers every table and every workflow.

The first touch and the follow‑ups, with message examples

Send the connection request with a short note, or with none. A note works when it names something real about the person in a sentence or two, with no pitch, and no note at all beats one that could have gone to anyone. The best first lines fit one person only, such as "Saw the new Leeds office. New region, same two people on outbound?"

Once they accept, send one message with one easy question, and keep the pitch for when they have answered. If they do not answer, follow up two or three times, a few days apart, and make each follow‑up add something: an idea, a number, an example from a team like theirs. Stop the moment they reply. If they have not accepted in about a week and you found their work email, write to them there instead.

A GetSales workflow sends the visit, the note and the follow‑ups from your sender profiles, paced within each account's limits, and by default a reply pauses the follow‑ups.

Limits, and the tools you can drop

LinkedIn sets the limits on how many requests and messages an account can send, and it can change them. Guides quote weekly figures for connection requests, but LinkedIn promises none, so treat any number as a ceiling to stay well under, never a target. The habits that keep an account in good standing are plain: write to fewer people with better reasons, so more of them accept, give a new account a slow start, spread the sends over the working week, and withdraw requests nobody answered after a few weeks.

In GetSales every step is paced within each account's limits, inside the workflow's sending window, and a new LinkedIn account gets a gradual ramp.

By hand, LinkedIn prospecting usually runs on four tools held together with spreadsheets, and every move between them is a copy, a paste and a chance to write to the same person twice:

Say who you sell to and Creator plans the search, the table, the work emails, the first lines, the workflow and the AI SDR, then asks before anything starts.

Measure acceptance, replies and meetings

Count four numbers for each search or segment, never only for the account as a whole. Then change one thing at a time, the list, the note or the first message, and give each change a couple of weeks and about a hundred people before you judge it.

Good acceptance with weak replies means the message needs work. Weak acceptance with good replies means the note or the list does. And answer fast, since a reply left for a day goes cold. The four numbers:

• Acceptance rate, the requests accepted over the requests sent: when it is low, the list or the note is off
• Reply rate, the people who answered over the people who accepted: when it is low, the first message asks too much or says too little
• Positive replies, the answers that want to talk, which matter more than the count of all answers
• Meetings booked, and later the deals that came from them

Each sending step of a workflow shows what it sent, who replied and the rate, and every reply from every sender profile lands in Unibox, where the AI SDR drafts each answer and each scenario sets whether it sends on its own or waits for your OK.