How to win LinkedIn outbound in 2026 as a founder

The harsh truth: why so many founders fail at outreach

Blub. You received a message.

You open your LinkedIn DMs and you already know what's coming. But you give it a shot.

Then, pure disappointment. Just a waste of time. A wall of text about something you do not care about.

Em dashes. Meaningless convoluted sentences. A fake compliment. A calendar link.

Closed in three seconds.

If you do not care, why should anyone else.

Four things are killing founder outreach right now.

People are annoyed by AI slop

More DMs are generated by AI than ever. They are boring and they all sound the same.

The rising aversion is justified.

Missing authority. Just another stranger

People get ten pitches a day. They do not care, because they never heard of you before.

Never saw your face or name in any context that matters to them. They skip and forget five minutes later.

Wrong person at the wrong time

Most messages are a result of laziness. The sender skipped the research.

They do not know if the recipient is a good fit, or if the offer would be helpful right now.

LinkedIn is more competitive than ever

On a growing platform there are tons of businesses promoting their stuff. People can smell a sales pitch from three kilometers against the wind.

A charming opener about your business may have worked two years ago. Now it is a dead end.

The good news: most senders are doing the lazy thing. The bar to stand out has never been lower. You just have to not be them.

The founder discount (we will keep coming back to this)
You can say things a sales rep cannot. You can admit you are early. Ask weird questions. Send a Loom from your kitchen. Quote someone's own post back at them without sounding like a stalker. Use this. It is your unfair advantage over every SDR in your prospect's inbox.

To tackle the four problems above, there are eight straightforward steps. Some sections include external tools and prompts you can paste into Claude or ChatGPT to spin things up faster.

How to use this guide

Treat this as a worksheet, not an article. Your progress saves automatically when you check things off. Some sections give you prompts that you can use with any AI you like (Claude, ChatGPT, Codex) to get more out of the step. The prompts are designed to help you think, not to generate messages for you. Big difference.

If a section does not apply to you, skip it.

1. A quick profile review

No worries, this does not become one of those "YOUR PROFILE IS SUPER IMPORTANT" cliché lectures. We stick to the basics.

But why does it matter at all: if your message, comment or post resonates, people will check your profile. So it has to not embarrass you.

Friendly profile picture. A smile beats a "confident founder gazing into the middle distance" shot by a wide margin. I promise.

Banner that says something about what you do. (A landscape is not interesting.)

Two or three relevant posts from the last six months, with visuals. You do not need to be an influencer. You need to look alive.

Featured section pointing to one genuinely useful free thing, if you have one.

Yes, that is it. You could optimize the title, the about section, the work history. But the truth is, most people scan and rely on visuals.

If you are looking for banner inspiration, check the Figma template pack or browse the Canva banner templates.

If you want a deeper diagnosis, run the free Social Selling Index calculator on your profile.

2. Use your ideal customer profile (and test three)

You probably already heard of this. To keep it short: your ICP is the buyer you are going after. The role, the company, the team they sit on.

Here is the thing nobody says out loud. Try multiple ICPs.

Many founders specify one ICP, run outreach, get nothing back, and conclude outreach does not work. Please do not. That is the worst thing you can do at this stage.

Especially in the early days, the most important thing is product market fit. Outbound right now is not pipeline generation. It is product market fit research disguised as sales.

Run three ICPs in parallel. Kill the losers. Double down on the one that replies.

Write down three distinct ICPs to test (not three flavors of the same one).

For each ICP, fill in: role, company size, industry, and the trigger event that makes them care this quarter.

Build a list of 50 named people per ICP.

Decide your kill threshold per ICP. A number, not a feeling.

I am a founder building [one sentence about your product]. My current ICP hypothesis is [role] at [company type]. Do not generate ICPs for me. Instead, ask me ten sharp questions, one at a time, that pressure test whether this ICP is real, urgent, and reachable. After my answers, tell me which of three ICPs (current plus two adjacent ones we discover) you would test first, and why.

Founder discount
You can say "I am testing whether founders like you would care about this" out loud, in the actual DM. A sales rep cannot. Honesty is a feature here.

3. Find the right people

Sourcing is boring. It is also where most founders burn the most time. Three options, ranked by pain:

Sales Navigator. Worth it once your ICP is sharp. Useless before. Build one search per ICP, save it, work the list down.
Regular LinkedIn search. If you do not have Sales Navigator yet, the free people search works. Blunter filters, slower to scan, but no paywall. Better than nothing on day one.
Engagement based sourcing. Find a post by a thought leader your ICP follows. The people commenting are warm. They self selected as interested in the topic. Highest signal source on the platform and almost free.
Event, podcast and community lists. Speakers and attendees of niche events convert better than cold lists by a wide margin.

Build one saved Sales Nav search (or equivalent list) per ICP.

Grow each list to at least 50 named people.

Scrub the list of obvious bad fits before you start messaging.

I am building a Sales Navigator search for [ICP role] at [company type]. Help me write the boolean filters: titles to include, titles to exclude, seniority, headcount, industries, geographies. Then give me three "second guess" filters that might surface a better adjacent ICP I have not thought of.

4. Warm up before you DM

Single biggest lever in this whole playbook. Almost every guide buries it.

Do not send cold cold DMs in 2026. Be visible to your prospect three times before you ever message them.

The math is brutal.

A first DM from a stranger gets opened, scanned for two seconds, dismissed. The same DM from "the person who left a thoughtful comment on my post last week" gets read.

The 3-touch warm sequence

Day 1: follow them. Like one recent post. Not the most recent. The one before. Shows you actually read.
Day 3 to 7: leave one real comment on something they wrote. Real means: an opinion, an example, a counter point. Not "great post!"
Day 7 to 14: send the connection request. No note, or with one line that references your comment.
Day 14 onwards: now you can DM. They have seen your name at least three times. You are no longer a stranger.

What counts as a real comment?
Real: "We tried this in 2024 and it backfired because X. Curious if you saw the same."
Not real: "So true!" / "Great insights!" / "This!"
If a bot could have written it, do not post it.

Founder discount
Founders can comment "we are a tiny team figuring this out, your post helped." Sales reps cannot do that credibly. You can.

5. Write the opener (kill the slop)

You did the warm up. Now you have to actually write the thing. This is where it falls apart for most founders.

The four fingerprints of AI slop

Em dashes. Nobody types these on a phone. If your DM has them, a model wrote it.
"I hope this finds you well." Or any variant. Filler. Cut it.
Vague compliments. "Love what you are building." About what, exactly?
Three paragraph walls. A founder writing in earnest writes short.

The opener framework

One sentence of specific observation (something only someone who actually read their stuff would notice).
One sentence of relevant insight (why that thing matters, or what it made you think).
One sentence of soft ask (low cost, optional to engage).

Sixty words, max. If you cannot say it in three sentences, you have not figured out what you want yet.

Three tests before you hit send

The delete-the-first-line test. Delete your first sentence. Does the message still work? If yes, your first line was filler. Send the trimmed version.

The specificity ratchet. Every sentence either gets more specific or gets cut. "I saw your post" becomes "I saw your post about pricing churn." "About pricing churn" becomes "about pricing churn after a free trial extension." Specific or gone.

The "say it at the bar" test. Read it aloud. Would you say this to them if you bumped into them at a conference? If you would feel ridiculous, the message is wrong.

Here is my draft DM: [paste].

Rewrite under these constraints: no em dashes, max 60 words total, three sentences, no compliments, no "hope this finds you well" filler. Keep the specific observation, sharpen it. Then explain in one line what you cut and why.

Founder discount
Quote their own words back at them. "You wrote last month that [exact phrase]. We hit that exact wall in February. Curious if you ended up solving it the same way." A sales rep cannot do this without sounding off. You can, because you are actually in the trench.

6. The soft ask

"Jump on a quick call?" is dead in 2026. You are asking for the most expensive thing your prospect has (time on a calendar) before you earned a coffee's worth of attention.

Better soft asks, ranked by how often they get a yes

"Want me to send the one pager?" (cheap, useful, async)
"Would a 90 second Loom be more useful than a paragraph?" (you do the work)
"Curious if you would push back on this take, no agenda." (flatters intelligence, no commitment)
"Mind if I send you the playbook we wrote on this?" (gift, not extraction)

All four have one thing in common: the cost to the prospect is under two minutes, and you give before you take. That is the whole game.

My offer is: [one sentence about what you do].
The prospect is: [role and what they care about].

Generate three soft-ask variants for my first DM. Constraints: each ask costs the prospect under two minutes, each one gives something before asking for anything, none of them say "jump on a quick call." Then tell me which of the three you would test first and why.

7. Follow up without nagging

Most follow ups say "just bumping this." This is a confession that you have nothing new to add. Stop sending them.

Every follow up does exactly one of four things

New angle. A different way the thing might be relevant to them.
New artifact. Something useful you made or found that fits their situation. Send it. No ask attached.
New question. One sharp question that proves you have been thinking.
Soft breakup. You acknowledge they have not replied. You move on. The door stays open.

If your follow up does not fit one of those four, do not send it.

Things that are not follow ups

"Just bumping this to the top of your inbox."
"Wanted to make sure you saw this."
"Circling back on the below."
"Following up." (Especially as the entire message.)

A sane cadence

Day 0: first DM.
Day 4: new angle or new artifact.
Day 11: new question.
Day 25: soft breakup.
Six months later: genuine re-engagement if their situation changed.

If you have not gotten a reply by day 25, you will not. Move on with grace.

Here is my first DM: [paste].

They have not replied. Write follow up #2 using the "new artifact" pattern: I will send them something useful (Loom, one pager, benchmark) with no ask attached. Suggest three artifacts I could realistically make in under 30 minutes. Pick the best one and draft the follow up around it.

For each ICP, draft a 4-touch follow-up sequence (Day 0, Day 4, Day 11, Day 25).

Tag each touch with the value it adds: new angle, new artifact, new question, or soft breakup.

8. Measure, then iterate

This is the step everyone skips, because it is the one that proves your message is bad.

Four numbers worth tracking

Reply rate (anyone who replies, including "no thanks").
Positive reply rate (replies that open a real conversation).
Meeting booked.
Qualified opportunity (the meeting was actually with a fit).

On benchmarks

You will see "17 to 30% reply rate" claims everywhere. Ignore them.

The number is meaningless without cohort, ICP, sample size and message style. A 30% reply rate from a known thought leader to their warm network is not the same as 30% from a stranger.

A useful internal heuristic instead: send 50 messages per ICP before you decide anything. Below 50, you are reading noise. Past 50, the pattern usually becomes obvious.

When to kill what

Reply rate fine, positive reply rate low: your ICP is wrong. Kill the ICP.
Reply rate low across the board: your message is wrong. Rewrite the opener, keep the ICP.
Both low: kill the ICP and rewrite. Start over with a sharper hypothesis.

Set up a simple log: date sent, ICP, opener variant, replied, positive reply.

Send 50 messages per ICP before drawing any conclusions.

Calculate reply rate and positive reply rate per ICP and per opener.

Write next week's hypothesis based on what the numbers say.

AI prompts as scaffolding, not output
Every copy-paste prompt in this playbook helps you think. None of them write the final message for you. On purpose. The single fastest way to sound like every other DM in your prospect's inbox is to let a model write end to end. Use AI to interrogate your ICP, pressure test your opener, brainstorm artifacts. Then write the actual DM yourself, in your voice, on your phone, typos and all.

How we actually use this at Sonarly

We live in this problem too. So before you start applying the eight steps, here is the version we are running right now, on ourselves.

Our current top ICP

B2B SaaS founders who sell to other B2B companies. Not B2C SaaS. Not agencies. Narrow on purpose.

We test three ICPs at a time and this one is the current winner. The other two are kept alive at low volume in case the answer changes next quarter.

The warm-up artifact we send most often

This workbook. The thing you are reading right now.

We send it ahead of any ask. No follow up required. No calendar link attached.

If you arrived here from a Sonarly DM, the message said something close to: "saw your post on X, we wrote a workbook for B2B founders on outbound, want it? No agenda."

That is the soft ask from step 6 in action. The cost to you: a click. What you got: the whole playbook. We took nothing first.

What the numbers look like

About 50 to 60% connection acceptance. About 30 to 40% reply rate within the cohort.

Now apply the benchmark warning from step 8 to ourselves. This is a specific cohort (B2B SaaS founders selling B2B), with the full warm-up sequence applied, with a workbook as the artifact.

The number is real, but it is ours. Yours will look different until you find your own combination.

What we learned

Switching the ask from "want to jump on a quick call?" to "want the workbook?" did more for our reply rate than anything else we tested. The artifact is what carries the message. The message just has to point at it.

Want the sequence templates we use?

We turned our full 4-touch sequence into an Excel sheet you can copy. The exact day-by-day actions, the artifact swaps, the kill-switches we use when an ICP is dying.

What now

Pick one ICP. Send ten messages this week, written by you, following the framework. Track the four numbers. Come back next week.

That is the whole job for the next seven days. Anything else is procrastination.

If you want tools that help with the parts that are actually worth automating (sourcing, tracking, warm up sequencing), our free tools are a decent place to start. None of them write your messages for you. That is on you. That is the point.