A weak LinkedIn campaign rarely has one broken line of copy. The offer may be vague. The list may include people who will never buy. The sender profile may look like a pitch before the prospect reads the message. Or the sequence may stop before most replies arrive.
GetSales co-founder Petr Kaliuzhny and Extrovert founder Oleg Sobolev shared data from both sides of that problem. GetSales measured more than 6 million connection invitations across 20,000-plus accounts in Q2 2026. Extrovert analyzed more than 150,000 connection requests, 800,000 direct messages and 850,000 comments from its users.
Use these numbers to diagnose where prospects drop off: targeting, acceptance, replies or follow-up.
Start with the right benchmarks
In GetSales' Q2 2026 network data, the median sender had a 21% connection acceptance rate and a 22% reply rate after acceptance. The top 10% of senders reached 42% acceptance and 38.5% replies.
GetSales Q2 2026 anonymized network data. The 42% and 38.5% figures describe the top 10% of senders, not a guaranteed before-and-after lift.
Treat these as diagnostic ranges, not universal promises. Petr noted that crowded categories such as custom software development can have lower acceptance rates because prospects see similar offers all day.
If your acceptance rate is well below 20%, inspect the list and sender profile before rewriting every message. If people accept but do not reply, inspect the offer, delivery and follow-ups. The gap shows how widely results vary among senders.
Give people a reason to answer
A generic description of your service is not an offer. “We offer LinkedIn automation at a good price” gives the prospect nothing concrete to grab.
Petr recommends building a no-brainer offer: specific value for a specific audience, with enough substance that the person can make a quick decision. In one GetSales experiment, a single email offering unlimited LinkedIn and email automation went to 2,000 people and booked 40 meetings in the same week. That result is specific to that offer and audience. The takeaway is to make the value obvious before adding personalization.
Value-first openers can also lower the effort needed to reply:
- “I put together a guide on how to double your LinkedIn acceptance rate. Want me to share it?”
- “We audited how often GPT and Claude surface your brand versus competitors. Want to see it?”
- “A client wants experts in your domain. Some make $10,000 a month on the side. Want an invite?”
These examples work because “yes” is easy and the promised value is clear. The asset or opportunity still has to be real. A dressed-up pitch with no useful give will not become better because it asks permission.
Once the prospect says yes, deliver what you offered and continue with a relevant question. That creates a conversation before you ask for a meeting.
Follow up, but know when the cold sequence is over
GetSales found that the first message received about a 15% reply rate among leads who reached it. The second produced about 7.5%, the third 4.5% and the fourth 1.7%. These rates are per message, not numbers you should add together.
GetSales per-message reply rates among leads who reached each step. The percentages are not additive.
Petr's practical recommendation is around three follow-ups in a cold sequence. One message leaves replies behind. Six or eight touches can feel like harassment and lead to blocks. After the cold sequence, move the prospect into a slower nurture motion rather than continuing the same chase.
Delivery matters too. Petr advises against putting a link or Loom video in the first LinkedIn message before a reply. Links can trigger LinkedIn warnings or filters. Oleg added that a truly personalized video may make sense for some mid-market or enterprise accounts, but his team found the effort hard to justify for routine outreach.
A better workflow is to ask whether the person wants a short personalized video. Record it only after they answer. You get consent, avoid spending time on people who are not interested and remove the link from the cold opener.
The same human-delivery rule applies to the sequence itself:
- clean company names, job titles and other imported fields before they enter a message;
- avoid obvious placeholder sentences such as “I came across your profile and saw you work at {company}”;
- do not visit, like several posts, connect and message within minutes;
- vary messages instead of sending one identical block at scale;
- monitor open rates where available, since a sudden drop may mean messages are landing outside the primary inbox.
Use automation for repetitive steps, but clean the variables, pacing and copy before anything reaches the prospect.
Use signals only after you qualify the list
Petr described auditing a campaign aimed at plumbers and local businesses that included tech CEOs and unrelated roles. The campaign had a 2% acceptance rate. Better copy would not repair the list.
Before launch, check:
- Buying role: Does this person own or influence the problem?
- Company fit: Is the company relevant even if the title looks right?
- Current employment: Does the person still work there?
- Geography and language: Can you communicate naturally in the market you selected?
- Reachability: Does the profile have a photo, a credible network and signs of real activity?
- Platform activity: Do they post, comment or react often enough to see your outreach?
Oleg's data makes the last point concrete. People posting two to five times a week accepted 43% of requests, compared with 25% among profiles that had not posted in a long time. This overlaps with the warm-up effect, so do not multiply the lifts. Use activity as a list-quality filter.
Signals can improve timing, but relevance still comes first. “Attended a webinar about LinkedIn conversions” can explain why a related playbook is useful now. “Liked a post” is not a reason to buy. Petr's framework is to use timely signals to strengthen a relevant offer, not replace one. A signal can explain timing; it does not prove purchase intent.
Use more than one route to the account
If a relevant prospect does not accept your connection request, the account does not have to disappear.
Petr recommends trying InMail, especially for open profiles where it is free, and using paid InMail selectively behind a strong offer. For prospects you cannot reach on LinkedIn, find and verify a work email, then send the same value-first offer at low volume.
Existing first-degree connections deserve their own path. They already know you or have seen your content, so do not automatically end a campaign when your system detects an existing connection. Review the conversation history and send a relevant message when there is a real reason to restart it.
Your profile is part of the message
Prospects often inspect the sender before accepting. A confusing headline, missing photo or generic banner can sink a strong campaign.
Petr's profile checklist is straightforward:
- use a clear, open photo, ideally with a visible face and natural smile;
- write a headline that communicates one understandable offer, achievement or point of view;
- avoid piling GTM, RevOps, AI and every tool you use into one line;
- use the banner to support what you do and why it matters;
- publish useful posts so new connections keep seeing evidence after they accept.
A blank connection request is a sensible default when you have no meaningful note. Petr found the acceptance rates for notes and blank requests were similar after recalculating his data, so he chose not to publish a hard comparison. A real referral or specific reason to connect can make a note useful. Generic filler adds risk without adding context.
Warm the request while your name is still fresh
Extrovert's data shows how familiarity changes acceptance. A cold request from a stranger was accepted 27% of the time. After one or two comments on that person's posts, acceptance rose to 41%. More comments did not keep increasing the result; after four or more, acceptance fell to 36%.
Extrovert observed 27% acceptance for cold requests and 41% after one or two comments—a 14-percentage-point association, not proof of causation.
Timing matters. Among warmed requests, those sent within 24 hours of the last comment had a 44% acceptance rate. Waiting a month brought it down to 33%.
The practical play is small:
- Choose an active, well-qualified prospect.
- Leave one or two useful comments on posts where you have something real to add.
- Send the connection request while your name is still familiar.
- Stop commenting if you are forcing it. Familiarity helps; a sudden wall of attention feels strange.
If prospects rarely publish, Oleg suggests engaging in their orbit: comment on posts from colleagues or industry voices they already follow and engage with. Extrovert's data showed a 41% acceptance rate after this kind of indirect warm-up, close to the direct-comment result. Since there is no guarantee the prospect sees an indirect comment, use it as a visibility tactic rather than pretending it is a direct interaction.
The orbit model means contributing to posts the prospect may already see. The exact 41% comparison comes from a smaller observational cohort and does not guarantee that any individual prospect saw the comment.
Keep the first message useful and easy to answer
Extrovert measured first-message reply rates by length. Very short messages of 50–100 characters received 16% replies. Messages between 150 and 300 characters reached about 30%. Messages longer than 800 characters reached 21%.
The middle range gives you room for two or three sentences: one specific observation about the prospect, one useful idea and one easy question. It is long enough to create a reason to answer but short enough to avoid a full pitch.
The subject matters more than polishing the sentence. In Extrovert's data, messages that referenced the prospect's post received 5–8 percentage points more replies than messages centered on the sender. The same approach becomes even more useful in follow-ups. A product launch, new hire, conference, post or other current event gives you a natural reason to restart the conversation.
For example, ask how a recent launch went rather than sending a third version of your value proposition. Once they answer, you can continue the conversation and introduce the commercial angle when it fits.
Stay visible after the cold sequence
Cold sequences usually end before many prospects are ready to buy.
Extrovert tracked conversations where the first message had gone unanswered for two weeks. When the sender went silent, 2% of those conversations later returned. When the sender kept lightly commenting and followed up when there was relevant news, 19% returned. The sample was modest and observational, so treat the result as directional. Stay present with relevant touches rather than extending the same cold sequence.
Comments, DMs and your own posts work together here. A comment keeps your name visible. A contextual DM reopens the thread. Useful content gives the prospect something to learn from between direct conversations.
Speed matters once they do reply. Across more than 155,000 exchanges in Extrovert's data, 83% of conversations continued when the sender answered within an hour, versus about 66% when the delay exceeded four hours. Some of that spread may reflect naturally active conversations, but the operational lesson is sound: route prospect replies into an inbox someone checks and respond while the exchange is live.
Keep a human between AI and the prospect
AI can help clean variables, summarize a post, find a useful angle and draft comments or DMs. It cannot reliably judge every social context.
Oleg said he approves about 70% of his suggested comments as written and edits roughly 30%. The edits may be small: remove a phrase, change the rhythm or add a line break. Sometimes the right decision is not to comment, especially on illness, grief, personal news or an inside joke the model does not understand.
Extrovert uses the prospect's post and profile plus the sender's style, company context and talking points to draft suggestions. The user reviews and sends them. That boundary matters because a slightly wrong public comment can damage the sender's reputation in front of the prospect's whole network.
Use AI to reduce effort per prospect. Keep the human responsible for relevance, taste and timing.
A practical LinkedIn workflow
Put the pieces together in this order:
- Benchmark the current motion. Separate low acceptance from low post-acceptance replies.
- Fix the offer. Give a narrow audience a concrete reason to answer.
- Clean the list. Verify role, company, employment, geography, reachability and LinkedIn activity.
- Check the sender. Make the photo, headline, banner and recent content support the offer.
- Warm priority prospects. Leave one or two good comments and connect while your name is fresh.
- Write a 150–300 character opener. Mention something specific about them and ask one easy question.
- Run a short cold sequence. Use around three follow-ups, then stop pushing.
- Add other channels. Try relevant InMail or a low-volume value-first email when LinkedIn does not connect.
- Move quiet prospects into nurture. Comment lightly and follow up when there is actual news.
- Reply fast and review every AI-assisted touch. The system can suggest; the sender owns what goes out.
FAQ
Should I add a note to a LinkedIn connection request?
Use a note when you have a real referral, shared context or specific reason to connect. Otherwise, a blank request is a reasonable default. Petr recalculated GetSales data and found notes and blank requests had similar acceptance rates overall, so there was not enough evidence for a universal winner.
How many LinkedIn follow-ups should I send?
Petr recommends around three follow-ups in the cold sequence. GetSales' per-message reply curve declines from about 15% on message one to 7.5%, 4.5% and 1.7% on the next messages. After the cold sequence, switch to slower, contextual nurture rather than continuing to push the same offer.
Should I send a Loom or another link in the first message?
Petr advises against links before the first reply because they can trigger LinkedIn warnings or filters. Ask whether the prospect wants the resource or personalized video, then send it after they answer. Oleg noted that personalized Loom videos may still make sense for selected mid-market or enterprise accounts when the potential value justifies the manual work.
How can I tell whether a prospect is active on LinkedIn?
Check when they last published an original post and how often they post, comment or react. Oleg cautioned that Sales Navigator's “posted on LinkedIn” filter can include reposts or old posts. Extrovert exposes posting recency and frequency for prospects, and Clay can also enrich activity data.
Can AI comments run fully automatically?
Oleg's answer was no. Even fine-tuned models can miss tone, private context or sensitive subject matter. Use AI for drafts, then have a person edit, approve or skip each comment.
Does the day of the week matter for connection requests?
Not in Extrovert's dataset. Acceptance stayed around 30% across all seven days in roughly 127,000 requests. Extrovert also found that 89% of acceptances happened within 14 days, so old pending requests contribute little after two weeks.
