ResearchAnalysis

Personalized vs Template LinkedIn Messages: Company and Role Tokens Did Not Beat First Name

Our analysis of 355 distinct first messages sent to 58,207 accepted LinkedIn connections found that adding company or role tokens did not beat first-name-only personalisation: 14.62% of recipients replied against 17.99%, and 13.47% against 18% when every campaign was weighted equally.

Dataset. Reply to the first message after acceptance by the personalisation tokens it used, across 58,207 accepted LinkedIn connections and 355 distinct first-message texts, from requests sent 8 January 2025 to 16 October 2026, pulled 16 November 2026.

By Jordan Kwan · Published Nov 16, 2026

TL;DR: Merging a prospect's company or job title into the first LinkedIn message did not lift replies. Messages with first name only drew 17.99%; adding company or role tokens drew 14.62%. Results for AI-written lines are mixed, so this study makes no claim about them.

Our analysis of 355 distinct first messages sent to 58,207 accepted LinkedIn connections found that adding company or role tokens did not beat first-name-only personalisation: 14.62% of recipients replied against 17.99%, and 13.47% against 18% when every campaign was weighted equally. Our analysis

Key Takeaways

  • First messages using only a first-name token drew a 17.99% reply rate (18% campaign-weighted).
  • Adding a company or role token drew 14.62% (13.47% campaign-weighted). Both measures point the same way.
  • First-name-only was the dominant format: 259 of 355 distinct messages and 74.45% of recipients (derived).
  • AI-written or custom-field lines show 14.25% pooled but 18.32% campaign-weighted. The measures disagree, so no claim is made about them.
  • 63 recipients whose company or title token would have rendered blank were excluded before any comparison.

What did we measure?

Personalisation is the most repeated advice in cold outreach, and "personalized LinkedIn messages statistics" is a common search. Most of what comes back is about personalisation in the broad sense: research, relevance, a genuine reason to reach out. This study measures something narrower and countable: the merge tokens in the first message after a connection is accepted.

Reachium's outreach sequences store each message's text with its placeholders, such as , or . That makes it possible to group first messages by which tokens they use and compare reply rates, across 355 distinct first-message texts.

How was it measured?

Dataset window. Connection requests sent between 8 January 2025 and 16 October 2026, pulled 16 November 2026, matured at least 30 days. InMail sends are excluded.

The first DM. The first message node inside the "connection accepted" branch of an outreach sequence, read as stored at the pull.

Denominator and outcome. Matured, accepted requests that were sent the first DM. "Reply to the first DM" means the person replied while the first DM was still the last message sent; sequences stop on a reply, so the reply belongs to the first message, not a follow-up. "Replied at any point" is shown alongside. Our analysis

Buckets. No tokens; first name only; first name plus company or role (, or ); and AI-written or custom-field line (a token generated per recipient or filled from a custom field).

Exclusions. Accepted requests whose first DM was never sent (10,402); requests whose copy had been edited after sending (55); campaigns whose first DM carried more than one variant text; and the blank check described next. Clean sample: 58,207 accepted people, 355 distinct first-message texts, overall reply to the first DM 17.05%. Our analysis

The blank check. A , or token renders blank when the recipient's profile has no company or title on file, which would turn a personalised message into a broken one. 63 recipients were in that position. They were excluded before any comparison. Our analysis

The effective sample is messages, not people. Every row reports distinct messages and a campaign-weighted reply rate (campaigns with 50 or more recipients, each weighted equally). A bucket with fewer than 10 distinct messages is labelled too few messages to read.

What did we find?

Personalisation in the first DMPeople who got the first DMReply to first DMReplied at any pointDistinct messagesCampaign-weighted replyLargest workspace share
No tokens (too few messages to read)3428.77%11.11%610.88%51.75%
First name only43,33417.99%26.97%25918%20.52%
First name plus company or role7,47114.62%24.94%5713.47%27.16%
AI-written or custom-field line7,06014.25%24.04%3318.32%29.26%
Reply to the first LinkedIn message, first name only vs adding company or role
First name only17.99%
+ company or role14.62%

Company and role tokens did not beat first name. Pooled, the token bucket trailed first-name-only by 3.37 points; campaign-weighted, by 4.53 points (both derived). The comparison rests on 57 and 259 distinct messages, both well above the readability floor. Replied at any point shows the same order, 24.94% against 26.97%. Our analysis

AI-written lines: no claim either way. Pooled across people, the AI-written or custom-field bucket replied at 14.25%, below first-name-only. Weighted by campaign, it replied at 18.32%, slightly above. A pooled rate below the weighted rate means the larger campaigns in this bucket replied less than the smaller ones, and the data cannot say which picture describes AI lines in general. This study does not say AI-written lines raise or lower replies. Our analysis

No tokens: too few messages to read. Six distinct messages, with one workspace accounting for 51.75% of its recipients. The 8.77% figure is not a finding about untokenised messages.

What does this mean for personalisation?

The finding is narrow and worth stating precisely: merging a company name or job title into a first message did not outperform addressing the person by first name. It does not show that relevance, research, or a real reason for writing are worthless. Those are not what a token measures. A token inserts a field; it does not make the sentence around it more relevant.

One plausible reading is that recipients recognise merged fields for what they are. An illustrative opener such as "I saw you work at [Company] as [Title]" (written for this page, not taken from the data) is a familiar template shape in cold outreach. Another is that teams who reach for company and role tokens write differently in other ways. The data constrains the theories without settling them. Token use travels with message length and with how the message ends, both measured on the same 58,207 people.

The blank check matters for anyone running their own version of this analysis. Profile fields can be missing: a companion study of how complete LinkedIn profiles are measures that directly. A token sent to someone with no title on file produces a visibly broken message, and including those sends would have made the token bucket look worse for the wrong reason.

Elsewhere on the platform, a note on the connection request and one follow-up are measured separately. The full funnel these first messages sit inside is in the outreach benchmarks. Teams reworking their personalisation can compare setups on Reachium's pricing page.

How do you use this in practice?

An SDR whose first message opens with the prospect's company and title is writing in the bucket that replied at 14.62% here, against 17.99% for first-name-only messages. The data does not say removing the tokens would lift that SDR's replies, only that the extra tokens did not come with more replies across 57 distinct messages. Our analysis

A team lead evaluating an AI personalisation tool has no answer from this study. The bucket's pooled rate (14.25%) and campaign-weighted rate (18.32%) point in opposite directions, and AI output cannot be checked for blanks from stored data. Our analysis

A RevOps lead building a list for a or template can see the failure mode the blank check guards against: any recipient without that field on file receives a gap where the personalisation was meant to be. Our analysis

The Carlisle Capital Group case study shows one team's outreach numbers on the same platform.

What are the limits of this study?

Observational, not an experiment. Workspaces chose their own copy and tokens. Nobody was assigned a personalisation level, so targeting, offer and the rest of the message ride along. A/B variants are excluded, not analysed.

Tokens are not personalisation in full. A message with only a first-name token can still be highly specific, and a message with three tokens can be generic. This study counts placeholders, not relevance.

AI and custom-field output is unchecked. These tokens are generated or filled per recipient, and stored data cannot show whether any rendered blank or wrong. The blank check covers , and only. Our analysis

Messages, not people, are the sample. 355 distinct texts. The company-or-role bucket has 57 and the AI bucket 33.

Concentration. No readable bucket's largest workspace exceeds 29.26%. The no-token bucket is 51.75% one workspace and is not read.

Reply is not interest. This study counts replies, not positive ones. Teams that want help writing first messages that read as personal can talk to the agency team.

FAQ

Does personalization increase reply rate on LinkedIn?

Adding company or role tokens did not: 14.62% replied against 17.99% for first-name-only messages, and the campaign-weighted figures (13.47% against 18%) agree.

Are personalized LinkedIn messages better than templates?

It depends what "personalized" means. Merged company and role fields did not beat first-name templates in our analysis of 355 messages. Genuine research and relevance are not what tokens measure.

Do AI-personalized LinkedIn messages get more replies?

This study cannot say. AI-written lines replied at 14.25% pooled but 18.32% campaign-weighted, so the measures disagree.

What happens when a LinkedIn personalisation token is blank?

The message goes out with a gap. 63 recipients in this data would have received a blank company or title, and they were excluded before the comparison.

Sources

How to cite this study

Reachium (2026). Personalized vs Template LinkedIn Messages: Company and Role Tokens Did Not Beat First Name. Reply to the first message after acceptance by the personalisation tokens it used, across 58,207 accepted LinkedIn connections and 355 distinct first-message texts, from requests sent 8 January 2025 to 16 October 2026, pulled 16 November 2026. https://www.reachium.io/research/personalized-vs-template-linkedin-messages

Related research

Your competitors are automating.
You're still copy pasting.

Try Reachium

Live in 10 minutes.