Why does a high LinkedIn reply rate not mean more meetings?

On LinkedIn, replies are the easy part: in our published studies, between 29% and 44.8% of the people messaged wrote back. Conversion is not easy, and it varied by audience and offer even when the agent and the reply rate stayed the same. Measure meetings held per person messaged, and test audience and offer before wording.

Why does a high reply rate not mean a good campaign?

A reply is any message back. It includes "thanks for connecting", a thumbs up, a question about what you do, and a polite no. Only some of those are a buyer.

In the wealth advisor network study, 44.8% of people messaged replied, but 33.7% replied with something real. About 44% of first replies were a pleasantry or a quick answer (study). In the foundry study, a list of people who engaged with competitors' pages replied at 58.2%, and 57% of those replies came from recruiters, vendors and peers rather than buyers (study).

A model that writes a relevant opener and answers in two minutes will get replies on most lists. That says the opener works. It does not say the offer sells to this audience.

What do reply rates look like across our studies?

These are Sellinger run campaigns on LinkedIn. Each row has its own base and its own definition, so read each row against its own study. Do not average them.

The reply rates sit in a narrow band. The outcomes do not, and they do not follow the reply rate. The highest reply rate in the table belongs to the study that counts calls booked, not held.

What changes conversion when the agent stays the same?

The same agents, on the same account, with the same wording style, produced very different results for different audiences and offers.

Three things stand out.

• The audience set the ceiling. On our own account, people who engaged with posts by AI SDR vendors and automation creators booked 2.6% to 3.7% meetings per person messaged. People who engaged with enterprise sales voices booked under 0.4%.
• The offer can cancel the audience. An agency added a Connected TV option to its pitch. Replies continued, but the 181 conversations where it came up produced no meetings (Uberwood).
• A higher reply rate can mean a worse list. The foundry's intent signal list replied twice as often as its named account searches and produced fewer buyer outcomes per person messaged.

Do speed and wording fix it?

Not much. On our own account the median reply time fell from about 19 minutes to about 2 over a year, and meetings per conversation did not rise. Speed is table stakes. Opener style moves replies: observations out replied questions by 10 to 15 points in the same month, at equal or better meeting rates. That is a reply rate gain, not a conversion lever. The list was the bigger lever (Sellinger case study).

What does conversion per reply show that the reply rate hides?

Take the foundry's two lists. From the published rates, the searches for named target manufacturers (449 people messaged, 27.6% replied) produced about 124 replies and 24 buyer outcomes, which is about 1 outcome for every 5 replies. The intent signal list (141 people messaged, 58.2% replied) produced about 82 replies and 6 buyer outcomes, about 1 for every 14 replies. These counts are our own arithmetic from the published rates, rounded. The list with the lower reply rate was the better list.

That is why we look at two denominators. Outcomes per person messaged tell you what the list is worth. Outcomes per reply tell you how much of the conversation is real. A campaign can look healthy on the first and weak on the second, or the other way round.

What should you ask a vendor about results?

When a vendor shows you a reply rate, ask these five questions:

1. What was the denominator: people messaged, connections accepted, or messages sent?
2. How many of the replies were real, and how was "real" decided?
3. How many meetings were held, not booked, and what proof is there?
4. What did the highest and the lowest performing audience produce with the same setup?
5. Who read the threads behind the numbers?

If the answer to the last question is "nobody", the numbers are labels, and labels can overstate.

What should you measure instead?

Fix the definitions before launch and keep them.

• [ ] People messaged (the denominator), not just replies.
• [ ] Real replies: a question, interest or a reasoned objection, counted apart from "thanks" and reactions.
• [ ] Meetings booked and meetings held, kept apart. See Count meetings honestly.
• [ ] Everything per audience and per offer, not per campaign. Tag the list source on every lead.
• [ ] Group size. Our comparisons use groups from 141 to 4,312 people. With a few hundred in a group, a difference of one conversion is noise.
• [ ] Read the threads. CRM stage labels can over and under count outcomes. Several of the studies above were rebuilt by reading the conversations, not by trusting the labels.