Conversion rate

To convert is to move someone from one state to another: from stranger to contact, from contact to meeting, from meeting to customer. The conversion rate measures the share of people who take that step. On its own the phrase means nothing. You always have to say which stage to which stage.

In e-commerce people usually mean a website's conversion rate: the share of visitors who buy. In B2B sales there is a chain of them, one per step of the funnel, and their product gives you the final result.

Splitting it this way prevents a common mix-up. Two teams announcing '5% conversion' may be talking about unrelated things: one about visitors who became leads, the other about meetings that became customers. Before comparing two figures, always ask for the numerator and the denominator.

The basic formula: conversion rate = number of people who reached stage B, divided by the number who started from stage A, times 100. If 200 prospects received your first message and 60 replied, your message to reply conversion rate is 30%.

Two precautions change everything. First, compare cohorts: today's replies partly come from last week's messages. Take the messages sent during a closed period and count their replies, even the late ones. Second, wait long enough. In MimikFlow's 2026 LinkedIn Prospecting Observatory (published in French), only 48.6% of replies arrived within 24 hours of the first message, and 84.8% within the week.

The cumulative rate multiplies the intermediate ones. For example, with 30% acceptance, 35% replies among people contacted and 15% meetings among those who reply, you get about 1.6 meetings per 100 invitations. The same math shows why a few points gained on one step ripple through everything below it.

Four rates describe most of it: invitation acceptance, reply to the first message, reply to meeting and meeting attendance. Each one points to a different lever: targeting, the message, the conversation, the meeting follow-up.

On the cold cohort of the 2026 Observatory (50,930 invitations sent at least 30 days earlier), acceptance reached 30.2%, replies 9.7% and observed meetings 1.6%, all relative to invitations. Relative to the people actually contacted, the reply rate of strangers found through search rises to 35.6%. Same population, two denominators, two very different numbers.

The final rate varies a lot with the target. In the study, consultants and freelancers who were contacted replied 57.1% of the time, directors and department heads 30.9%. A rate is only good or bad compared with your own market.

MimikFlow tracks these steps for every prospect, from invitation to a meeting booked in your calendar. A meeting only counts once it is actually booked, never when a link is merely sent, so your rates stay cautious rather than flattering.

Start with the step that is weakest against your benchmark, not the one that feels easiest to change. Low acceptance is fixed through targeting and your profile. A low reply rate is fixed through the first message: in the 2026 Observatory, first messages under 200 characters got a 50.5% reply rate, against 19.6% above 700 characters.

For the reply to meeting step, reread your conversations. Proposing a call too early scares people off, too late lets interest fade. Speed matters too: 22.0% of replies arrive within the hour, and a reply handled the next day has lost its momentum.

Change one variable at a time and let it run long enough to get readable volumes. On 30 messages, a difference of three replies teaches you nothing.

The composition effect is the most common trap. The 2026 Observatory gives an example: across the whole dataset, messages without a question seemed to convert better. In reality they were sent more often to warm prospects. Among strangers only, the gap almost disappeared: 34.7% without a question, 35.7% with one.

Another mistake is improving one rate at the expense of the next. Widening the target raises acceptances and can lower the meeting rate. The rate that really matters is the one linking your effort to revenue.

Finally, a meeting rate measured by a tool is often a floor. Meetings arranged by phone or email escape automatic tracking, and it is better to know that before concluding a campaign does not work.