LinkedIn DM reply rates, measured against email on the same accounts

What is a LinkedIn DM reply rate?

A LinkedIn DM reply rate is replies of any kind divided by direct messages sent on LinkedIn, over a stated window, on a stated segment. That is the whole definition, and almost every disagreement about the number is really a disagreement about the denominator.

Three decisions sit underneath it and nobody states them. Whether automatic away messages count. Whether a prospect who got three messages and answered once contributes one reply against one send or one against three. And whether the denominator is messages sent or prospects messaged at all.

That is not a rounding difference: a sequence sending three messages per prospect reports a per prospect rate roughly three times higher than its per message rate on identical activity. Two teams can run the same campaign, report honestly, and land on numbers that look like different channels.

Our convention: replies of any kind, including negative and out of office, divided by messages sent. The least flattering option, which is why we use it.

What reply rate should you expect from cold LinkedIn DMs?

The figure we publish, with its condition attached, is roughly 9%, from one white label programme running across a set of advisor workspaces, and we are not turning that into a range. A range built from one data point plus judgement is a guess wearing a decimal point, and most ranges circulating here come from companies that sell the sending tool.

More useful than a benchmark is knowing which inputs move the number. Three do.

• Who the sender is relative to the recipient. The same message from a founder and from a title that reads as a sales function produce different reply rates. Largest single input, and it is a staffing decision, not a copywriting one.
• How the connection happened. Someone who accepted your request has given you a small signal. Someone reached by InMail has given you nothing, and a warm opener to them reads as a false claim of familiarity.
• How specific the reason for contact is. Segment level specificity is enough. Individual level specificity that was obviously generated is worse than none.

What did we measure, and under what conditions?

We measured three rates on one client programme where LinkedIn and email ran from the same senders, against the same population, in the same window. That last condition is the point.

The engagement was a white label programme running across a set of advisor workspaces. These figures come from the control cell, recorded live, not reconstructed afterwards. The senders were recognisable to that audience as peers, which flatters the acceptance rate and presumably the reply rate too.

Two things about this dataset are still open on our side and we would rather name them than hide them. The totals, meaning DMs sent, emails sent and replies on each channel, plus the exact date window, have not been pulled from the campaign record. And we have not confirmed whether the roughly 9% is replies per message sent or replies per accepted prospect. That second one changes the denominator and the calculator depends on it, so the calculator uses the more conservative of the two readings and prints a footnote saying so.

Those gaps stay visible until somebody closes them. A page criticising everyone else for publishing rates without denominators does not get to skip its own.

Why is a same account comparison worth more than a cross study one?

A same account comparison holds the sender, the list, the offer and the window constant, which leaves the channel as the only thing that varied. Every other comparison of these two channels leaves at least four things varying at once.

Put your LinkedIn reply rate next to an email reply rate from somebody else's study, measured on a different audience, in a different year, with a different offer and a different denominator. That is not a comparison. It is two unrelated numbers side by side, and the ratio between them means nothing.

Almost nobody publishes one, because almost nobody runs both channels against a single list at volume. It needs the same senders, one list, one window, and enough infrastructure that neither channel is throttled by something unrelated to it.

What does this reply rate not measure?

It does not measure interest, quality or revenue. It measures how often a message got an answer, and "not interested, please remove me" is an answer. It is not a booking rate either, and the gap between a reply and a booked meeting is where most of the real work sits.

It is also not a positive reply rate. A positive reply ratio is positive replies divided by total replies. A positive reply rate is positive replies divided by sends. Putting a healthy looking ratio next to a low per send rate implies an advantage the data does not support, and it is the most common measurement error in this industry, including in our own older marketing. The full argument belongs to our sibling property AllboundPros, which owns funnel conversion definitions for the group.

What is Channel Yield, and why does a higher reply rate not mean a better channel?

Channel Yield is per message reply rate multiplied by the monthly volume one prepared account can achieve on that channel. Every public comparison of these two channels drops the second term, which is where LinkedIn loses.

LinkedIn wins the first row by a wide margin and loses every one after it. A six times higher per message response is not a six times better channel when the capacity term moves the other way by more than six, and on most programmes it does.

Which is why the question of whether to move budget from email to LinkedIn has no answer at this level. The two are not substitutes, and sequencing them is a different discipline that belongs to our sibling property MultichannelPros. This page publishes the measured rates and how to read them against capacity, not which channel to run.

How do you turn a reply rate into a monthly number for your own accounts?

Four multiplications, in this order, and the first one is not yours to choose.

Which of the four funnel numbers actually responds to a rewrite?

Here is how that plays out live. We run two motions, using the group's own names. WideNET is high volume systematic angle testing across the full addressable market. Spearhead is signal triggered work on the hottest slice. Neither raises the ceiling, so a WideNET test on LinkedIn runs on smaller cells and takes longer to reach a decision than the same test on email.

That matters because of how we kill things. The group runs fixed kill and scale thresholds on positive replies per send, published with their denominator by our sibling property AllboundPros. Applying those per send thresholds to LinkedIn would kill campaigns on noise, because LinkedIn cannot produce the send counts that make a sub one percent threshold meaningful in a reasonable window. That is a LinkedIn specific adjustment and it is worth stating plainly: LinkedIn cells get judged on acceptance and reply first, and on positives only once volume justifies it.

An internal monitoring desk watches reply rate by sequence daily, and the client approves messaging and lead lists before anything sends. That gate is why a bad angle costs a week, not a month.

Where does this benchmark fail, and who should not use it?

This benchmark fails as soon as you change the sender or the audience, which is most of the time. The honest list of its limits.

It is one dataset. One client, one vertical, one control cell, one window. We would rather publish one number with its full condition than a range with none, but one is one. If a multi client LinkedIn DM reply dataset with consistent denominators can be assembled across programmes, that becomes this site's second original benchmark, and until it exists a range would be judgement dressed as data.

The sender was recognisable. That raised the acceptance rate and almost certainly the reply rate too. If your senders are unknown to your market, plan below this figure, not at it.

It says nothing about total volume. LinkedIn did not send comparable volume to email on that programme and could not have. Anyone using these numbers to argue LinkedIn produced more replies in total has removed the constraint that makes the comparison meaningful.

It is useless if your buyers are not on LinkedIn. Field operations, skilled trades, plant level manufacturing, most public sector procurement. If the person who signs does not open LinkedIn in a normal month, this is a fact about a channel they are not in.