Cold email or LinkedIn DM, when you have ten hours a week

You have ten hours a week for outbound and you cannot run both channels properly. So you go looking for the number that settles it, and you find plenty of numbers. They do not agree with each other.

Nobody has measured this. Every side-by-side reply-rate comparison of cold email against LinkedIn DMs is published by a vendor selling one of the two, none of them disclose a denominator, and the figures contradict each other by up to five times. What can be established, at source, is which ceiling you hit first. LinkedIn meters your first touches. Email meters your reputation. At ten hours a week those two constraints behave completely differently, and for most people at this volume that makes email the channel to spend the hours on. The exception is specific and worth knowing, and it is at the end.

What the reply-rate numbers actually are

Here is every comparative figure found in a full pass across the pages ranking for this question, with what each one discloses about how it was produced.

Read the email column on its own. One vendor says one to two percent and another says 5.1 percent, and both present their figure as the average. The LinkedIn column has the same problem in the other direction.

Two datasets in this field do disclose a real sample size, and neither of them compares the channels. Saleshandy reports a 3.7 percent overall reply rate across 53.1 million emails in 60,000 sequences between January and June 2026, all of it from its own platform's users. Belkins, working with Expandi, reports a 7.2 percent average across 14,077 tracked contacts and 15.1 million touchpoints through 2025, all of it LinkedIn. Put those two next to each other and you have not compared anything: different populations, different years, different definitions of a reply, and each one measuring the channel its author sells software for.

One of the figures above does not survive a source check at all. LinkedCamp attributes its 8 to 15 percent to a "LinkedIn State of Sales Report 2024." The most recent State of Sales report published under LinkedIn's own domain is the 2022 edition. No 2024 edition exists there.

What LinkedIn meters, and the one number it publishes

LinkedIn charges you on first touches, and it does so twice.

The first meter is the invitation. LinkedIn will restrict your account from sending connection requests, and its help pages name two triggers: sending a high volume in a short period, and having a high share of invitations ignored or marked as spam (LinkedIn, a551012). A restriction typically lasts one week. What LinkedIn never states anywhere is the weekly number, which is covered in more detail in how LinkedIn's connection limits actually work.

The second meter is the commercial use limit, which governs searching rather than messaging. Filtered searches and People Also Viewed spend it. Searching by name and browsing your own first-degree connections do not. It resets at midnight Pacific on the first of each month, LinkedIn does not display your counter, and LinkedIn states it cannot lift the limit on request. The full breakdown of what does and does not count is in what counts against your LinkedIn search limit.

So before you write a word, LinkedIn has already capped how many new people you can approach this week and how many you can find this month, and it publishes neither figure.

There is exactly one absolute reply-rate percentage anywhere on LinkedIn's own domain. Recruiters using Recruiter or Recruiter Lite must maintain an InMail response rate of at least 13 percent on 100 or more messages in a 14-day period. Read what that is before you use it. It is a compliance floor that keeps a recruiter out of a sending suspension, not an average of anything. It applies to a hiring product where recruiters message candidates about jobs, not to sales outbound. And its definition of a response includes an invitation acceptance or a "Not Interested" click, which is a wider net than the word reply usually implies.

LinkedIn's talent blog does publish real reply-rate research, drawn from tens of millions of InMails sent between May 2021 and April 2022. Every figure in it is relative. Shorter InMails do 22 percent better than average, personalised ones about 15 percent better than bulk. The average those percentages are calculated against is never given. That is the same pattern as the invitation cap and the search cap: LinkedIn publishes the shape of the thing and withholds the number.

What email meters, and why the published cap is the wrong number

Email works the other way around. Nobody limits how many people you can approach. What gets limited is your ability to reach anyone at all, and it is governed by complaints rather than volume.

Google's published thresholds are specific. You become a bulk sender at "close to 5,000 messages or more to personal Gmail accounts within a 24-hour period", and once you have that status it does not expire. Google asks every sender to keep their reported spam rate below 0.10 percent and to never reach 0.30 percent. A bulk sender at or above 0.30 percent is ineligible for mitigation until the rate has stayed under it for seven consecutive days (Google, Email sender guidelines). Authentication requirements took effect on 1 February 2024.

Microsoft's ceiling is higher and just as public: 10,000 recipients per mailbox per rolling 24 hours, and 30 messages per minute, on a page whose own revision stamp reads 13 May 2026.

Now do the arithmetic on the 0.30 percent. Three complaints in a thousand sends puts you at the ceiling. At forty emails a day, one complaint a week is roughly 0.36 percent. The published cap of thousands per day is a number you will never approach, and the number that decides your fate is one you can trip in your first week. The practical sending schedule that follows from this is in how many cold emails a day is safe on one domain.

Which ceiling binds first at ten hours a week

Ten hours a week of genuine manual outbound is roughly 40 to 80 personalised first touches. Take that number to each channel.

On LinkedIn that is 40 to 80 invitations a week, and nobody outside LinkedIn can tell you how close to the ceiling that puts you. LinkedIn publishes no weekly number, shows you no counter, and describes a restriction as something triggered by volume in a short period and by the share of your invitations that go ignored. You are working toward an unmarked line, and the first reliable signal that you crossed it is a week of not being able to send. Add the monthly search allowance on top, since finding those 40 to 80 people takes filtered searches, and the same activity is metered twice.

On email the same 40 to 80 a week works out at eight to sixteen a day. Google's bulk-sender threshold is close to 5,000 a day. You are two orders of magnitude below it, and further still below Microsoft's mailbox ceiling. Volume never binds, and it would not bind if you tripled your hours. What binds is the complaint rate, and the complaint rate is a function of who you picked rather than how many you sent. Ten well-chosen sends a day from a warmed domain will sit under 0.10 percent indefinitely. The same ten sent to people who have no reason to hear from you can put you over 0.30 percent within weeks, because the ceiling is a rate rather than a count.

So at ten hours a week the choice is email, and the reason is not that it performs better. Nobody knows whether it performs better. The reason is that LinkedIn's constraint is fixed before you start and email's constraint responds to how well you choose, which means the hours you have are worth more on the email side.

That is the actual difference between the two channels at this volume, and it is a difference in what you can control. LinkedIn's constraint is external and fixed. Email's constraint is internal and responds to your list quality. If your targeting is good, email gives you room and LinkedIn does not. If your targeting is bad, email punishes you for it and LinkedIn simply hides the ceiling while you hit it.

What changes if you have a network rather than a list

One case flips this. If the people you want to reach are already your first-degree connections, LinkedIn stops metering you: messaging existing connections costs no invitation, and browsing your own connections does not spend the commercial use limit. A founder with 3,000 relevant first-degree connections and ten hours a week is in a different position from one starting cold, and for them LinkedIn is the cheaper channel by a wide margin.

That case is narrower than it looks, because it depends on whether those connections are still in the roles you connected with them in. The reply rate on a message to someone who changed jobs eighteen months ago is not the reply rate on a warm contact.

For everyone else, the choice at ten hours a week comes down to this. Email does not cap your volume and does cap your carelessness. LinkedIn caps your volume before you start. Pick email, spend the hours you saved on the list, and treat LinkedIn as the place you go after somebody has already replied. What actually moves the number in either channel is who is on the list, which is covered in who actually replies to cold email.