Between 27 July and 14 September 2026 we sent 2,338 cold connection requests from 23 LinkedIn profiles. Some of those profiles belong to the person who owns the company being sold. Some belong to salaried people inside those companies. Some are rented, which is to say they belong to a real person who agreed to lend their account and has nothing riding on the outcome.
Same targeting, same pacing, same copy rules, same queue. Founder-led outbound usually arrives as one whole recommendation, and running all three kinds of profile through one machine lets us price the pieces separately. Below is what changed when the name on the profile changed.
Three kinds of profile, one machine
To make the comparison honest, we counted only invitations older than seven days, because acceptance arrives fast and a fresh invitation looks like a refusal. That leaves 693 matured invitations across four founder or CEO profiles, 820 across nine employee profiles, and 416 across six rented profiles.
One caveat before the numbers. These groups write to different audiences, because a founder tends to sell into a different room than the employee two floors below them. Treat every figure here as a measurement of our practice rather than a law of the platform.
The cold invitation lands about the same whoever signs it
This is the part that surprised us, so it goes first.
Acceptance on those matured invitations came in at 24.4% on employee profiles, 20.3% on founder profiles, and 18.0% on rented ones. Reply rate measured against everyone invited was 3.5% on employee profiles, 3.5% on founder profiles, and 1.9% on rented ones.
The first step landed the same either way. A stranger deciding whether to accept a connection request is reading a headline, a photo and maybe a shared connection, and at that moment the word "founder" sits among the ordinary signals. The companies we sell into have plenty of founders in their own inbox already.
That is good news for anyone building a team: your first sales hire can open doors at the same rate you can, as long as their profile is dressed as an operator and the targeting is the same. The founder's advantage sits further down the thread, and every number below measures it.
Speed is the first real difference
Once a person decides to answer, the founder's profile gets answered sooner. Median time from invitation to first reply was 1.2 days on founder profiles, 3.2 days on employee profiles, and 4.0 days on rented ones, across 28, 33 and 8 replies respectively.
A day and a half of median latency is worth real money in a pipeline where the follow-up cadence is measured in days. It also compounds: the faster the first reply arrives, the more likely the second message lands while the original context is still in the person's head.
Our reading is that a message from the person whose name is on the company reads as a decision someone made, and it gets answered at the priority of a decision. A message from a role reads as a process, and processes wait.
Half the founder's answers come from people we never wrote to
This number changed how we plan campaigns.
Across the whole period the founder profiles collected 63 replies. Only 28 of those came from people we had invited. The other 35 came from people who arrived on their own: they connected, they wrote first, they came back months after a conversation that went nowhere. On employee profiles the same split was 33 invited and 11 arriving on their own. On rented profiles it was 8 invited and zero arriving on their own, against 416 matured invitations.
Six rented profiles, 416 matured invitations, zero people who came looking. A rented profile can send. It can accept. It can hold a conversation. Inbound stays with the person: somebody arrives at a profile because of who owns it and what they are publicly known for, and on a rented account that belongs to the renter's own career.
So founder-led outbound has two halves. The founder's profile runs a cold campaign at roughly ordinary rates, and alongside it runs a second channel that costs nothing per message and that only the profile's owner can operate. Budget for the first, build for the second.
A founder who posts turns an audience into a list
The clearest version of that second channel is sourcing from engagement.
When a person posts publicly, the people who react and comment become a list, and it is a list with an event attached: this person read this specific thing on this specific day. We have collected 1,322 such leads. Because that requires someone posting under their own name, 1,289 of them sit against a single profile.
Most of that pool is still sitting there. Only 85 have reached an invitation, because the fit check is tight and 124 were excluded outright. Of those 85 invitations, 29 were accepted (34.1%) and 11 produced a reply (12.9%).
Put that against our general directory-sourced line over the same period: 1,729 invitations, 363 accepted (21.0%), 43 replies (2.5%). The engagement-sourced line replies at roughly five times the rate. Eighty five invitations is a small sample and we are stating it as a small sample, but the direction has held for two months, and three of the five calls booked in our base came from that pool.
The mechanism is unglamorous. Someone who reacted to a post two days ago has a real, checkable, recent fact attached to them, and that fact is about a topic they chose. Personalisation stops being research and becomes bookkeeping. This is also why the channel belongs to whoever is willing to post: the raw material is created by the act of publishing, and it decays within about a week.
The conversation goes somewhere only the founder can take it
Reply counts flatter everyone equally, so we measured what happens after the reply.
Threads that received at least one inbound message ran to an average of 12.5 messages on founder profiles, 6.7 on employee profiles, and 4.8 on rented ones. Of 77 founder threads, 29 collected three or more inbound messages from the other side. On employee profiles that was 13 of 44.
Then we looked at subject matter. Across 71 people who replied to a founder profile, 17 raised commercial terms in their own words: price, budget, fee, retainer, rate. On employee profiles that was 5 of 37. And 18 of those 71 founder threads brought up a call, a meeting or a calendar, against 6 of 37 on employee profiles.
The founder's asset here is authority. When a person raises price on message two, a founder answers on message two. Everyone else takes the question away, asks, comes back, and the conversation cools in the gap. We have written the same rule into our own machine: the engine handles scheduling, factual questions and polite closes on its own, while price, terms and anything outside the published offer go to the person who can settle them. On a founder's profile that path is one person long.
The practical version: if a founder is going to lend their profile to a campaign, they should be reachable inside a working day for the threads that reach a commercial question. A founder profile that answers at employee latency gives up most of what it brought.
The one thing worth protecting
All of the above rests on one profile that money cannot replace, so it gets a policy of its own.
We split every change into two kinds. Mechanics is what the platform sees: whether a note rides along with an invitation, the daily volume, a new type of action, a new channel. Copy is what the human sees, and it reaches the platform as plain text. Mechanics get proven on a rented or disposable profile first and arrive at founder profiles already settled. Copy can be tested anywhere, including on the founder's own account, because the downside of a weak sentence is a weak sentence.
The reason is one line long. A throwaway persona profile we ran to test mechanics was closed by the platform on its ninth day, having sent eight invitations. The same restriction on the profile that owns your network and your inbound would cost the channel that the last three sections were about, and no amount of money buys it back.
In our own database a flag carries this instead of a habit: registering a mechanics experiment against a protected profile fails at the write, and a check goes red the moment that flag appears on an account after the fact.
What to run from Monday
Three things, in order of how quickly they pay.
Send from whoever is available, and stop paying a premium for the founder's signature on the cold step. At 24.4% against 20.3% acceptance, an employee's own profile opens doors at full rate. Spend the founder's scarce hours on the threads that have already answered.
Post under a real name, then work the reactions. That source arrives with a recent, checkable, self-selected fact attached, and it belongs to whoever's name is on the posts. Keep it inside a seven day window and put the same fit gate on it you put on everything else.
Give the commercial questions a one person path. Measure how long your threads wait when somebody asks about price. If the answer runs past a day, the founder's profile is working as a nameplate, and the depth numbers above stay out of reach.
If you want this running with the sourcing, pacing and reply handling already built, the service levels page shows where that sits, and MCP access connects your own model to the same stack so you can drive it yourself.