The weekly report a LinkedIn outreach campaign should produce

The four stages, and the only correct way to divide them

Four counts, four rates, and the denominators are where most reports go wrong.

• Accept rate = invites accepted ÷ invites sent.
• Reply rate = people who replied ÷ invites accepted. Not divided by invites sent.
• Meeting rate = meetings booked ÷ people who replied.
• End-to-end rate = meetings booked ÷ invites sent. Report it alongside the other three, never instead of them.

Worked through on your own numbers, the arithmetic is simple. If you send 200 invitations, 60 are accepted, 9 of those people reply and 3 book a meeting, your accept rate is 30%, your reply rate is 15%, your meeting rate is 33% and your end-to-end rate is 1.5%. Those figures are arithmetic on one hypothetical week, not targets — the point is the formula, and specifically that the reply rate was divided by 60 and not by 200.

The denominator is not pedantry. Reply rate over accepted tells you whether your first message works on people who already agreed to connect, which is a thing you can act on. Reply rate over invites sent blends two entirely different failures — bad targeting and bad copy — into one number that cannot be fixed because it cannot be diagnosed. A report that quotes only the blended figure is hiding which half is broken, whether or not anyone intended it to.

The report, field by field

One table, one page, same shape every week. If the shape changes, week-on-week comparison is gone.

The last two rows are what separate a report from a dashboard screenshot. A weekly report with no stated change is a status update, and status updates do not improve campaigns.

Cohort weeks, not calendar weeks

Attribute results to the week the invitation went out, not the week the acceptance arrived. Acceptances trickle in over days and sometimes weeks, so dividing this week's acceptances by this week's invites mixes two populations and produces a rate that is wrong in both directions — flattering when sending drops, punishing when it rises.

The practical consequence is that your two most recent weeks always look worse than they are, because they have not finished maturing. Mark them as provisional in the report rather than explaining it verbally every Monday. A provider who cannot explain why last week's accept rate went up after the meeting is reporting on calendar weeks and does not know it.

The same lag applies to replies and, more painfully, to meetings, which can be booked from a thread that started a month earlier. Keep the cohort attribution all the way down the funnel, and accept that the end-to-end number for any recent week is an underestimate.

Why this report contains no industry benchmarks

Because a published benchmark is drawn from a population you are almost certainly not in. Seniority of the people being contacted, geography, the sender's own profile, the size of the ask and the category all move these rates more than any copywriting decision does. Importing a number from an unrelated dataset makes a bad decision feel evidenced, which is worse than having no number at all.

Build your own baseline instead, and it takes four weeks. Hold the list type, the sequence and the sender constant. Record each stage weekly using cohort attribution. At the end, take the median week — not the average, which one freak week distorts — for each of the four rates. That set of four numbers is your baseline, and from then on every result is read against it rather than against someone else's slide.

When a vendor or an article quotes you a rate as an expectation, ask three questions: whose profile was sending, who was on the list, and what exactly was being asked for. If those three cannot be answered, the number is decoration. This is also the honest reason our outreach service reports against your own first month rather than against a headline figure — the first month is the only comparable dataset that exists.

The diagnostic ladder: weak stage to likely cause

Fix from the top. A rewritten first message is wasted effort if the list is wrong, because the people reading it were never going to be interested in any version of it.

One change per week, and one variable at a time. Two changes in the same week produce a report nobody can read, because you will never know which one moved the number. If the accept rate is the weak stage, the profile itself is usually the cheapest thing to fix first — profile optimisation changes the accept rate without touching the list, and the accept rate is upstream of everything else. If the bottom of the funnel is weak while the top is healthy, the problem is usually the offer rather than the channel, and that is a lead generation question rather than a messaging one.

Numbers to leave out, and one exception

Remove anything that moves without anyone deciding anything.

• Profile views. They correlate with activity, not with outcomes, and they go up when a campaign is going badly as well as well.
• Total connection count. It is a stock, not a flow, and it only ever increases.
• Impressions, reach, and any other borrowed advertising metric.
• Messages sent, as distinct from sequences completed. Sending more messages to the same person is not progress.
• Daily activity charts presented as results. They are operational data, not performance data.

The exception is safety. Invitations sent per day, the proportion of pending invitations withdrawn, and any warnings or verification prompts received all belong in the weekly report — but in a separate block clearly labelled as operational. A campaign whose accept rate is climbing while its daily volume is climbing faster is not improving, and only the operational block will tell you that. Free calculators and planning tools sit on our tools page if you want to sanity-check volumes before a campaign rather than after.

Using the report on a Monday

Fifteen minutes, three questions, one decision. Anything longer and the report becomes a performance rather than a control.

1. Which stage is weakest against our own baseline, not against last week? Week-on-week noise is large at these volumes; the baseline is the comparison that means something.
2. What did last week's single change do to that stage? Answer it before proposing a new one.
3. What is the one change we are making this week, and which stage do we expect it to move?

Write the answer to question three into the report itself, in the field reserved for it. Next Monday, that line is what you are grading. A campaign run this way improves slowly and provably; a campaign run on a dashboard with no stated changes improves by luck and nobody can say why.

Key takeaways

• Divide reply rate by accepted invitations, not by invitations sent, or you cannot tell targeting from copy.
• Attribute every stage to the week the invitation went out, and mark the two most recent weeks as provisional.
• Build a baseline from the median of your own first four weeks instead of importing an industry benchmark.
• Fix the funnel top down and change one variable per week, or the next report is unreadable.
• Every weekly report should end with one named change and the stage it is expected to move.

Frequently asked questions

What metrics should a LinkedIn outreach report include?

Four counts — invitations sent, accepted, replied and meetings booked — plus the four rates derived from them, the sequence and list segment in use, two or three verbatim replies, and the single change being made next week. A separate operational block should carry daily sending volume and any warnings received.

Is reply rate calculated on invitations sent or on connections accepted?

On connections accepted. Dividing by invitations sent blends a targeting problem and a copy problem into one number that cannot be diagnosed. Report the end-to-end figure — meetings divided by invitations sent — separately if you want a single headline number.

What is a good accept rate or reply rate on LinkedIn?

There is no honest universal answer, because seniority, geography, sender profile and the size of the ask move these rates more than copy does. Run the same list, sequence and sender for four weeks, take the median week for each stage, and use that as your baseline. Any figure quoted without those four variables attached is not comparable to yours.

How often should a managed outreach campaign be reported on?

Weekly, in the same format every week, with cohort attribution so results land in the week that caused them. Monthly reporting is too slow to catch a broken sequence, and daily reporting encourages reacting to noise at these volumes.