Cohort Analysis for LinkedIn Outreach: Tracking Reply Rates Over Time

LinkedIn outreach campaigns generate three numbers everyone watches: acceptance rate, reply rate, and booked meetings. But a single campaign average hides the most useful signal in your data: whether performance is getting better or worse over time, and exactly when prospects lose interest.

Cohort analysis solves this by grouping connection requests or message sequences by the week or month they were sent, then tracking each group forward through its lifecycle. A January cohort might show 28% acceptance after 30 days and 7.2% reply rate after 60 days. February's cohort, measured the same way, might show 26% and 8.1%. That two-point acceptance drop and one-point reply gain tells you something changed, and cohort tracking tells you when.

This guide walks through how to build cohort analysis for LinkedIn outreach, what metrics to track, and how to use the resulting data to improve targeting, sequencing, and timing.

What is cohort analysis for LinkedIn outreach?

A cohort is a group of connection requests or outreach sequences that share a common start date. Cohort analysis tracks how each group performs over time, measuring acceptance, reply, and meeting rates at fixed intervals after the send date.

In email marketing, cohort analysis is well established. Mailchimp describes it as dividing users into groups with shared characteristics and charting data for each group to determine the success of business functions. The same method applies to LinkedIn: group requests by send week, then measure what percentage of each group accepted, replied, and booked within 7 days, 14 days, 30 days, and 60 days.

The key difference between cohort analysis and campaign-level reporting is that cohorts isolate the effect of time. A campaign average mixes requests sent yesterday with requests sent two months ago. Cohort analysis separates them, so you can see whether last week's batch is outperforming or underperforming the one from January.

Why track LinkedIn performance by cohort instead of by campaign?

Campaign-level reporting answers the question: how is this campaign doing right now? Cohort analysis answers: is this campaign getting better or worse, and when do prospects stop engaging?

A campaign that shows 26% acceptance today might look weak. But if last month's cohort is sitting at 24% and this month's is at 28%, the trend is upward and the campaign is improving. Without cohort segmentation, that improvement is invisible.

Cohort analysis also reveals maturity effects. Reachium's 2026 benchmark study tracked 204,847 connection requests. When they measured the full set including recent requests, acceptance sat at 26.34%. When they restricted the analysis to requests sent at least 30 days earlier, acceptance rose to 27.11%. The difference is recency bias: requests sent last week haven't had time to be accepted yet, and counting them as failures understates performance.

Tracking cohorts by send date isolates this effect. Each cohort is measured at the same intervals, so newer cohorts are compared to older ones at the same stage of maturity, not at different points in their lifecycle.

Which metrics should you track in a LinkedIn cohort analysis?

The three foundational metrics for LinkedIn outreach cohorts are acceptance rate, reply rate, and meeting rate. Each has a clear denominator, and cohort analysis tracks all three over time.

Acceptance rate is accepted connection requests divided by requests sent. This measures whether the request itself landed. Across Reachium's 180,155 matured requests, acceptance was 27.1%. Expandi's 13.2 million request dataset from May 2025 to April 2026 reported 28.5%. Both figures sit in the 26% to 29% range that multiple studies have confirmed as the platform baseline.

Reply rate measures whether an accepted connection produced at least one inbound message, divided by accepted connections. Reachium reported 27.6% of accepted connections replied. That same figure, divided by requests sent instead of acceptances, drops to 7.5%. The denominator matters: 27.6% and 7.5% describe the same dataset, and both are correct. One divides by accepted connections, the other by requests sent.

Meeting rate is connections with a booked meeting divided by accepted connections. Reachium reported 1.99% across their matured cohort. Belkins tracked 14,077 LinkedIn outreach records and reported 1.3% of connected prospects booked a meeting. Both studies used platform-tracked bookings, so the true rate is higher by an unknown amount because conversations that moved to email or calendar links pasted manually were not counted.

How to build a cohort analysis spreadsheet for LinkedIn campaigns

Building a cohort analysis spreadsheet starts with exporting your outreach data. You need four columns: contact ID, request sent date, acceptance date, and first reply date. If your platform tracks meeting bookings, add that as a fifth column.

Step one: assign each contact to a cohort based on the week or month the request was sent. A request sent January 8, 2025 belongs to the January 2025 cohort. A request sent February 3, 2025 belongs to February 2025.

Step two: calculate the number of days between the request sent date and each downstream event. For accepted connections, subtract the sent date from the acceptance date. For replies, subtract the sent date from the first reply date. This gives you the maturity interval for each contact.

Step three: count how many contacts in each cohort reached each milestone within fixed intervals. For example, count how many January cohort requests were accepted within 7 days, 14 days, 30 days, and 60 days. Repeat for replies and meetings.

Step four: calculate the rate for each interval by dividing the count by the cohort size. If 200 requests were sent in January and 54 were accepted within 30 days, the 30-day acceptance rate for the January cohort is 27%.

Step five: build a pivot table or chart with cohorts on one axis and time intervals on the other. Each cell shows the rate for that cohort at that interval. This structure makes it easy to compare performance across cohorts and spot trends.

The resulting table might show January at 27% acceptance after 30 days, February at 29%, and March at 25%. That two-point variance is the signal cohort analysis is designed to surface.

How long should you wait before measuring a cohort?

Reachium's methodology used a 30-day maturity cutoff for headline figures. Requests sent at least 30 days before the data pull were counted; requests sent more recently were excluded from the matured cohort. The reason is straightforward: a request sent last week has not had time to be accepted yet, and counting it as a failure understates recent performance.

The study also reported figures for the full dataset including recent requests. Acceptance across all 204,847 requests was 26.34%; acceptance across the 180,155 matured requests was 27.11%. The 0.77-point difference is the maturity effect made visible.

For cohort analysis, measure each cohort at the same intervals. If you're tracking 7-day, 14-day, 30-day, and 60-day windows, do not report the 30-day figure for a cohort that is only 20 days old. Wait until the cohort reaches 30 days, then measure. This keeps comparisons valid across cohorts.

Belkins' study tracked acceptance times and found that median time to acceptance was 25.4 hours. 46.7% of accepted requests were accepted within a day, and 92% within 30 days. This confirms that a 30-day window captures the vast majority of acceptances that will ever occur.

What patterns should you look for in LinkedIn cohort data?

The most valuable patterns in cohort data are row-over-row changes and drop-off timing. Row-over-row changes show whether newer cohorts are performing better or worse than older ones. If February's 30-day reply rate is 8.2% and January's was 7.1%, the trend is upward. If March drops to 6.5%, something changed and the trend reversed.

Drop-off timing shows when prospects stop engaging. If 28% of a cohort accepts within 7 days, 29% within 14 days, and 30% within 30 days, most of the acceptance happened early and the rate flattened. If reply rate shows 5% within 7 days, 7% within 30 days, and 7.2% within 60 days, replies trickle in slowly over two months.

Sharp early drop-offs often indicate a problem with the connection request itself or the first message. If a cohort shows 10% reply rate within 7 days and only 11% within 60 days, most prospects who will ever reply do so immediately. Those who don't reply in the first week are unlikely to reply later, which suggests the follow-up sequence is not working.

Conversely, if a cohort shows 4% reply rate within 7 days and 9% within 60 days, half of all replies arrive after the first week. That pattern suggests follow-ups are effective and the sequence is long enough to allow for delayed responses.

How to use cohort analysis to improve LinkedIn outreach campaigns

Cohort analysis turns data into decisions by isolating the effect of changes. If you adjust your connection note in March and March's cohort outperforms February's by three points on reply rate, the new note worked. If it underperforms, revert or test a different approach.

The same method applies to targeting. If you shift from targeting Directors to VPs in April and April's acceptance rate drops five points, the new audience is less receptive. If reply rate climbs, the audience is harder to connect with but more engaged once connected. Cohort analysis separates these effects.

Timing analysis also benefits from cohort tracking. Reachium measured reply timestamps and found that 51.5% of first replies landed between 09:00 and 18:00 UTC. Weekend reply volume ran 42% below weekdays. If your cohorts show different patterns, adjust send schedules accordingly.

Finally, cohort analysis reveals the return on follow-ups. Reachium tracked 43,084 accepted connections and found that 14.7% of replies arrived after a follow-up. Follow-ups converted 13.9% of prospects who had gone silent after the first message. If your cohorts show lower conversion from follow-ups, the follow-up copy or timing needs work.

What tools and platforms support LinkedIn cohort analysis?

Most LinkedIn outreach platforms export data that includes request sent date, acceptance date, and reply date. That raw data is enough to build cohort analysis in a spreadsheet. Tools like Google Sheets and Excel support pivot tables, which are the standard way to structure cohort data.

For teams that want automated cohort reporting, platforms like Reachium, Expandi, and others with built-in analytics can track cohorts natively. The advantage is that the data is pre-structured and the platform handles the interval math.

Data integration tools like Fivetran centralize data from multiple sources, which is useful if your outreach runs across more than one platform or if you want to combine LinkedIn data with CRM data. Once centralized, you can build cohort reports in a business intelligence tool like Power BI, Tableau, or Looker.

Common mistakes when running LinkedIn cohort analysis

The most common mistake is mixing denominators. A reply rate of 27.6% divided by accepted connections is not comparable to a reply rate of 7.5% divided by requests sent. Both figures can describe the same dataset, but they answer different questions. Always state what you're dividing by.

Another mistake is comparing cohorts at different maturity stages. If you compare January's 60-day reply rate to February's 30-day reply rate, the difference might reflect time rather than performance. Measure every cohort at the same intervals.

Ignoring sample size is also a problem. A cohort with 50 requests might show 40% acceptance, but that's 20 acceptances. A cohort with 1,000 requests at 28% acceptance is 280 acceptances. The smaller cohort's higher rate might be noise. Treat small cohorts as directional, not definitive.

Finally, cohort analysis is backward-looking. It tells you what happened, not what caused it. If March's cohort underperforms, the analysis shows the drop but not the reason. You still need to review what changed in targeting, messaging, or timing during that period.