# How the numbers on this blog are measured

Where the data comes from, what each count means, when a batch is old enough to be counted, how recency is read, and what is left out. The rules behind every table here.

Every table on this blog comes from the same place and follows the same rules. This page is the reference the experiments point to, so each post can stay short. When a rule changes, it changes here first, and the change is dated at the bottom.

## Where the data comes from

The numbers come from Scout running on my own LinkedIn account, prospecting for real: real people, real connection requests, real replies, for products I actually sell. Nothing comes from other users of Scout. The app keeps its data on the user's computer and sends nothing to us, so there is no aggregate to draw from, and there will not be one. What you read here is one account, written down honestly.

## What each count means

- **Visited**: a profile that was opened and read before any decision was made.
- **Connection request**: a request sent with a short note written from that profile. A visited profile that did not fit got no request.
- **Accepted**: the person accepted the request. Counted when the acceptance appeared, not when it was noticed.
- **Replied**: the person wrote back at least once after accepting, in any direction.
- **Interested**: a reply that, read in the conversation, was marked as interest. It is a judgement, made per conversation, and it is the loosest count on this blog.

Rates are always given with their base: "59% of accepted" and "25% of requests" are different numbers for the same people.

## When a batch is old enough to count

Acceptances keep arriving for about a week after a request, and replies for about two. A batch enters a table only after that time has passed: one week for acceptances, two for replies and interest. Numbers published earlier than that are marked with the date they were counted.

## How recency is read

"Posted recently" is what the activity tab of the profile shows at the moment the profile is read: the date of the most recent post, in the bands LinkedIn displays (days, weeks, months). The cuts used here are 7 and 30 days. A profile with no visible activity counts as "no post", not as "unknown".

## Groups, rounding and sample size

A group gets its own row only with at least ten people in it; smaller groups go into "other". Ratios like "2x" are rounded to the nearest whole number and the exact rates are in the table next to them. Medians are preferred to averages wherever a few slow cases would distort the mean.

## What is left out

- Profiles that were visited but never got a request are not in any comparison of acceptance or reply.
- People who asked not to be contacted are counted once, as a reply, and never again.
- Nothing about who the people are: no names, no companies, no message text. Only aggregates.

## How pages are updated

Experiments are living pages. When new weeks of data come in, the tables and the dated sentences are updated in place, at the same address, and the change is listed in the changelog at the end of the post, with the date. The update date is also what search engines receive. Nothing is republished under a new address.

## Changelog

- 2026-10-08: first version.

---

Canonical: https://www.scoutwith.me/en/method
Published: 2026-10-08