Outreach produces a lot of numbers. Five of them tell you what to change. This guide defines each one, shows how to work it out and says what a weak result usually means.
The five numbers
| Metric | How to calculate it | If it is low, look at |
|---|---|---|
| Acceptance rate | Accepted requests divided by requests sent | Who you are targeting, and what your profile says about you |
| Reply rate | Contacts who replied divided by contacts who accepted | The first message: is it specific, short and easy to answer? |
| Positive reply rate | Interested replies divided by all replies | Who you are targeting, and what you are offering |
| Meetings per 100 contacts | Meetings booked divided by contacts reached, times 100 | The step from reply to meeting: how fast you answer, and how you ask |
| Restriction events | A count of warnings or restrictions on the account | Volume and pace. Any event is a signal to slow down |
A worked example
These numbers are made up to show the arithmetic, not to suggest what is normal. Suppose you send 200 requests, 80 are accepted, 20 of those people reply, 8 of the replies are interested and 3 become meetings.
- Acceptance rate: 80 / 200 = 40 percent.
- Reply rate: 20 / 80 = 25 percent.
- Positive reply rate: 8 / 20 = 40 percent.
- Meetings per 100 contacts: 3 / 80 x 100 = 3.75, counting the 80 people reached.
Always say what you divided by. A reply rate against requests sent and a reply rate against requests accepted are different numbers, and mixing them is how teams fool themselves.
Change one thing at a time
If you change the audience, the message and the timing together, you will not know which one moved the result. Change one, run it long enough to see a difference, then change the next. If your tool supports it, test two versions of a message against each other.
What to ignore
- Total requests sent. Volume without results is a cost, and more volume raises the account risk.
- Profile views. They show activity, not interest.
- Other companies' reply rates, for the reasons above.
For how volume relates to account risk, read Is LinkedIn automation safe?
Questions
- What is a good reply rate?
- There is no reliable universal figure. Track your own reply rate over time and test one change at a time.
- Should I count automatic replies?
- No. Count only replies from the person, and say in your notes how you counted so results stay comparable.