Linkedin AI Tagging Methodology
This article was written as an email for Linkedin automation clients when we released our new reporting dashboards. Email tools already have sentiment analysis included, but no Linkedin tools include that as a feature.
We had to create (essentially) a new automation process to have AI read the incoming responses and tag them - then push the data out to the client reporting dashboard.
Why go through the trouble?
Because it’s not enough to say “You’re getting replies. Therefore you’re getting leads.”
I wanted to see the quality and sentiment of the replies to know if we’re on the right track. It also provides my clients with a clearer accounting of who they need to follow up with and reach out.
A secondary benefit is that we could now notify the Linkedin clients when they got a positive reply, like we do with our Email clients, and therefore keep them focussed on the real leads.
Here's how we categorize responses:
- Our AI system reads and labels responses in these categories:
- INTERESTED: The person wants to connect or meet
- CURIOUS: The person has questions or wants more information
- WRONG PERSON: We reached out to someone who isn't the right contact
- UNCLEAR: We can't tell for sure what the person means
- NEGATIVE: The person says no to our message
Important things to know:
- Each response gets its own tag - we don't look at entire conversations
- Well-written messages help us get genuine interest from people
- Messages we draft get truly interested responses - not simply polite "nice to connect" replies
- Poorly written sequences can make the number of "interested" responses look better than they really are (ironically), because of the “nice to connect” or “glad to connect” message
- If you think your response numbers look off, let's talk about your campaign together
Thanks,
Franc Berrones
ClarityAdvisory