How would you test the LinkedIn feature "People you may know" when you don't have any data to base your decision off of?

Qn clarification - People you may know is a feature by Linkedin to enable making connections. It has a subset of how the algorithm has picked people - eg school alumni you may know, people in Mumbai you may know, people in Internet industry you know, people who follow who you follow, etc. The same can be accessed under "Network" in menubar after the "Invites"/connection rrquests section. This feature is also displayed when user is joining Linkedin for the first time based on hi/her email address/phonebook contacts. Is this correct? Any other use case of this feature I should be aware of?

Business goal of this feature - Enables higher connections in professional circles which the user would have missed out on, thereby enabling more traction on the platform and higher engagement. User goal - Helps user make connections for networking / following professional profiles / seeking or making referrals / seeking or providing mentorship / sharing content etc / staying connected on Linkedin.

User journey: Three alternate journeys for users to interact with this feature:
1) Clicks on Network menu -> Visits the people you may know section post invites section
2) Signs up and at final stage is suggested list of people you may know with options to select some/all/skip altogether
3) While scrolling on newsfeed, there is suggestions of "people you may know" to connect with

Metrics by customer journey:
1) Discoverability of the feature - # users who have opened Network menu and scrolled to People you may know section , # users who made atleast one connection through this feature, %connections formed through search/homepage/posts (basically alternate journey) which were part of suggestions
2) Active use of the feature - avg #connections / user made through the feature (quantity), % users who selected some/all connections while signing up instead of skipping
3) Engagement - # interactions( Messages exchanged / posts liked or shared or commented on as a result of connections made through this feature), % session time spent in interactions with connections made through this feature, Length of time spent on interactions with connections made through this feature

Metrics for improving feature algorithm:
1) Feature usage - % users by cohort of time since joining Linkedin platform (hypothesis: new joiners might use the feature more often to build connections), % of users by cohort of existing connections (hypothesis: users with lesser connections are likely to use more)
2) Accuracy of prediction - % of displayed "people you may know" request actually sent to (quality) for those who clicked on atleast one

North goal metric: Since primary purpose of this feature is to increase engagement on the platform, the metrics guiding impact of the feature would be:
a) length of time spent on interactions with connections made through this feature, and
b) %session time spent in these interactions