Generative A.I. Assisted Messaging

LinkedIn Recruiter Product

OVERVIEW

I led the Recruiter Messaging design team to ramp our first Generative AI assistance project (Project Genesis) to 1% of non-large LinkedIn Recruiter Corporate users. 

The Recruiter product launched in 2008, with continuous advancement in technology and features to help recruiters and hiring managers find the right candidate based on skills, shared values, and how and where people want to work.

The Messaging design team worked with cross-functional partners and were engaged in mid-2022-2023 to uncover the complexity of connecting all this at scale by using AI.

USER PROBLEMS

LinkedIn Recruiters key pain points:

  • Manually researching each candidate and personalizing every message is difficult to do at scale. 

  • Want to ensure they are staying within brand when messaging candidates.
     

  • Personalization of messages need to be simplified for initial and continuous contact.  

GOALS & OBJECTIVES

Launch AI-assisted messages in LinkedIn Recruiter that will help recruiters:

  • Save time by engaging with the right candidates, faster.

  • Standardize branded communication for company information.
       

  • Easily craft effective, personalized InMails to maximize candidate engagement.

RESPONSIBILITIES

As the Sr. Product Designer Leader, I managed:

A team of remote designers (1:1’s, Team meeting, Dept Meetings, Pillar Meetings, Design Crits)

  • Recruiter product (5) 

  • Enterprise Jobs (3)

Day-to-day collaboration with cross-functional teams (1:1’s, Pillar Meetings, Design Reviews)

  • Director + Principal + IC Product Managers

  • Engineering Managers + IC

  • UXR Managers + IC

  • Product Marketing Managers

  • Data Scientists

  • Recruiter Subject Matter Experts

  • Design + Technical Program Management 


Quarterly design planning cycle

  • Met with designers + cross-functional partners regarding design work

  • Reviewed the design scope + negotiated when needed

  • Assigned designers per design pillars

  • Stakeholder in product + engineering planning reviews

Design Process

DESIGN REVIEWS

My responsibility was to ensure during 1:1’s that the designer was implementing the To-Do’s + prepared for the review.

  • Recruiter team meeting -(Any Design Stage)-Design Show + Tell : Gain feedback from manager + core team peers

  • Design Crits-(Any Design Stage)-Held weekly for the Director level team to obtain feedback from the entire peer level + managers + director level. The feedback was received in the form of Comments, Ideas, Opinions, To-Do’s

  • Design Reviews- (Develop + Deliver) - Held weekly for designers + cross-functional partners

PRESENTATION PREP

My responsibility was to ensure that the designer + product manager were prepared for the requested presentation. Include practice runs with directors, quality edits, understanding the scope of the request. 

  • Departmental Office Hours- (Develop + Deliver)- Selected projects on a quarterly basis will present to the Sr. Director + all management leads

  • Design Jams- (Define + Develop + Deliver)- Selected projects on a quarterly basis will present to the VP + pillar management leads + invited cross-functional partners

  • Executive + VP Marketplace review- (Deliver) -  Selected on-going projects will present to the Executive level + invited cross-functional partners

Designs

Recruiters enhanced messaging with GenAI using Personalization; include details About You, Job Detail, Company Info, Edit Message, and Save to new template

82% of messages were rated as ‘good’ via in-product feedback, indicating an optimistic signal on the quality of the generated messages.

OUTCOMES & METRICS

40% InMail Acceptance Rate

Recruiters who personalized InMails with AI-Assisted Messages received a 40% increase in acceptance rates when compared to single, non-AI-assisted messages.

Strong Adoption

75% of contracts and 25% of seats (56 contracts and 260 seats) tried out Genesis

11% Faster Rate of Acceptance

Faster time to accept an InMail by using AI-Assisted Messages

Observation

16% of single non-templated InMail’s sent by pilot customers used Genesis content.

40% Increase In Accepted InMails

Recruiters who enabled Automated Follow-Ups saw an overall 39% increase in accepted InMails compared to those who manually follow up.

Solution

We improved our onboarding flow to drive adoption.

Lessons Learned

We implemented changes after the first ramp to a limited group.

Version A - Old Version

  • The user has to click draft personalized message to trigger the onboarding,

  • The disclaimer is right after the onboarding that requests users to accept. (Per legal required before)

  • No persistent entry point on the toolbar, when the user typed a message manually.

  • The draft personalized message wand will disappear

Version B - New Version

  • Onboarding shows when user triggers the composer, no need to click the draft personalized message

  • Removed disclaimer (Legal provided approval)

  • A persistent entry point on the toolbar, so user can access personalization anytime as needed.

Next

Design Systems