Enhancing recruiter’s outreach messages leveraging AI (WIP)

Context

For recruiters, they want to send unique outreach messages from LinkedIn Recruiter to attract top candidates. Recruiters will personalize and tailor their message using candidate profile information in their message. However, this process can be time-consuming and it requires intense research. Also, it’s hard to achieve at scale when reaching out to potential candidates in bulk. With this in mind, the product team created a concept using generative AI to scrape candidates’ LinkedIn profile information and create unique, personalized outreach messages at speed. Specifically, we hypothesized that integrating generative AI into Recruiter messaging could increase recruiters' productivity and produce unique, compelling outreach messages. 

Research goals

Concept goals

• Provide recruiters with a composing tool that scrapes data from LinkedIn member profiles;
• Reduce recruiters' time and effort to compose messages;
• Allow recruiters to create unique, eye-catching outreach messages to passive candidates with ease.

Usability goals

• Measure discoverability of the AI CTAs, Customize panel CTA, Auto-drafting options panel, and bulk editing CTAs in the inMail flow.
• Evaluate participants' ability to engage with the key data point pills interactions, up & down paragraph ordering arrows, and the Draft again CTA in the Customize panel, and horizontal navigation arrows in the bulk message carousel panel.
• Assess participants' ability to understand message drafting CTAs and notifications, Customize panel interactions, message editing, and LinkedIn feedback panel copy.

Research questions

• What makes a quality outreach message to passive candidates from recruiters?
• How can recruiters leverage AI to save time when composing outreach messages?
• What are recruiters’ perceptions of leveraging LinkedIn’s AI to compose outreach messages? Will they find the experience transparent, safe, clear, and controllable from a user standpoint?

Stakeholders

• Principal product manager
• Senior UX Designer
• Product marketing manager
• UXR managers
• Research Operations

Timeline

• 3 weeks total.
• 1 week to workshop with stakeholders, solidify concept design for testing, select recruit audience for the study, and design the study approach.
• 1 week to conduct research sessions.
• 1 week to analyze and synthesize data for the final research deliverable.

Recruit audience

For this study, we selected recruiters who source candidates or sourcers that frequently use inMails to reach out to passive candidates. We chose recruiters and sourcers as a population for two reasons:

• This audience focuses on using pre-built templates to send outreach messages in bulk.
• This audience is most likely to use AI to help craft unique personalized messages to reduce time.

Methodology & Research Approach

Research method

A hybrid approach of merging a concept and usability test was recommended to meet the needs of the study. We took this approach for two reasons:

• Performing a concept test would surface deep insights around the attitudes and perceptions of recruiters leveraging AI to craft outreach messages;
• Conducting a usability test would allow the product team to leverage tactical insights to iterate on the current design and successfully launch an MVP.

Research approach

We conducted 5, 1-hour moderated research sessions by testing three different scenarios using Figma. Additionally, we designed the study to allow for exploratory follow-up prompting through each scenario. We used the following scenarios in the study:

• Scenario 1: Locating and engaging with the AI-composed feature in Recruiter messaging.
• Scenario 2: Discovering and engaging with the Customization CTA and panel.
• Scenario 3: Discovering and enabling auto-generated AI-composed messages, and discovering and engaging with the auto-generated settings menu.

Sample size

We chose a sample size of n=7 to accomplish the needs of the research.

• 3 participants were staffing recruiters, 3 participants were in-house recruiters, 1 participant was a recruitment coordinator.
• Participants were US recruiters and sourcers.
• Participants averaged sending 30 inMails a month.
• A mix of demographics (i.e., age, gender, etc.) was chosen to create a diverse sample size.

Impact

The insights and recommendations surfaced from this study led to a 36% improvement in inMail acceptance rate from potential candidates after launching Recruiter AI-composer MVP. Additionally, insights from the study shaped future product strategy by planning to increase transparency with AI controls, develop flexibility through advanced AI customizable settings, and road-mapping future product features to increase advanced customization.