Designing a job referral platform from 0 to 1

Why job referral?

We saw a huge unmet need for job referrals in the market. Beyond the user demand, referrals also opened up more monetization paths than pure job-search ever could - sponsored posts, premium referrer tools, paid match acceleration.

For AiTou, this wasn't a feature - it was a product pivot.

What shipped, and what it moved

Job referral is messy - for both sides

I empathized with users by interviewing 9 job-seekers and 8 referrers to surface pain points across the existing referral flow.

Job-seekers

Usually look for referrals on LinkedIn by connecting with others or reach out to friends who already work there. The connecting and back-and-forth messages is time-consuming. It is also hard to track their requests or get responses timely.

Referrers

Most referrers said that it is difficult to find suitable candidates, they don't know where to post the referral information, and it is hard to manage all the requests via emails.

Who I designed for

Two distinct sides of the marketplace, each with their own jobs-to-be-done. The platform had to serve both - or it served neither.

Job-seeker persona

Referrer persona

Building bridges - redefining job referrals for both sides

We wanted to design a platform that lets job-seekers find matching referrals & request them (no more back-and-forth, no more wondering if anyone got the message), and lets referrers post referral info & manage requests easily (one place to find candidates, one place to respond).

Two sides, one platform - the design has to keep both moving

A platform that matches seekers with referrals - and helps referrers manage requests

Two coordinated flows: job-seekers browse referrals, request, and track status; referrers register, build a profile, post referrals, and manage incoming requests. Below is a walkthrough of each side, plus the design process I followed to get there.

For Job-seekers

Fill in a referral request, submit it directly to the referrer's inbox, then track or withdraw the request from the profile page.

Design Process

The flow I followed to ship this MVP - testing fed back into prototyping rather than waiting for one final review at the end.

Research → Ideation → Prototype → Testing → Implement (with a feedback loop back to Prototype)

How are other platforms doing job referral?

Before designing, I researched the landscape. Few platforms focus mainly on job referrals; the closest patterns came from Exponent and 1point3acres. I broke down each model's pros and cons to find where AiTou could differentiate.

Exponent · Anonymous request

Job-seekers submit referral requests blindly; they don't see referrer info or profiles. Referrers reach out if interested.

1point3acres · Public referral posts

Referrers post referral details on the platform to recruit candidates. Job-seekers browse posts and reach out to referrers directly.

Defining and building an MVP that matters

After identifying key gaps and opportunities in the market, we brainstormed ideas, prioritized features, and defined a focused MVP to address the most critical needs - protecting both sides' time and privacy without bloating scope.

Feature prioritization - high-impact, low-effort first

Scope brainstorm - what's in v1, what waits for v2

Mapping the user journey

Two flows, two distinct mental models. The job-seeker flow optimizes for finding and requesting; the referrer flow optimizes for posting and managing. Each side's pain points became checkpoints in the diagram.

For job-seekers

For referrers

Two rounds with real seekers

Recruiting

• Considers referral an important component when finding a job
• Currently seeking, or has applied a job referral in the recent 2 years (familiar with the process)
• Uses apps or websites to find referrals

Method - Think Aloud

Participants shared their screen (if remote) and thought aloud through the whole process. I recorded the screen and verbal answers, then walked them through three tasks:

• Take a minute to explore the site - what is AiTou about?
• You're seeking a Software Development Engineer referral at Google. Search and request the referral.
• You've requested a referral - now try to track its status.

Think Aloud - planning prompts and follow-up questions per page

1st round - 5 in-person participants

2nd round - 17 participants total

Each team member found 2-3 target users. We sent out a survey to gather quantitative feedback to balance the qualitative observations from round 1.

What changed after testing

Browse Referrals

Users couldn't tell which referrals were active vs. expired and missed the filter affordance. I rebuilt the list with status pills, clearer hierarchy, and a persistent filter rail.

Referral Card

What information should appear in the card before a click? Testing showed seekers wanted role, location, deadline, and referrer's seniority - not the referrer's name. Cards became scannable instead of dense.

Referral Details

The detail page balanced two competing needs: giving seekers enough context to decide, and protecting referrer privacy. I added an explicit "what referrers will see" section so seekers self-edit before sending.

Five remote sessions with active referrers

AiTou's existing referrer list became my recruitment pool. I ran 5 remote sessions where referrers screen-shared and thought aloud while posting a referral and managing incoming requests.

Referrers asked us to share screens and think aloud while completing the post + manage flow

What changed for the referrer side

Post Referrals

Referrers found the original post flow long and unclear about what was required vs. optional. I split the flow into focused steps, surfaced "required" cues up front, and added a preview before publish.

Privacy Concern

Referrers worried about exposing their email or phone. I added a "what's visible to seekers" toggle on every step and routed all communication through the platform until the referrer opts in.

A new system for the rebrand

Alongside the new flows, I built AiTou's first formal design system to support a parallel rebrand - typography, color, components, and motion. The system gave the engineering team a stable foundation to ship the MVP fast and a clear path to extend later.

Design system - color, type, components