I took ownership of interview scheduling in Alva’s T-Rex squad, from research and problem definition to product direction and design. I had no PM support on this project, so I also owned the MVP scope and success measures. This included the scheduling workspace we released and an agentic experience for scheduling through an in-app chat.
The business believed that helping with hiring admin could give more customers a reason to choose Alva. I took ownership of understanding that work: what recruiters were keeping in sync, where it cost them time, and what would make moving it into Alva worthwhile.
I traced recent scheduling experiences
5 internal and 7 customer interviews, following the steps, tools and workarounds people used.
I mapped the patterns across those interviews
I also reviewed 90 days of interview usage to understand where scheduling could fit in Alva.
Recruiters were keeping people, calendars and interview details in sync by hand. Chasing replies and repeating updates across tools took time away from candidates.
I grouped what I heard by the work around an interview, from planning a round to arriving prepared. This map brings those patterns together.
Scroll across to follow the five moments
| Interview journey | 01Plan the wave | 02Get times agreed | 03Lock the slot in | 04Adapt when it breaks | 05Walk in |
|---|---|---|---|---|---|
| ActionsWhat they do | Reserve managers’ time at kickoff. Batch interviews into rounds. | Check calendars, propose times or send a booking link. | Send invitations. Copy the details into the other tools. | Find new times, reissue invites and follow up on replies. | Check the invitation and look up the candidate and role. |
| GoalsWhat they want | Have time ready when candidates reach the interview stage. | Agree a time that works without days of messages. | Keep the agreed slot held and every record in sync. | Keep the round moving when someone’s plans change. | Arrive with enough time and context to prepare. |
| Feelings & thoughtsHow it feels | Uncertain whether candidates will be ready for the time reserved. | Anxious that the slot will disappear while waiting for a reply. | Cautious about trusting the booking without checking again. | Drained by repeating the coordination after a cancellation. | Caught off guard when the interview arrives without context. |
| Pain pointsWhat gets in the way | Timing is a guess. Calendar blocks need constant upkeep. | Availability still needs checking. Reserved time can appear busy. | Pending invites can leave slots open. Records fall out of sync. | Replies stall. A change means updating everyone all over again. | Back-to-back meetings leave little prep time. Invites lack role details. |
| OpportunitiesThe unmet needs | I need to plan interview time with less guesswork. | I need to agree times without chasing both sides. | I need to know the booking is secure and up to date. | I need to recover a booking without restarting the coordination. | I need time and context before the interview. |
Condensed research synthesis · Notes are paraphrased from the original experience map.
I mapped the needs behind interview admin. The business hypothesis was that solving them would give prospects a reason to choose Alva.
Increase conversion among prospects evaluating Alva for their interview workflow.
Hiring teams spend less time on interview admin.
The released interface brought interview scheduling into Alva. Recruiters could coordinate interviews there, reducing the administrative work spread across emails and other systems.
I also designed a way to schedule through an in-app chat with an AI agent. This case study follows three problems: planning interview time, agreeing on a time, and knowing whether the booking is secure and up to date.
Recruiters were reserving time weeks ahead, while still guessing when candidates would be ready. I treated that early planning as part of scheduling and explored how Alva could help prepare a whole interview round.
A recruiter wanted first interviews in one week and the next round in the following week. Keeping that pace meant predicting demand and blocking managers’ time by hand.
First interviews
Next round
Predict demand · Block managers’ calendars by hand
Bring planning ahead into the experience.
I proposed a forecast of the interview round, a way to reserve time, and conditions such as the week the recruiter wanted to keep interviews within.
The expected interview round
The week interviews need to happen
Protect time before invitations
Pre-booking already helped teams protect interview time. The opportunity was to support that routine and reduce the estimating and calendar work around it.
RejectedPre-booking was tied to job setup, while recruiters needed to revisit the plan as candidates progressed.
RejectedThe calendar still left recruiters to estimate demand and reserve interview time by hand.
RejectedThe list separated slots from the rest of the calendar, making it harder to judge where a whole interview round would fit.
Agreeing on a time meant checking calendars, confirming with managers and exchanging messages with candidates. Even time reserved for interviews could appear unavailable. I explored how the agent could use calendar context to suggest times recruiters could actually offer.
Free or busy did not explain what a slot was for. Recruiters also checked with managers because the calendar alone did not tell them which time they could use.
Reserved for interviews
Busy · Unavailable
The time could be taken while candidates booked.
Use the context behind the calendar block.
I explored using event titles and descriptions when presenting times, recognizing manually reserved interview slots and giving those reservations priority.
Prioritize interview reservations
The agent would need to interpret the purpose of the time already held. That gave the AI exploration a concrete problem to address: make the recruiter’s existing preparation useful when suggesting times.
Sending an invitation did not mean the interview was confirmed. Pending replies and details spread across tools left recruiters checking whether a booking would hold. I included that status in the chat and overview so recruiters could see what still needed attention.
The same interview could live in several places. Pending invitations and missing confirmations left recruiters checking and following up to find out what was actually booked.
Check replies · Follow up · Update the records
Make the booking status clear.
I proposed an overview widget and a detailed view showing where candidates and hiring managers stood in scheduling.
Open the details to see who still needs to respond.
Recruiters needed to distinguish an invitation sent from an interview confirmed. Keeping candidate and hiring-manager status together would make unanswered invitations and missing confirmations easier to spot.
Scheduling workspace
I wanted to know whether teams connected their calendars, got interviews booked and came back to schedule again.
This was the measurement plan. The release outcomes follow in the next section.
People with a connected calendar
divided byAll users in the organization
Measured per organization. I also wanted a separate count of connected people who weren’t recruiters.
Candidates booked
divided byCandidates entering a stage with an interview assessment
Calculate the ratio for each stage, then take the median by job and the median by organization.
Overall healthBooked candidates per organization
I wanted to see where people tried scheduling once and stopped.
I planned to compare smaller customers and enterprise separately, so the results would show where scheduling was useful to each group.
The release reduced the admin around interviews and helped close contracts that had been waiting on scheduling.
Less admin work to get interviews booked.
Recruiters spent less time going back and forth between candidates, hiring managers and separate systems. That supported our department’s “time to talent” goal by leaving more time for candidates.
Exact figures are confidential.
Many stalled deals became signed contracts.
Interview scheduling had been a major obstacle for customers considering Alva. Adding it helped our sales team close contracts that had been waiting.
Research, product direction
and design.
I owned the process without PM support. That meant defining the problem, running the research, setting the MVP scope and deciding how we would measure success. Customer Success helped recruit customers; I interviewed recruiters and hiring managers and mapped their workflows and workarounds.
Within T-Rex, I designed the scheduling workspace we released and explored scheduling through an AI agent. The research guided how the agent could help plan interview rounds, use reserved time and keep track of bookings.