UX/UI
Lookalike Feature
/ Novia101
Context: Novia is an AI-powered recruitment platform that helps hiring teams find the right candidates faster and with greater accuracy. It reduces manual effort and prevents strong candidates from slipping through the cracks.
The problem
Through conversations with recruiters, we discover a recurring gap in the hiring process.
While job requirements can be documented clearly, many hiring decisions are also influenced by an internal mental model of the "ideal candidate" - built from past successful hires, team dynamics, personality, or other intangible qualities.
Recruiters often know this profile instinctively but struggle to explain it
in words.
As a result, there is no way to communicate these implicit preferences to the product. This creates a disconnect between the candidates Novia101 identifies as strong matches based on explicit requirements and the candidates hiring managers feel naturally drawn to. The Lookalike feature explores how this additional layer of hiring intuition can be surfaced and translated into a meaningful product experience.
The Thought Process
The feature planning began as a series of exploration sessions with Novia's AI lead. While she investigates what the LLM can realistically generate, I focus on imagining how the design can be translated into an already exciting Role building process, without adding friction.
Scoping the MVP, we decide to focus t this point on the Explicit Lookalike (manager-driven input), while the Implicit (behavior-inferred) version is deferred to a later phase. Compliance is also considered from the outset, including how the system detects protected-class proxies (e.g., "Female only/Under 35").
The final input as imagined in the early stages:
The Solution
One core principle: help recruiters externalize an intuitive preference without making the process feel like filling out another form.
To reduce friction, the experience is broken into three lightweight, clearly numbered steps with visible progress and concise guidance throughout.
Step 1 - Define. Users choose the easiest starting point for them: a LinkedIn profile, an existing candidate from the system, or a CV. They can also describe the kind of person they have in mind in free text.
Step 2 – Refine. The LLM generates up to three targeted follow-up questions only when critical information is missing. Once completed (or skipped), it presents an editable summary of the captured persona and translates the extracted signals into structured skill and experience chips that users can review and adjust.
Step 3 – Validate. Instead of asking users to simply "trust the AI," the system generates fictional candidate personas representing different interpretations of their preferences. Users react to these examples, helping refine the profile while building confidence that the system has correctly understood their intent.
The Lookalike builder:
The Impact
As the feature is still in development, its impact is currently evaluated qualitatively rather than through production metrics. The concepts and interaction patterns are validated with the stakeholders who helped define the problem from day one, ensuring the proposed solution aligns with real hiring workflows and business goals.
The expected outcome is to give recruiters a more intuitive way to say "find me more like this," strengthening Novia101's "Fit, don't just match" positioning.
Success will be evaluated after launch using two primary KPIs:
Lookalike adoption - At least 30% of new jobs are expected to use the Lookalike feature within the first three months.
Reduced job requirement revisions – We expect recruiters to make fewer changes to the job requirements after reviewing the first batch of recommended candidates. Indicating that the initial hiring intent captured more accurately, the expectations.
UX/UI
Lookalike Feature / Novia101
Context: Novia is an AI-powered recruitment platform that helps hiring teams find the right candidates faster and with greater accuracy. It reduces manual effort and prevents strong candidates from slipping through the cracks.
The problem
Through conversations with recruiters, we discover a recurring gap in the hiring process.
While job requirements can be documented clearly, many hiring decisions are also influenced by an internal mental model of the "ideal candidate" - built from past successful hires, team dynamics, personality, or other intangible qualities. Recruiters often know this profile instinctively but struggle to explain it in words.
As a result, there is no way to communicate these implicit preferences to the product. This creates a disconnect between the candidates Novia101 identifies as strong matches based on explicit requirements and the candidates hiring managers feel naturally drawn to. The Lookalike feature explores how this additional layer of hiring intuition can be surfaced and translated into a meaningful product experience.
The Thought Process
The feature planning began as a series of exploration sessions with Novia's AI lead. While she investigates what the LLM can realistically generate, I focus on imagining how the design can be translated into an already exciting Role building process, without adding friction.
Scoping the MVP, we decide to focus t this point on the Explicit Lookalike (manager-driven input), while the Implicit (behavior-inferred) version is deferred to a later phase. Compliance is also considered from the outset, including how the system detects protected-class proxies (e.g., "Female only/Under 35").
The final input as imagined in the early stages:
The Solution
One core principle: help recruiters externalize an intuitive preference without making the process feel like filling out another form.
To reduce friction, the experience is broken into three lightweight, clearly numbered steps with visible progress and concise guidance throughout.
Step 1 - Define. Users choose the easiest starting point for them: a LinkedIn profile, an existing candidate from the system, or a CV. They can also describe the kind of person they have in mind in free text.
Step 2 – Refine. The LLM generates up to three targeted follow-up questions only when critical information is missing. Once completed (or skipped), it presents an editable summary of the captured persona and translates the extracted signals into structured skill and experience chips that users can review and adjust.
Step 3 – Validate. Instead of asking users to simply "trust the AI," the system generates fictional candidate personas representing different interpretations of their preferences. Users react to these examples, helping refine the profile while building confidence that the system has correctly understood their intent.
The Lookalike builder:
The Impact
As the feature is still in development, its impact is currently evaluated qualitatively rather than through production metrics. The concepts and interaction patterns are validated with the stakeholders who helped define the problem from day one, ensuring the proposed solution aligns with real hiring workflows and business goals.
The expected outcome is to give recruiters a more intuitive way to say "find me more like this," strengthening Novia101's "Fit, don't just match" positioning.
Success will be evaluated after launch using two primary KPIs:
Lookalike adoption - At least 30% of new jobs are expected to use the Lookalike feature within the first three months.
Reduced job requirement revisions – We expect recruiters to make fewer changes to the job requirements after reviewing the first batch of recommended candidates. Indicating that the initial hiring intent captured more accurately, the expectations.
Previous
Next