Why I stopped selling "95/100" vanity scores to job seekers (and built a Context Engine instead)

Hey everyone,

I wanted to share a specific realization I had while building in the HR-Tech space, specifically regarding resume scanners.

If you look at the market, it is flooded with tools that promise to "grade" your resume. You upload a PDF, and a script runs some heuristics (formatting, action verbs, length) and spits out a dopamine-inducing number: "Your Score: 95/100."

As a builder, these are great metrics to sell. They are sticky, they make the user feel accomplished, and they are easy to code.

The Problem: The "Placebo" Effect I realized that by optimizing for a generic score, I was essentially selling a placebo.

Here is the reality of the hiring side (ATS): A candidate could have a "98/100" resume structure, perfect grammar, great layout, Harvard formatting. But if they are applying for a React Native role and their resume is filled with Flutter experience, their "Match Score" on the recruiter's dashboard is near zero.

The generic scanner told them they were perfect. The reality was they were irrelevant.

The Pivot: Context > Syntax I decided to scrap the idea of a "Generic Score" for my project, cvcomp. I realized that a resume cannot be graded in a vacuum. It can only be graded relative to a specific problem it claims to solve (the Job Description).

I rebuilt the core logic around what I call a "Context Gap" approach:

Forced Context: You cannot scan a resume on cvcomp without providing the Job Description (JD). It forces the user to stop thinking about "Is my resume good?" and start thinking "Is my resume good for this role?"

Semantic Matching: Instead of just checking if keywords exist, we check if the experience aligns.

Reducing Friction: The biggest pain point I saw was the "Tab Switching Hell." Users would see a mistake in a report, go to Word/Canva, fix it, export to PDF, and re-upload. I built a Live Editor to bridge this—users accept AI suggestions or type manual fixes directly in the browser.

The Challenge for Indie Founders The hardest part of this pivot was actually user psychology. Users want that easy 95/100 score. Telling them "You are actually a 40% match because you missed these 3 key technical requirements" is a painful UX.

But I believe it's the only honest way to build in this space. If we aren't helping users get past the actual semantic filters of an ATS, we are just selling them false hope.

I’d love your feedback I’m currently refining the "Live Editor" flow. I’d love for some other makers to tear it apart and tell me if the UX makes sense to you, or if the "Gap Analysis" feels too harsh compared to standard scanners.

You can play with the tool here: cvcomp.com

Cheers
Prashastha