Case study · Operations redesign · 2024
4.6× Conversion rate lift · application → interview

A conversion lift, built in ten weeks.

Application-to-interview conversion went from 0.5% to 2.3% in ten weeks. The tech stack wasn’t broken. The operation underneath was. The five-leverage redesign that fixed it is the spine of Mezura’s method today.

0.5% → 2.3% conversion 10 weeks end-to-end −21% weekly cost 28+ hrs/mo eliminated 100% team retained
Documented outcome

The number, stated plainly.

4.6×
Conversion rate
Application-to-interview conversion: 0.5% → 2.3%, in ten weeks.

A 2024 sprint at a Y Combinator-backed company (YC W18 portfolio). Engaged as Lead Systems Architect, the operation was rebuilt across five leverages — the same five that now structure every Mezura engagement. No new product. No new team. No new capital.

The leak wasn’t in the platform. It wasn’t in the model. It was in the operation underneath — the five leverages every operating team runs on. We rebuilt those five. Conversion followed.

The leak wasn’t in the platform.
It wasn’t in the model.
It was in the operation underneath.
The business at risk

A product, a team, and capital. None were the problem.

The product is structured coaching that places candidates into engineering, product, and data roles. The business model is income share — the company earns only when a candidate is hired. That makes one number the gate to everything: application-to-interview conversion. Below a threshold, the unit economics collapse.

When the engagement began, that number was 0.5% — roughly one in two hundred applicants reached the placement stage at all.

Leadership had assumed the conversion problem was a product problem and had been iterating on the product without moving the metric. The conversion problem was an operating-model problem. The engagement was as Lead Systems Architect — to re-engineer the operations stack, not to advise on it.

The audit

Where the friction actually was.

Two weeks of diagnostic walked the funnel transition-by-transition. The losses weren’t in volume. They weren’t in coaching content. They were in five structural decisions every operating team makes — and every one of them was misaligned.

  1. Input quality · unstratified

    Applications entered the pipeline unevenly screened. Applicants who were not a fit for the program were accepted at the same priority as fit applicants. Coaches spent time on conversion paths that could not convert.

  2. Workflow design · single serial queue

    Every application moved through one queue, in order. No triage. No SLA. High-signal candidates and low-signal candidates received the same handling, at the same pace.

  3. Decision protocols · unwritten

    When to escalate, refer out, or decline was made on instinct. Inter-reviewer variance was high — the same applicant profile got different decisions from different coaches on different days.

  4. Automation layer · absent

    Reporting was 28+ hours of spreadsheet work per month. Leadership saw the funnel a month late. Routing, status communications, and scheduling were being done by humans — with no single source of truth tying applicant data, coach assignments, and outcomes together.

  5. Incentive alignment · misaligned

    Compensation rewarded activity — applications processed, hours logged — not interview yield. The system paid coaches to spend time, not to convert.

That’s the seventy percent. None of it was a tool problem. All of it was an operation problem.

All of it was an operation problem.
Not a tool problem. Not a product problem. Not a team problem.
Fig. 01 · Workflow redesign

From one queue, to three tracks.

The single serial queue was the structural cause of the conversion collapse. Replacing it with an intake scorecard plus three signal-stratified tracks — each with its own SLA — was the change that did the most work.

Before
Before: applicants flow through a single FIFO queue to a coach pool, then to interview, yielding 0.5 percent conversion. Applicants priority equal Single queue FIFO · same protocol Coaches by instinct Interview 0.5% — No triage. No SLA. Inter-reviewer variance high.
After
After: applicants flow through an intake scorecard that stratifies them into three parallel tracks with SLAs, then converge to interview, yielding 2.3 percent conversion.
The rebuild

Five leverages rebuilt in ten weeks.

Each fix maps one-to-one to the audit finding above it. The work was sequenced — not parallelized — because each leverage depended on the one before it being stable.

01
Input quality stratified

Built a structured intake scorecard. Applications that didn’t clear the threshold didn’t enter the coaching queue. ~30% of volume dropped before any coach touched it. Coach hours per qualified application doubled.

02
Workflow design three parallel tracks

Killed the single queue. Replaced it with three parallel tracks segmented by candidate signal strength. Each track had its own SLA and its own protocol. Triage became automatic.

03
Decision protocols codified

Documented the calls coaches were making by feel. Made them explicit. Built a four-question protocol any reviewer could apply consistently. Inter-reviewer variance dropped sharply.

04
Automation layer built

Airtable as single source of truth; Zapier as trigger engine; Google Apps Script running the intake scorecard, track-assignment logic, and weekly reporting; Slack handling coach assignment and SLA alerts. The 28+ monthly hours disappeared.

05
Incentive alignment rebuilt

Shifted compensation from activity-based to outcome-based. Volume processed dropped. Quality and yield rose. The incentive matched the outcome.

By the end of week ten, conversion had moved from 0.5% to 2.3% and held there. Weekly operating cost fell 21%. 100% of the operating team stayed through the change.

Sequencing

Ten weeks, sequenced.

Each leverage unblocked the next. Input quality stratified before workflow tracks could be sized; tracks were live before decision protocols had stable conditions to codify against; protocols had to be documented before they could be automated; automation had to run before incentives could be tied to its outputs.

W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 Input quality 01 Workflow design 02 Decision protocols 03 Automation layer 04 Incentive alignment 05 W1 · diagnostic W5 · conversion moves W10 · 4.6× sustained
Fig. 02 · Automation stack

Airtable, Apps Script, Zapier, Slack.

No new platform. No vendor-locked product. A four-tool stack — Airtable as the single source of truth, Google Apps Script running the intake-scoring logic, Zapier orchestrating the triggers and event flow, Slack closing the coach-assignment loop. The same architecture is reproducible inside almost any operating team.

Trigger
Airtable · new applicant record
An applicant submits the intake form. Airtable creates the row. That row is the single source of truth for the funnel from here on.
Action 01 · via Zapier
Google Apps Script · intake scorecard
Zapier triggers on the new record. Apps Script computes a signal score, assigns a track (A / B / C), and returns the result.
Action 02 · write-back
Airtable · update record
Score, track assignment, and status written back to the original record. The next step in the workflow sees a single canonical state.
Fan-out
Output 01 · Slack
Coach assignment & SLA alert posted to #intake-tracks.
Output 02 · Email
Templated confirmation to the applicant. Track-specific copy.
Output 03 · Live dashboard
Leadership funnel view, real-time. Per-track yield, visible for the first time.
Eliminated 28+ hours of monthly manual reporting. Funnel visible in real time for the first time.
What changed

Before, after, ten weeks apart.

Conversion
4.6×
application → interview
2.3% 0.5% 4.6× lift
In the one metric that gated the business.
Weekly cost
−21%
weekly operating cost
baseline w10
Lower cost, higher yield, simultaneously.
Reporting
28+ hrs
eliminated each month
Before After real-time
Funnel visible in real time for the first time.
Retention
100%
operator retention through the change
100%
No one left during the ten weeks.
What this proves about Mezura’s method

Five leverages. The same five, every time.

The 2024 engagement isn’t a one-off. The five leverages — input quality, workflow design, decision protocols, automation, incentive alignment — govern every operating team Mezura works with today. A clinic group. A property portfolio. A SaaS RevOps function. Different vocabulary, same structure.

Every Mezura engagement starts in the same place — a Chapter 1 Operational Friction Diagnostic. Ten business days. $25,000, firm. You get back a ranked, dollar-sized inventory of where your operation is leaking, a 90-day fix plan, and an implementation envelope for the rebuild.

If we don’t find friction worth fixing, we say so. If we do, you decide whether to keep going.

A clarity guarantee

If the diagnostic doesn’t produce a clear, dollar-sized map of where your operations are leaking, Mezura keeps working at no additional cost until it does. The $25,000 buys an answer, not an effort.

Next step

Find out where it’s leaking. On a free call.

A free 30-minute call with the founder — a direct read on whether, and where, you’re leaking, and whether the $25,000 Diagnostic is the right next step. No deck, no pitch.