Ali Kordasti

A weighted performance measurement instrument, under document control

The organization needed to know how well teaching was actually being delivered across three sites, continuously, without relying on anyone's impression. I designed a twenty-index instrument spanning three scored domains, each index weighted one to six, totalling 1,560 points, issued and revision-controlled as a formal document.

It drew on two independent data sources that could be checked against each other: structured lesson observation, and per-lesson assessment submitted by learners. Results were published as rankings visible across all sites, so every instructor could see their standing at any time rather than learning it once a year. Measurement became a live signal instead of an annual verdict.

The instrument asked whether teaching changed what learners could do, not whether sessions had taken place. That distinction runs through most of the work below.

Scaling protective mask output twentyfold in six months

As managing director of a manufacturer of food, health products and medical equipment, I took protective face mask output from 5,000 to 100,000 units a day inside six months (roughly 2.4 million a month) with twenty production staff.

The binding constraint was not the line. It was meltblown fabric, which at that moment was the global bottleneck for respirator manufacture. Rather than compete on a spot market where price and availability were both unreliable, I contracted directly with a producer for committed supply. That single move is what made the scale-up possible; meltblown and spunbond then ran on just-in-time delivery rather than working capital tied up in stock.

The range ran from single-use surgical and procedure masks through to certified respirators: two product lines certified to KF94 and EN 405:2001+A1:2009, including a children's respirator developed in house. Every unit was UV-sterilized before packaging, and distribution ran nationally through the largest domestic e-commerce platform under our own registered brand.

Standard work that ran 35 people without daily supervision, then transferred to a factory floor

Every role's daily, monthly and periodic tasks surfaced by date in a shared calendar, were reported against a checklist, and carried a bonus on completion. Thirty-five staff across three sites operated without daily supervision. Designing it took a concentrated period of very long days; running it did not, and that is precisely why I was able to hold concurrent executive roles for years.

The test of a process design is whether it survives a change of context. Years later I redeployed the same date-driven job-description-and-checklist system across twenty production staff at a manufacturing company. Different industry, different workforce, same mechanism, same result.

A multi-criteria decision platform every proposal passed through

Decisions were being made the way they are in most owner-run companies: by the owner, from intuition, with no record. I built a governance platform where any staff member could submit a proposal, have it evaluated by colleagues against defined criteria, and see it scored across weighted dimensions. More than a hundred proposals ran through it, and every strategic decision was routed down the same path, including my own, submitted on identical terms.

The scoring logic was prototyped in Excel and used for live decisions first. Only once it had proven itself in daily use did I build the web platform carrying proposal, scoring and dashboard functions across all sites, learning HTML, CSS and PHP to produce the first version myself.

Doubling enrolment conversion, piloted before it was scaled

Intake ran on walk-ins converting at 35%. I replaced it with a single session in which prospects were assessed on the spot and received their own results (a four-skill placement outcome and a recommendation) before they were asked to decide anything. They experienced the organization's assessment method rather than being told about it.

The redesign ran at one site first, measured against the existing 35% baseline, and was scaled only once the result held. Scoring and data entry happened in parallel with the session itself, so a multi-day turnaround became a same-session outcome.

Three physical sites to fully online delivery in one month, without losing a teaching day

When the pandemic arrived I closed all three locations permanently and moved the entire operation, roughly three hundred learners and every instructor, onto Moodle within a month. Not one teaching day was lost, and enrolment rebuilt to between four and five hundred over the following months.

The reason it worked was the training method rather than the platform choice. Instead of running sessions for staff and learners, I produced self-serve screen recordings they could use at whatever hour they hit a problem. It is the same principle I had already applied twice elsewhere: put the instruction where the person already is, at the moment they need it, rather than convening an event and hoping for attendance. Adoption is where process change usually fails, and it is the part I design for first.