PTC University — Learning Connector

The real choice: design the screens, or design the business model.

PTC sold its software on perpetual licenses: pay once, own forever. Around it sat five learning platforms — Learning Connector, LearningExchange, Precision LMS, Digital Guides, IoTU — five URLs, five logins, five lines on an invoice. Of 550k+ registered learners, only 350k+ ever came back. The official metric counted the gap and never asked why.

I spent three weeks inside the customer-success call recordings. Nothing was wrong with the navigation. Perpetual licenses meant no recurring revenue. No recurring revenue meant stale content. Stale content meant engineers learned on YouTube instead.

That was the fork. I could ship the redesign the brief asked for, or argue that the contract underneath it was the thing actually failing — and that no amount of cleaner navigation would fix a product nobody had a reason to come back to.

The safe version would have demoed beautifully and changed nothing.

The move in the brief was a cleaner skin over the same five products on perpetual licenses. It would have looked successful in a portfolio and changed nothing — a prettier version of the thing already leaving 200k+ of its registered learners inactive and pricing itself out of recurring revenue.

So the real deliverable was never a screen. It was a political argument for killing four products and rewriting the contract — with the screens built to back it up.

I made the call, then spent a year making the case.

The CRO had 60% of revenue sitting on perpetual licenses and said so loudly. I took the phased version to the President of PTC University, and it stuck: new customers on annual subscription from Q3 2017, existing ones protected for 24 months, then migrated. The redesign took a quarter. The argument took a year.

Twelve months on: 64% of new bookings were subscription, against 0% in Q3 2017. The grandfathered cohort held at 78% through its first migration. Perpetual was 60% of total revenue when I started.

Every screen after that earned its place by making the case real:

One login, or no login. With four engineers and one PM, I shipped single sign-on across all five platforms in 14 weeks and rebuilt the IA as a graph, not a tree — a learner on Precision LMS now saw "engineers like you also learned on IoTU" without leaving the surface. Cross-traffic became the biggest acquisition channel the under-used platforms ever had.

Nobody searches the way marketing writes. The catalog mirrored the release calendar — "Precision LMS 5.0 — New Features." Engineers think in problems, not version numbers. I lost that review twice and won the third with the search-query log from logged-out users: every query was problem-shaped, zero were feature-shaped. Enrollment rose 28% the quarter after launch, against the prior four-quarter rolling baseline.

A user guide with a shipping cost. Every product shipped with a 200-page printed manual — printing, freight and warehousing cost PTC roughly $1M a year, and locked the content to a release cycle. I led the digital guidebook that replaced it: print volume fell 92% in year one, the $1M became recurring savings against the 2016 print budget, and the guide became a living surface tied to the subscription.

Translation was an architecture decision. I built the IA knowing German and Russian run 30% longer than English — short labels, shallow hierarchy, no text baked into images, 30% width buffers. Eleven locales, each WCAG AA, validated at 99% automated pass plus manual screen-reader and keyboard testing per language. The side effect: designing for accessibility made every interaction fast on the 3G handsets engineers in emerging markets actually carried. Mobile went from 4% of sessions in 2017 to 38% in 2019, across 80+ countries.

What it cost: I shipped the wrong homepage first.

The first unified landing screen showed all five product families to everyone, ordered alphabetically. I thought it would feel modern.

An engineer in Pune with a Precision-only license logged in and found 80% of the surface advertising platforms she had no right to open. Session length fell 19% in week one.

Three weeks later I keyed personalization off the license: the homepage now showed only what the customer's company had bought, with everything else behind one "Explore" link. Session length recovered, then ran 31% above the pre-launch baseline.

The user's contract is the most accurate filter you'll ever ship. I learned that by getting it wrong first.

The bet carried other costs too: a 24-month grandfather promise to existing customers that I couldn't walk back, and two lost reviews before the query log won the third. I priced those in on purpose. A wrong homepage you can fix in three weeks; a redesign that quietly preserves a broken contract, you can't.

The receipt.

Every number with baseline + window

The principle it left me with.

Most platform redesigns stall because the design team isn't allowed to ask the contract question sitting underneath them. Every move here — subscription, SSO, use-case content, license-keyed personalization, the living guidebook, locale-first architecture — wore UX clothing over a business-model skeleton.

Without that read, this would have shipped as a cleaner skin over the same five products on perpetual licenses — a prettier version of the thing that was already leaving 200k of its registered learners inactive and pricing itself out of recurring revenue. The redesign would have looked successful and changed nothing.

I run that check on every AI product now. When adoption stalls, "make it easier to find" is sometimes the answer — but "who pays for this, and does paying make them use it?" is what finds the real ceiling.

This is the kind of problem I solve full-time.

If your product has a gap between a right answer and an acted-on one, let's talk — 30 minutes, no pitch.