Mark Progano Portfolio
Eight numbered case studies, each split into Context, What was needed, Work I did and What Changed, from a lease-insurance backlog tool to two self-built live products (Please Ramble, AppBoards).
Location
Didn't say. We're respecting the mystery.
Skills
What makes it work
A breakdown of the choices that make this portfolio stand out.
Problem-solution structure that mirrors PM thinking
Every case study expands into the same four labelled parts: Context, What was needed, Work I did, What Changed. The fixed "Work I did" slot forces a first-person account of his own contribution, and "What Changed" makes each entry end on a result instead of a deliverable.
Case-study titles framed as specific problems, not generic labels
The numbered "Selected Works" list (six entries, "001" through "006", with a "+ Show All" toggle revealing 007 and 008) reads "Making a brand kit useful after the handoff," "Clearing a multimillion-dollar backlog," "Finding answers in a legal-sized haystack," and "Trusted as the product person's product person" — each title states the actual friction solved instead of a generic "Redesign Project" label, which makes six very different engagements memorable at a glance.
Numbers only where there is a real number
Metrics show up only in the "What Changed" line, and only when there is one: the backlog tool cut processing from weeks to days with roughly 85% higher team productivity, the regulations search took lookups from hours to minutes, and the PM-in-residence mentoring lifted session satisfaction 22%. The self-built products (Please Ramble, AppBoards) skip invented numbers and point to the live product instead.
A self-aware, funny FAQ that still does real qualifying work
The closing FAQ accordion includes "What about AI, do you know AI? (AI. AI. AI...)" alongside serious questions like pricing and industries served. The joke doesn't undercut the page — it sits next to real answers about process and cost, which makes the whole page read as a person rather than a template.
What Product Managers can take from this
Specific, actionable tips to apply to your own portfolio — no generic advice.
Answer three questions in every PM case study: what was the problem, what specifically did YOU do (not the team), what changed as a result?
Title your case studies after the specific friction solved ("Clearing a multimillion-dollar backlog") instead of a generic project label — it's more memorable and it previews the story before anyone clicks in.
Include metrics before AND after. "Improved retention by 23%" is a claim; "retention was 41%, we targeted 50%, we shipped X, retention is now 64%" is a story.
Write PM case studies for a non-PM reader. The hiring committee often includes design, engineering, and executive stakeholders who are evaluating communication ability.
A little personality in your FAQ (a joke question alongside the real ones on pricing and process) makes a portfolio feel like a person, not a template — as long as the real answers are still there.
Projects
Titles like "Clearing a multimillion-dollar backlog" and "Finding answers in a legal-sized haystack" do the summarizing before the click. Mark Progano numbers eight case studies, six up front and two behind a "Show All" toggle, and names each after the friction it removed, not the deliverable.
004 — Clearing a multimillion-dollar backlog
For a lease insurance and risk business, a core property disposition workflow lived in spreadsheets and email threads. Cases piled into a large backlog, each could take weeks, and millions of dollars were effectively stuck. Mark sat with the operators, walked real cases end to end to learn the actual rules, gray areas and workarounds, then designed a focused internal tool that pulled the right data into one place.
The work starts with sitting next to the people doing the job, not with a spec. Learning the gray areas and workarounds before designing is what makes an internal tool get used instead of worked around.
Weeks to days on processing time, ~85% productivity gain — both stated on the card
Research method is concrete: walking real cases end to end with operators
Card follows the site's fixed Context / What was needed / Work I did / What Changed format
005 — Finding answers in a legal-sized haystack
A nonprofit helping communities and local agencies understand environmental and water regulations across multiple states and counties needed to answer questions like "what applies here, for this river, in this county" without hand-hunting clauses in PDFs. Mark identified the key sources with their team, built a small crawling pipeline to pull and normalize public regulations and guidance, indexed the corpus with embeddings and metadata for plain-language, topic or location search, and added filters and saved searches for recurring questions.
A small, scoped retrieval system aimed at one recurring question shape, with saved searches for reuse — applied AI sized to the problem rather than an AI headline.
Crawl → normalize → embed → search pipeline described end to end
Location-aware queries ("this river, in this county") drive the metadata design
Outcome stated as minutes instead of hours, no inflated metric
003 — Please Ramble: letting clients talk their way to a better brief
A self-initiated product for collecting context people can say out loud but struggle to write. A creator asks one question or a short sequence and shares one link; respondents answer aloud one question at a time, and each answer is transcribed and kept organized by question.
A shipped, paid-for product owned end to end, not a concept. It shows a PM who can take an observed friction (thin, over-edited written answers) all the way to a working product.
Live product with accounts and billing, linked from the case study
Problem framed from both sides: thin written answers vs. unstructured voice notes
Site nav lists an MCP page alongside use cases and pricing
007 — AppBoards: turning disposable prototypes into a shared workspace
Prototype variants were easy to generate but lived scattered in chat threads and local files. Mark built a real-time service, a cross-platform command-line agent and a browser viewer that keep multiple HTML directions on one board; local changes sync automatically, collaborators join through one link without an account, and the board exports as one file.
Tooling built around how prototyping with AI coding agents actually happens, then dogfooded in his own practice.
Sits behind the "+ Show All" toggle as the seventh case study
No-account join via one link, single-file export
Works with any coding agent that has file access, per the live site
More Product Manager portfolios
Ready to build your portfolio?
Follow the Product Manager roadmap — skills, projects, and timeline to get hired.