Senior Product Manager at HVAC.com, a Magenta Technologies brand and a joint venture of Trane Technologies and Red Ventures. The interview process started conversational, then turned into a two-hour take-home product exercise about connecting homeowners with top-quality contractors in under a day. Five remote rounds, a full-day onsite at Red Ventures HQ in South Carolina, and an offer I accepted under pressure. A large enterprise data compliance company was still working through their final round with me at the same time, but their hiring window was running behind. Here are the details on the role I actually took.
The brief came over email after the recruiter and hiring manager rounds. Reproduced here, lightly redacted.
I had questions before I started. The biggest ones were about how HVAC.com monetizes, contractor vetting, and whether there were demos of the current journey. The hiring manager's reply was the most useful possible answer:
Translation: the role is real, the scope is open, and the test is whether you can make defensible calls without a brief. That changed the assignment from "answer these five questions" to "show how you'd reason about this whole problem."
The job listing itself was the first signal. The role is titled "Senior Product Manager (who plays the Hurdy-Gurdy)" with a virtual band gag and a list of disqualifiers that includes "Operate a John Deere MS11G Series Small Manure Spreader." That is not the language of an org that wants a corporate deck. It's the language of an org that wants a real person with a clear point of view. So I built the deck to be opinionated, not safe.
HVAC marketplace decisions have analogues: Kopperfield for EV charger installation, EnergySage for solar. Both run a homeowner through a structured questionnaire, then surface qualified providers based on that input. Both also handle scheduling immediately after match. I walked through both flows, captured screenshots, and used them as the comparative reference. Solar in particular felt like a similar problem that had already started to be solved, so I went in that direction. Looking at someone else's working answer to the same problem is the fastest way to argue for a direction without internal HVAC.com data to cite.
Both target the "under one day" goal directly. Together they collapse the matching-to-scheduled-service window from days to minutes.
BDD format, Given/When/Then. Two scenarios per option (efficient matchmaking + contractor confirmation for matching, hassle-free scheduling + reminders and notifications for instant scheduling). Specific enough that an engineer reading it could ballpark a sprint estimate. That was the explicit ask in the brief.
The hiring manager told me to assume and document. I assumed referral-fee monetization (contractor pays for the successful match), then wired Referral Fee Revenue into both feature scorecards. Primary KPIs were conversion rate (questionnaire to match for matching, time to schedule for instant scheduling) and user satisfaction. Secondary KPIs covered match accuracy, time from match to service, and the referral revenue line.
For the prioritization question the brief explicitly called out, I built a RICE table comparing the two proposed features against five future-roadmap candidates (reminders/notifications, AI-powered recommendations, "HVAC.com Express" guaranteed match, payment integration, partnerships). Spelled out my assumptions for each axis (0-10 scale on Reach, Impact, Confidence, Effort) and put the actual math under the table so the panel could push back on individual numbers without burning the whole framework.
For the record: my first pass at the math was wrong, and I caught it before submitting. The corrected version had Instant Scheduling at the top, Matchmaking second.
The brief asked how I'd validate the hypothesis. I closed with an A/B testing approach: segment users between the existing flow and the new questionnaire-plus-scheduling flow, instrument both for engagement, conversion, and time-to-schedule, run user feedback surveys, and use the cohort comparison to decide whether to expand the new flow.
I'm sharing two versions side by side. The first is what I actually sent the hiring manager in 2024, exported from iA Presenter and dropped on my personal site. Putting it on the web wasn't required. The brief said "short presentation," and most candidates would attach a PDF. I wanted to wow them, and I wanted to put a flag down on my own site as a portfolio surface for live work. Look at the layout in the first embed below. It's fine. It's also clearly someone figuring out what their site can do. The current version is the same content, rebuilt with two more years of taste. Together they're the most honest answer I can give to "do I think my work gets better over time."
Open the original in a new tab ↗
I presented the deck live to the hiring manager. The conversation skipped past "is this thorough enough" almost immediately and went straight to debate over which of the two options to ship first, what assumptions I'd made about contractor inventory, and how I'd validate the under-one-day claim. That's the conversation the brief was set up to produce. We had it.
From there the process moved fast. Conversations with the future colleague, engineering leader, and the CEO. The signal was good throughout. The CEO's only remaining question was whether I'd be hands-on enough for a small org with no clear scope yet. So they flew me to Red Ventures HQ in South Carolina for a full-day onsite to find out.
The agenda, in order: HR director to start, then CEO and CFO together, a collaboration session with the CEO, CFO, my would-be hiring manager, and my future product peer, lunch with the HVAC team, the senior director of engineering, an engineering manager, and back to HR to wrap. I met the engineering leader twice. I left for the airport by 4 PM.
The point of the day was not the deck. The deck had already done its job. The point was to make sure I'd be effective in the room with the people I'd actually work with.
Recruiting told me I could take my time deciding. Reasonable. I had another process running in parallel at a large enterprise data compliance company, with one round left there. Then the hiring manager got insistent that I accept theirs. Not aggressive, but the "take your time" runway shortened with each conversation. Under pressure, I accepted.
The other process was running behind its own promised timelines around the same time. I withdrew from it. I started at HVAC.com.
The honest read: the enterprise data compliance role might have been the better strategic fit. The teams were larger, the surface area bigger, the stretch closer to where I was trying to grow. I won't know now, because I withdrew before that process closed.
Run both processes in parallel until both finish. The cleanest version of this is: accept HVAC.com's offer, finish the other process, and deal with whatever fallout the other timeline produces. Worst case, the other offer doesn't materialize and I'm in a job I want anyway. Best case, I have two real offers in hand and can make a real decision. What I did instead was let urgency from one side collapse my optionality on the other, and that's the part I'd reverse.
Push back on "you can take your time" when it stops being true. The recruiter said one thing, the hiring manager's behavior said another. Naming that gap directly ("I want to make sure I'm reading the actual urgency here, not the polite version of it") would have given me cleaner ground to negotiate one more week. Even if the answer was "no, we need an answer Friday," that's still useful information.
Keep prep capped at the brief. The exercise said no more than two hours. I went over on production polish. The substantive thinking fit inside two hours. The extra time on slide aesthetics didn't move the needle in the room. Inside the actual hiring window, that time would have been better spent prepping for the onsite or doing diligence on the parallel process.
Document my own assumptions before researching analogues. I jumped to Kopperfield and EnergySage early because they were good comparatives. The stronger version writes my own first take in 20 minutes, then uses the comparatives to pressure-test it. I wrote the answer; the comparatives confirmed it. They'd have done more if I'd let them disagree with me first.
Treat the prototype as a working artifact for the role, not a one-time deliverable. The 2024 version was a one-shot. The version on my portfolio today is the same content rebuilt with two more years of craft. The lesson I took forward, and the lesson that produced this whole "Slides I Made for Free" project: the deck for the interview is also a piece of public work. It can keep getting better after the interview is over. That ongoing iteration is itself a signal about how I think.
The artifact is the live prototype. The source repo is also linked.