.svg)
Most teams trying to prove a tool's ROI make the same mistake: they report everything they can pull, hoping enough data adds up to a convincing story. Allison Julander, Senior PMM at PDQ, did the opposite. She started with an ambitious list of metrics and cut almost all of them, keeping only what she could defend without hesitation in front of leadership.
By the end of this piece, you'll have the same filter she used to decide what to keep, and the exact reporting format you can copy to present your own results.
Why this matters
Proving the ROI of an interactive demo tool is one of the most common questions PMMs and marketers face, and it's rarely about having more data. Leadership doesn't need every metric a platform can generate. They need a small number of numbers they can trust, understand instantly, and repeat to someone else without a caveat attached. A metric that requires a paragraph of explanation, even a well-intentioned one, tends to lose credibility the moment someone asks a follow-up question.
What PDQ started with
PDQ sells IT management software to a technical, skeptical audience: system administrators who trust proof over marketing claims. Their Storylane demo lives embedded and ungated on a core product page, meant to educate buyers earlier in the funnel rather than close a late-stage deal.
Allison began with three broad goals and a metric for each: track conversions on the page and elsewhere in the same session, show that demo viewers reached a product qualified state faster than everyone else, and confirm the demo's design itself wasn't holding people back using time spent and completion rate.
Almost none of it survived contact with reality.
What she cut, and why
The converted elsewhere in the same session metric got cut first. The intention was good: capture people who saw the demo and purchased later without clicking a CTA inside it. In practice, it was hard to explain in one sentence, and a metric leadership can't repeat back to someone else stops being useful, no matter how sound the logic behind it.
The cohort comparison, checking whether demo viewers hit a product qualified milestone faster than non-viewers, ran into a more mundane problem: the internal data wasn't clean enough for a fair apples-to-apples comparison. Qualitative feedback from sales, asking whether prospects arrived at calls more informed, didn't surface anything sales actually noticed.
Time spent and full completion rate got cut for a subtler reason. PDQ's demo runs 20 to 30 steps, since their audience specifically wanted depth and realism over brevity. For a demo that long, someone converting halfway through is a win, not evidence the demo failed to hold attention. The metric was measuring the wrong thing for this specific use case.
What she kept, and why it survived
What remained was small and boring by design. For every metric she kept, Allison could answer two things in one sentence each: what it actually measures, and why it's fair to claim.
The page conversion rate, before and after the demo went live: this measures whether adding the demo changed how many visitors converted on that page, split out by funnel stage so it's clear whether the shift shows up at the top, middle, or bottom of the journey.
CTA clicks inside the demo: this measures direct engagement with a call to action embedded in the tour itself, a signal with no ambiguity about what happened.
Influenced ARR: this measures revenue from contacts who viewed the demo at some point before purchasing. Allison was explicit that this is influence, not full attribution, since a purchase involves far more than one asset. That distinction, spoken out loud every time she presents it, is part of why leadership trusts the number.
Each of these has one property in common: Allison could explain what it measured in a single sentence, and defend it if someone pushed back. She never claimed the demo alone closed a deal. The language was always influenced or encouraged, language that holds up under scrutiny instead of overselling the tool's role.
What PDQ actually reports on, month over month
Allison's recurring report tracks a small, consistent set of metrics, each with its definition restated directly on the slide so nobody has to remember what a term meant from the last meeting.
This same structure repeats every month. Nothing exotic gets added just because it might be interesting. If a metric doesn't earn a permanent place in this table, it doesn't get reported on an ongoing basis.
Reporting it the way leadership already thinks
Rather than inventing a new framework for leadership to learn, Allison mapped these metrics onto the funnel stages her leadership already used for every other channel: top, middle, and bottom of funnel, color-coded the same way. Every report restated plain-English definitions next to each metric, since interactive demos were still a new asset type for her organization and she didn't want to assume anyone remembered what a term meant from the last meeting.
The results spoke clearly inside that format. Visitors who viewed PDQ's product tour converted at nearly double the rate of those who didn't, a 92% lift (6.14% versus 3.19%), and about one in five visitors interacted with the demo each month, month after month.
The defend it or cut it framework
You don't need PDQ's exact metrics to apply the underlying method. For every metric you're considering reporting, ask three questions: can you explain what it measures in one sentence, can you explain why the number is fair to claim in one sentence, and if the number moves next month, can you explain why?
The test isn't whether a metric sounds impressive. It's whether you could explain it out loud, unscripted, if someone in the room asked a follow-up question.
The takeaway
Proving ROI isn't about finding the most sophisticated metric available. It's about ruthlessly cutting anything you can't defend, keeping what's left simple enough to repeat, and reporting it in a format leadership already trusts. Before you add a metric to your next report, ask yourself the same question Allison asks herself: if this number moves next month, can I explain why? If the honest answer is no, that's not a data problem to solve later. It's a sign the metric doesn't belong in the report yet.