Scaling Voice of Customer Sprints to Redefine ICP for Intuit ProConnect Tax

The core problem was that we did not actually understand this new high-value cohort. These are bigger, more complicated firms with layered workflows, more stakeholders, and unmet needs that look nothing like the smaller shops we knew cold. Everything else downstream depended on cracking that. Get the cohort right and the ICP definition falls out of it, messaging follows, and you have a real cross-functional strategy spanning product, marketing, and sales. Get it wrong and you are guessing in three directions at once.

The obstacles were real. PMMs already carry heavy ownership, so carving out time for true customer insight often feels impossible. Without a repeatable system, ICP stays a guess, messaging stays shaky, and GTM rests on hope instead of evidence. We also had to navigate seasonality, since accountants are nearly unreachable mid-January through mid-April, overcome the say-do gap, and recruit both customers and prospects at volume.

I treated voice-of-customer work as concentrated sprints. Each quarter I aligned with leadership and cross-functional thought partners on one meaty topic, then built a hyper-specific discussion guide and rigid firmographic criteria (firm size, services offered, tax returns filed, geography, workflow details). I recruited a 30/70 mix, roughly 30% existing customers and 70% prospects, to reduce bias, setting criteria tight enough that sessions only happened with the right profiles. Customer advisory boards were useful for speed but not for net-new learning. Freer prospect outreach worked better.

In sessions I rapidly built trust with an "uncommon commonality," immediately clarified I was not sales, and offered to share healthy and unhealthy workflow patterns I had seen across hundreds of peer firms. Accountants almost always said yes. I started with three to five extra quantitative questions beyond basic Salesforce firmographics so I could put them in a box and recall similar prior conversations, then shifted to open qualitative questions that let them talk 50% of the time about goals, pain, and next-year priorities. The guide stayed a guide. I followed depth over checklist completion and used five-whys probing. I aimed for roughly 300 customer and prospect conversations a year, mixing formal interviews with industry-event meals where the richest insights often surfaced.

The post-call workflow is where this got dramatically faster. I record calls in Zoom, then pull the transcripts straight into Claude and give it access to the full set on the back end. That means synthesis runs at scale instead of one conversation at a time. I strip PII first, then work the corpus for themes I missed, counter-hypotheses, and patterns that only show up across dozens of firms. I validated qualitative findings with analytics (to juice sample size), sales, and product, then ran targeted cross-functional share-outs, sometimes three in two weeks, positioning everything as learnings and not my opinion. I played point-guard: spotting juicy insights, looping in the right partners, and letting them drive next steps on campaigns, enablement, or product changes.

The part people underrate is what happens once the program is actually running. A real VOC engine keeps your understanding current instead of letting it rot between planning cycles, and it turns into a flywheel for the rest of your PMM life. Competitive insight, positioning, enablement, roadmap input, campaign angles. They all get faster because the evidence base is already there and already fresh.

I treated voice-of-customer work as concentrated sprints. Each quarter I aligned with leadership and cross-functional thought partners on one meaty topic, then built a hyper-specific discussion guide and rigid firmographic criteria (firm size, services offered, tax returns filed, geography, workflow details). I recruited a 30/70 mix, roughly 30% existing customers and 70% prospects, to reduce bias, setting criteria tight enough that sessions only happened with the right profiles. Customer advisory boards were useful for speed but not for net-new learning. Freer prospect outreach worked better.

In sessions I rapidly built trust with an "uncommon commonality," immediately clarified I was not sales, and offered to share healthy and unhealthy workflow patterns I had seen across hundreds of peer firms. Accountants almost always said yes. I started with three to five extra quantitative questions beyond basic Salesforce firmographics so I could put them in a box and recall similar prior conversations, then shifted to open qualitative questions that let them talk 50% of the time about goals, pain, and next-year priorities. The guide stayed a guide. I followed depth over checklist completion and used five-whys probing. I aimed for roughly 300 customer and prospect conversations a year, mixing formal interviews with industry-event meals where the richest insights often surfaced.

The post-call workflow is where this got dramatically faster. I record calls in Zoom, then pull the transcripts straight into Claude and give it access to the full set on the back end. That means synthesis runs at scale instead of one conversation at a time. I strip PII first, then work the corpus for themes I missed, counter-hypotheses, and patterns that only show up across dozens of firms. I validated qualitative findings with analytics (to juice sample size), sales, and product, then ran targeted cross-functional share-outs, sometimes three in two weeks, positioning everything as learnings and not my opinion. I played point-guard: spotting juicy insights, looping in the right partners, and letting them drive next steps on campaigns, enablement, or product changes.

The part people underrate is what happens once the program is actually running. A real VOC engine keeps your understanding current instead of letting it rot between planning cycles, and it turns into a flywheel for the rest of your PMM life. Competitive insight, positioning, enablement, roadmap input, campaign angles. They all get faster because the evidence base is already there and already fresh.

The research, well over 100 formal interviews plus hundreds more conversations annually, gave us major penetration into the up-market segment because we finally understood how those firms actually run. That understanding produced new customer personas, journey maps, and data-backed value propositions tailored to their specific pain points, plus sales enablement collateral, competitive insights, and clear product roadmap and marketing campaign implications, both short and long term, for each cohort.

We also got far tighter alignment across product, marketing, and sales, which created a more cohesive go-to-market motion. The biggest personal win was breaking down internal silos. Once teams shared the same evidence base we could move with speed, and that speed made the work fun again. A side benefit: a fair number of prospects I stayed in touch with later converted and emailed to say the earlier conversation helped. Voice of customer is now the weekly engine of my PMM work.