Running product discovery when you can't A/B test

Or, how to navigate when you don't have GPS

You’re at a product conference. The speaker says “run an A/B test.” Everyone nods. You sit there thinking: I have 50 customers. They know my email and my CEO’s phone number. My sales cycle is 18 months, a ‘cohort’ is two people and five versions of the product. My users are behind a firewall I need security clearance to access. Who exactly am I A/B testing?

The standard discovery playbook of continuous experimentation, quantitative validation, and rapid iteration assumes a world where data is abundant, cycles are short, and you can test in production. That world exists. It’s just not your world.

This post is for the enterprise PM who’s been told they’re doing discovery wrong because they can’t run experiments on 10,000 users. You’re not doing it wrong. You’re navigating without GPS.

My blog post images are sourced on Pixabay from human creators.

Navigation without GPS

B2C product discovery has GPS: real-time positioning from millions of data points. You always know roughly where you are. Take a wrong turn? Reroute in seconds. The data is abundant, the feedback loops are tight, and statistical significance is achievable before lunch.

Enterprise discovery has a compass, a hand-drawn chart, and a few locals who each know their stretch of coastline. You navigate by triangulation - taking bearings from multiple imperfect reference points, and finding your position where they converge. You never get GPS-level precision. But a local who knows the reef is worth more than a satellite overview, and the navigator who can read tides, landmarks, and currents will find the harbour when the GPS-dependent sailor is lost the moment the signal drops.

Here’s the reframe that matters: in enterprise, each signal is more valuable, not less. A single customer interview in enterprise can reveal more about the market than a thousand survey responses, because enterprise customers have been pondering their problem, can describe workflows in detail, and tell you things no analytics dashboard captures: the political dynamics, the regulatory constraints, the reason they bought your competitor and regretted it. Sometimes it’s hard to schedule interviews with the right people, but it’s never hard to get them to talk. You need to focus on finding those right people, the ones who know the reefs like a local.

Every conversation is discovery

The books will tell you to schedule regular discovery interviews with customers. That’s good advice. It’s also an incomplete picture of how enterprise discovery actually works.

In enterprise, discovery happens every time you talk to a customer, not just in formal interviews. The trick is recognising it, and being ready to learn from it.

I’ve run professional services engagements where I helped customers install and configure the product. Those weren’t “discovery calls,” they were implementation sessions. But by asking about their environment, feeling the rough edges myself, and watching where the product met reality, I learned more about our customers’ actual workflows than any scheduled interview could have surfaced. You see the gap between what the product assumes and how the world actually works, and you feel it with your own hands, not through a screen share.

I’ve run workshops and webinars where I’d ask the audience about their context so I can “tailor this presentation to your situation.” That’s a facilitation technique. It’s also discovery in disguise. When 30 people in a room tell you what they care about and why, you’re collecting bearings. Different organisations, different roles, different priorities. The patterns that emerge across them are gold.

And I’ve jumped on customer support calls with angry customers. Those conversations, when you sit and listen to someone who is frustrated enough to escalate, teach you what people really care about and why. Spend time to go beyond the feature they’re requesting to the problem they’re trying to solve. In one case, I spent an hour on a call with a customer who was ready to walk. By the end of the call, I understood not just their issue but the organisational pressure driving it, something no NPS survey would have surfaced. We saved the account, and I won the adoration of CSM and Sales for months afterward, which, let me assure you, is a very useful form of social capital.

The principle: every customer conversation is a bearing. The formal discovery interview, the implementation session, the support escalation, the casual chat at a conference. You’re always triangulating. The question is whether you’re structured enough to capture what you learn and connect it to the broader picture.

Sales as your discovery network

Enterprise PMs have a discovery asset that B2C PMs can only dream of: a sales team that talks to prospects every single day.

Most product teams treat sales as a feature-request pipeline. “Customer X needs Y to close the deal.” That’s a data point, not a direction. But when five salespeople across different territories independently describe the same pain point (when you ask ‘why’ enough times), the same competitive loss pattern, the same objection they can’t overcome - that’s a strong bearing.

The investment is relational. Sit in on sales calls. Read deal post-mortems. When a deal is lost, pick up the phone. Talk to the prospect if you can, but also to the salesperson. Ask what happened. Ask what the prospect said. Ask what they said between the lines. Make it clear you value their intelligence, and they’ll bring you more of it. Ignore them, and they’ll route around you directly to engineering, to the CEO, to whoever will listen.

Win/loss analysis in enterprise is genuinely hard. Government officials often can’t speak beyond what’s been formally released. Regulated industry buyers are prohibited from accepting a dinner or a coffee. But it’s absolutely worth working the CRM, identifying the deals worth investigating, and picking up the phone. If you manage to speak to someone who chose your competitor, that single conversation can be worth more than a quarter’s worth of analytics. Treat it accordingly.

Design partners: the enterprise version of beta testing

Once you’ve triangulated the problem, enterprise solution discovery also works differently from the books. You can’t ship an MVP to millions and measure adoption. Instead, you work with design partners: a handful of customers who co-develop the solution with you, and give you feedback every step of the way.

I did this formally in one role, and it produced the most solid product foundation I’ve ever worked with. The process looked nothing like a B2C beta. We started with deliberately unstructured, open-ended discovery calls, to avoid seeding expectations. Then deeper dives into specific workflows. Then Figma prototypes, each accompanied by a list of questions we wanted answered (not “do you like it?” but “what would you expect to happen here?” and “where does this break your current process?”). Then demos of something still far from the prototype, but working enough to interact with real data and validate direction.

What made it work wasn’t the process. It was the relationship. We had a few people from their side, they had a few from ours, all sitting around trying to collaborate to solve a problem. Not vendor and customer. Partners. Those relationships lasted years and shaped not just that feature but the product’s entire direction.

The danger with design partners, and the navigator analogy holds here, is that a local pilot who guides you into harbour is invaluable for this harbour. Their knowledge of the next port is limited. Design partners give you deep, specific feedback, but their needs may not generalise. Not all user feedback is created equal: expert users push for power features, beginners want hand-holding, and the silent majority tells you nothing at all. You need design partners, AND you need your own chart.

Building conviction, not statistical significance

The hardest part of enterprise discovery isn’t gathering signal. It’s deciding when you have enough to act.

In B2C, you can define statistical significance and let the numbers decide. In enterprise, you’re building conviction from qualitative data and pattern recognition. One data point is better than none, but don’t get anchored on it. Collect customer anecdotes like you’re OCD; make it your obsession. You’ll be rolling them out again and again in stakeholder meetings, in engineering brainstorming, in board strategy reviews, and there is nothing like a real customer story to build empathy and alignment across an organisation.

Triangulation in enterprise isn’t as formal as actual navigation. It’s a constant process of connecting dots, of keeping those apparently unrelated anecdotes in mind until the pattern emerges. The professional services engagement where you felt the rough edge. The three salespeople in different regions hitting the same objection. The RFP that asked for something nobody offers yet. The support call where the real problem was three layers beneath the ticket. You hold these loosely, and one day they click together, and you know with the kind of conviction that no p-value provides where to point the product next.

This is where experience matters most, and where it connects back to building taste. The reps you’ve done, the patterns you’ve seen across multiple products and industries, the instinct that tells you “this is real” versus “this is one loud customer” - that’s product sense, and it’s built from exactly these kinds of qualitative signals, accumulated over years.

Practical navigation

For enterprise PMs doing discovery:

Structure your signal collection. Don’t leave insights scattered across call notes and Slack messages. Build a lightweight system for capturing and tagging discovery insights by source, theme, and confidence level. It doesn’t need to be complex; a shared document with consistent tags will outperform a fancy tool nobody uses. The point is making patterns visible across sources and over time.

Invest in sales relationships. Your sales team is your largest, most distributed discovery network. Build the relationship where they bring you insights, not just feature requests. Sit in on calls. Read the CRM. Make it clear their intelligence shapes the product, and they’ll give you more of it.

Triangulate before committing. One customer asking for something is a data point. Three customers and two salespeople independently describing the same pain is a bearing. The same pattern showing up in RFPs, support tickets, AND customer interviews? That’s a fix. Act on it.

Name your uncertainty. Enterprise discovery never reaches certainty. That’s fine. “We have high conviction from four independent sources” carries more weight than “a customer asked for this.” Being honest about confidence levels earns trust with stakeholders and makes prioritisation transparent. It also makes it easier to course-correct when new signal arrives, because you never pretended you had the final answer.

Uncharted territory

I should note: the principles behind continuous discovery are absolutely sound. You should be constantly talking to customers, collecting evidence, and validating assumptions. The enterprise version isn’t a rejection of those principles. It’s an adaptation to a world where the nature of conversations and the scale of evidence are fundamentally different. Once you sign on to high-end B2B, reality diverges from the textbooks. The principles hold. The practice looks nothing like the diagrams.

The enterprise PM navigates what the military calls VUCA: volatility, uncertainty, complexity, ambiguity. Uncharted territory. Not a known map, but a place you’ve never seen before. There’s fog, and it’s raining, and you can hear but can’t see the competitors over the other side of the hill.

But that’s still better than silence. Silence is the scariest data point in enterprise discovery. It means your customers aren’t talking to you - or worse, they’re talking to someone else. The navigator who’s out in the weather, taking imperfect bearings in rough conditions, is in a better position than the one sitting in port waiting for perfect visibility. It never comes. Set sail. Triangulate as you go. Adjust course when the signal sharpens. That’s the craft. After 20 years, I can tell you it never gets boring.