Data scarcity in B2B: how to make good decisions when there isn’t enough data, by Lorena Salgado
We’ve become used to wearing a watch that tells us our heart rate, how much we’ve walked and whether we’ve slept well. Our phone tells us how much screen time we’ve had and what we’ve spent it on. We also know how much time we’ve spent reading over the past week, month or year. When any of these metrics get worse, it feels like we’re doing something wrong.
When we pay so much attention to numbers in our daily lives, how could we not do the same at work? On top of that, every role is now expected to be “data-driven”, because what isn’t measured can’t be improved (we’ll see later why that isn’t always true).
When it comes to B2B digital products, the amount and quality of data are rarely enough to make decisions. And we can’t wait six months for a dashboard to start talking. The key to making better decisions in this context is collaboration.
Why we don’t have data in B2B and why that’s fine
I’ll admit it: when I moved from B2C, with its fireworks-filled dashboards and tens of thousands of data points per metric, to B2B, where sometimes there isn’t even enough data to draw a straight line , I went through several phases of panic, uncertainty and, eventually, certainty. Certainty that I was doing things wrong.
A typical B2B chart that tells you nothing.
Over time, I realised that data scarcity is normal in B2B. We work with a handful of accounts instead of thousands or millions of users, events happen infrequently (annual renewals, long periods of inactivity), and A/B testing is rarely clean (there’s not enough volume or comparable traffic, and client-level variability skews everything).
All of this is fine, as long as we know how to spot signals.
Signals are concrete, repeated pieces of evidence with context that help guide decisions, without promising mathematical precision. Let’s look at how to use them with common sense.
How to work with signals
The hardest part of paying attention to signals is that they’re not as objective as data. It’s easy to get lost in the noise and fail to distinguish between what really needs our attention and what doesn’t.
I find it helpful to have clear criteria to decide whether a signal is objective and actionable — or whether it should simply be ignored.
- Who it comes from: feedback from an occasional user doesn’t carry the same weight as feedback from a decision-maker. Nor does input from your ICP compared to an outlier account. The source matters a lot, so it’s always useful to record role and segment.
- Minimum repetition: if the same problem or idea shows up across different roles, appears in multiple conversations, or comes from different accounts, it’s worth paying attention. A single anecdote is not a signal.
- Impact: what happens if we don’t address it in the next 90 days? Does it block work, waste time or money, create risk or cause embarrassment for the user? If the answer is “nothing important”, it’s probably not a priority.
- Verifiable evidence: tickets, emails, notes, videos, screenshots or any other tangible evidence make a signal far stronger than “someone told me”.
By considering these four aspects, we can decide which signals to follow and which to ignore. If we want to add a bit more rigour, we can score each aspect and use that to rank initiatives.
There are a couple of other things to bear in mind when working with signals:
- A signal describes the problem or task, not the solution (“we need X”). If it jumps straight to a solution, we’re missing information about what actually needs solving. Grouping by tasks or scenarios is always more useful than grouping by features.
- Signals can, and often should, be linked to a metric we want to improve. Tagging each signal with the lever it affects (e.g. activation, churn, sales win/loss, adoption by role) helps guide focus and prioritisation. It’s not a dashboard, but it does the job.
At this point, you might be thinking: this all sounds great, but where do these signals come from? Let’s get into that.
Where signals come from and how to capture them
Signals obviously come from talking to customers, prospects and so on. But product teams can’t be everywhere. That’s why collaboration matters, properly set up collaboration. Each team sees a different slice of reality, and if we collect those slices using the same language, they start to fit together.
Here’s what each team typically captures that can be useful:
- Customer Success: quarterly reviews, renewals, upsells. Particularly valuable are moments of user embarrassment (fear of making mistakes, delaying tasks) and recurring workarounds.
- Sales: objections that block deals, phrases that keep coming up in demos, immediate win/loss insights (“we win when we show X”, “we lose when Y is missing”).
- Marketing: responses to campaigns and webinars, social media comments, competitor claims that raise doubts, internal website searches.
- Support: top ticket categories, spikes in questions after a release, copy errors or confusing flows that require tutorials.
- Operations: manual steps that always break, intermediate spreadsheets, heavy integrations, permission mappings customers don’t understand.
- RevOps / Finance: recurring reasons for churn or downsells, discounts granted due to product gaps.
For this information to reach Product without getting lost, collaboration needs to be frictionless: a single place to log signals, a shared format, and less than two minutes per entry. A Signals Bank, where teams always record the basics (who/role, task, evidence and impact), turns scattered opinions into comparable entries with traceability and visibility for everyone. What used to be noise becomes actionable input that feeds decisions and the roadmap.
If you want a Notion template for a Signals Bank, you can download it here. The first five columns are filled in by the team reporting the signal; the rest are completed by Product.
Signals Bank template in Notion
In the end, the quality of our decisions as a product team depends less on having perfect data and more on seeing the whole picture. Customer Success sees what happens after the sale, Sales sees why people don’t buy, Marketing sees which messages resonate, Support sees where things get stuck, and Operations sees what breaks internally. If we only talk to one of these teams, we’re making decisions based on a very small slice of reality.
Collaboration here isn’t a nice-to-have, it’s the only way signals can be representative. Without it, we’re still deciding based on isolated anecdotes… even if we call them “insights”.
How to encourage collaboration without adding yet another tool
For information to reach Product without getting lost, collaboration has to be almost invisible. If sharing a signal takes more than two minutes, it won’t happen. The Signals Bank helps, but it needs to be as easy and rewarding to use as possible.
A few ideas that work well:
- Appoint a “champion” in each team to collect the essentials and upload them regularly.
- Build signals into existing rituals. For example, always spend five minutes at the end of a meeting deciding what goes into the Signals Bank that week.
- Close the loop by showing what happened with the input. Nothing kills collaboration faster than feeling like you’re throwing things into a black hole. Sharing openly which signals influenced the roadmap, a specific improvement, or even a decision not to act, reinforces that contributing is worthwhile.
- Recognise good signals. Something as simple as “signal of the quarter” in a team meeting, or calling out the person who brought it in. Make it clear that providing useful input is valued as much as closing a ticket or a sale.
The goal is for every team to see that sharing signals is part of doing their job well, not just another bureaucratic process or report to fill in.
Conclusion: making good decisions with what we actually have
Working with small samples is normal in B2B. What we can’t do is keep acting as if magical dashboards will one day appear and make decisions for us. They won’t.
What will help is building a system where signals don’t get lost: clear criteria to filter them, a single place to collect them, and honest collaboration across teams to see the full picture. There will be as much data as there is; but we can have far more signals than we think if everyone knows what to look for and where to put it.
If we manage to get reliable signals and real collaboration, being “data-driven” stops being a pose and starts meaning something genuinely useful: making better decisions, faster, with what’s actually on the table.
Lorena Salgado - Product Manager
Want to know a little more about who wrote this article? 👇
Product Manager with a background in Environmental Sciences and experience in marketing and business strategy. Today, she focuses on making the technology she helps build more meaningful for people and for the world.