How to uncover what your customers actually think (without a £100k research budget)

Hi 👋

Something that still surprises me after 20+ years in marketing is that most businesses make critical decisions based on what they think their customers want, rather than what customers are actually telling them.

I see it constantly. Teams sitting in meeting rooms debating messaging, copying competitors, or worse - letting the loudest voice in the room decide. Meanwhile, their actual customers are out there, literally spelling out what would make them buy.

My expensive wake-up call

When I was at Skype, Starling Bank and Yolt, we had proper research budgets. Focus groups, surveys, interviews, the works. It gave us confidence in our decisions because we knew exactly what our audience wanted.

Then I started working with startups. No budget for fancy research. At first, I tried recreating the same process on a shoestring. The a-ha moments were thin on the ground..

That's when everything changed.

Your customers are already telling you everything (but you still need to talk to them)

Here's what I discovered: your audience is having conversations about their problems right now - on LinkedIn, Reddit, in G2 reviews, on industry forums. They're using their own words to describe their pain points, their ideal solutions, what they've tried that didn't work.

This isn't theoretical market research. It's real buyer language in the wild.

Previously, this used to means days if not weeks of work, individually going through each review and online conversation. But thankfully, now there’s a quicker way.

But here's the thing before we go through this - and this is really important - this data doesn't replace talking to actual customers. Nothing beats a real conversation for understanding the why behind the what.

What this approach does brilliantly is help you prepare for those conversations. It tells you what questions to ask. It helps you spot patterns across hundreds of customers that you'd never see from just 10 interviews. And it validates what you're hearing in those one-to-one chats.

The magic happens when you combine both

We now use this scraped data to:

• Identify the right customers to interview
• Know exactly what to dig into during conversations
• Validate whether what we heard in 5 interviews applies to 500 customers
• Spot gaps between what customers say publicly and what they share privately

The results? One client met their annual customer target in 7 weeks. Another saw engagement rates triple. Not because we stopped talking to customers, but because we got smarter about what to talk to them about.

Making it work for small teams

The companies I work with - typically post-seed to Series B - need results fast but can't afford to get it wrong. They definitely don't have time for 12-month research projects.

That's why we've developed a lean, repeatable process that any team can use. It's the same approach that helped Hassan & Yasin move from "marketing gut feel" to evidence-based decisions they could actually action.

Here's our 5-step framework:

1. Find your audience's watering holes Where do they actually hang out online? Industry forums, review sites, specialist subreddits. Use AI to help identify these spaces quickly

2. Gather the raw data We use simple scrapers to pull reviews and discussions. Nothing fancy - Google Colab and basic prompts do the job

3. Set up your AI properly Feed ChatGPT or Claude your company context, the Jobs-to-be-Done framework, and be crystal clear about what you're looking for

4. Analyse with a critical eye Get AI to quantify pain points and pull exact quotes. But always verify - ask "Did you analyse every row?" and "What assumptions did you make?"

5. Validate with real conversations Use these insights to have better customer interviews. Ask about the specific pain points you've identified. Dig into the emotions behind the patterns you're seeing

A real example from our work

When we did this for a B2B SaaS client recently, we discovered through review analysis that customers kept mentioning wanting "something that just works without me having to think about it."

But it was only through follow-up interviews that we understood WHY - they'd all been burned by complex implementations before. That emotional context changed everything about how we positioned the simplicity message.

One messaging change. 3x improvement in demo bookings.

The perfect research stack for small teams

Here's what works:

• Weekly: 2 customer conversations (30 mins each)
• Monthly: Scrape and analyse 100+ reviews/discussions
• Quarterly: Run a focused survey based on what you've learned

This gives you constant insight without the massive research budget. The scraped data tells you what's happening at scale. The conversations tell you why. Together, they give you confidence.

Try this tomorrow

Want to see how this works? Here's a quick experiment:

• Pull 50 competitor reviews from G2, Capterra or wherever your customers are
• Identify the top 3 frustrations mentioned
• Schedule calls with 3 of your customers and specifically ask about those frustrations
• Notice how much richer the conversation becomes when you know what to probe on

Once you experience this combination, you'll never go back to either approach alone.

The mindset shift that changes everything

For small teams, comprehensive research feels impossible. But here's what I've learned: you don't need perfect research, you need smart research.

By combining AI-powered analysis of existing data with strategic customer conversations, you get the best of both worlds - scale AND depth, patterns AND stories, what AND why.

As Yasin told me after implementing this: "It's not just the recommendations, it's knowing WHY they're the right ones."

Remember: AI can tell you what 1,000 customers are saying. But only a real conversation can tell you what they're feeling.

What marketing assumptions could you test with this combined approach? Hit reply - I'd love to hear what you discover.

Talk soon,

Lucy 👋

P.S. We're still massive advocates for regular customer conversations - in fact, we think every team should be doing at least 2 per week. This data approach just makes those conversations 10x more valuable!