The template, the value and the cognitive trap to avoid. If you’re building a product, a feature or a new flow, the first question to ask isn’t “how long will this take?” or “what tech stack should I use?” It’s “Why am I confident anyone will care?”

That’s a question SA UX lead Hilde Franzsen forces founders to answer before she lets them touch a keyboard.

Hilde is the Brand and Marketing Director at Inkplot, a Stellenbosch-based creative agency that works across brand strategy, UI/UX, web design and development, and digital marketing. She’s worked with a long list of founders and product teams, and she’s seen what happens when people skip this step: months of work, polished products, and nobody using them.

Here’s her workflow for writing a value hypothesis: the two short sentences that catch you before your own optimism decides for you.

A value hypothesis isn’t a research project; it’s a quick two-sentence statement that forces you to commit, in writing, to a specific claim about who your product helps and how you’ll know if you’re right.

Hilde’s test: “If you can’t articulate your app’s value clearly enough to write it in these two short sentences, then you definitely can’t communicate it in the 50 milliseconds people will use to judge your interface.”

Hilde’s template has three fields: “We believe that [your value proposition] is valuable to [your specific audience]. We will know this is true when we observe [behavioural signal].”

That’s it. The discipline of forcing yourself into this exact structure is the whole point. It stops you from hiding behind vague claims and forces you to commit to something specific you can actually test.

This is where most founders go wrong on their first attempt. They write something like “We believe our AI-powered insights dashboard is valuable to founders.”

That’s not a value proposition. It’s a feature description.

Hilde’s point: “AI-powered is a feature, not a value. Nobody wakes up at 3 am thinking I wish I had more AI-powered insights. They’re much more likely to wake up thinking I have no idea which of my three revenue streams is about to dry up. That’s the value. If you can name the anxiety and not the technology, you’re on the right track.”

Better version: “We believe a weekly plain English summary of their top three revenue risks is valuable to founders running early stage SaaS products.”

“Founders” is not an audience. “Small businesses” is not an audience. “People who want to be more productive” is not an audience.

Your audience needs to be specific enough that you could actually find five of them this week and put your product in front of them. If you can’t, the audience definition isn’t tight enough.

Good examples of audiences specific enough to test:

Each of these you could realistically recruit for a test next week. “South African founders”, you could not, because the surface area is too big to find a representative sample.

This is the field most founders skip or fudge. They write something like “we’ll know it’s true when users like it” or “when we see strong engagement.”

A real behavioural signal is something you could literally watch happen in a room. Hilde’s example: “At least 60% of participants in testing sessions say they would act on a specific recommendation.”

You could sit with five founders, show them the summary, and watch whether they say they’d actually do something with your product/tool. That’s observable. That’s a real signal.

Other behavioural signals that work:

If you can’t imagine watching the signal happen, it’s not specific enough yet.

Once you’ve written your hypothesis, you don’t need a research lab to test it. You need five people who match your specific audience and a willingness to actually listen to what they say.

Show them the value (a description, a mockup, a paper prototype — whatever the simplest version is). Watch their reaction. Listen for the behavioural signal. Don’t coach. Don’t explain. Just observe.

As Hilde puts it: “You could literally sit in a room with five founders, show them the summary, and watch whether they say yes, this I would actually do something with. That is something testable.”

If the signal happens, you’ve validated the hypothesis. If it doesn’t, you just saved yourself months of building something nobody wanted. Either way, you’re ahead.

SA founders typically have less runway and smaller teams than their US or European counterparts. You can’t afford to build for three months and find out at the end. The value hypothesis is the cheapest possible bet you can place on whether the work is worth doing at all, and SA founders need cheap bets more than most.

Secondly, AI has made it dangerously easy to ship without thinking. You can go from a vague idea to a deployed product in an afternoon. That makes the discipline of writing a value hypothesis more important, not less. The cost of building has collapsed. The cost of building the wrong thing has not. Catching yourself before you build is now the highest-leverage thing you can do as a founder.

Hilde’s framing: “Your brain is wired to seek evidence that agrees with you and to quietly ignore evidence that doesn’t. The value hypothesis is the structural habit that catches you before your own optimism makes the decision for you.”

Once you’ve written your value hypothesis, every product decision gets easier and faster.

It takes 30 minutes to write, but it can save you the months of building the wrong thing.

Writing a value hypothesis is one habit from a much longer session. In Hilde’s full Founder Collab masterclass on From Solutioneering to UX: How to Prioritise Designing the Right Things, she walks through her complete system for:

You can get access to Hilde’s full playbook, plus 40+ other masterclasses from South African operators, founders and experts when you join The Founder Collab.

This news first appeared in our 30 April ‘26 edition on Tim Treagus startup lessons building Yazi.

Learn how to get better leads from Google in SA and how to break the ice before your first sales meeting.

Get more SA tech and business news and subscribe to The Open Letter.