How do we gather evidence that actually tests a new venture’s assumptions?

Innovation Management
Corporate Innovation
Innovation
Growth Forge
Strategy

Evidence gathering is the most important part of developing a new venture. Everything before it, the strategy hypothesis, the models and the prioritized assumptions, is a set of beliefs about what might work. Evidence is what tells a team which of those beliefs hold. The work that produces it happens in the real world: conversations with customers, mock-ups put in front of them, letters of intent, pilots. No software does any of that. A model can show which beliefs matter most. Only evidence from outside the building can show whether they are true.

What software can do is make that work more effective and efficient, and its results more useful for what comes next: the next iteration of the strategy and the next investment decision. Evidence gathering often goes wrong not because teams do too little of it, but because nobody is clear which assumption a piece of evidence was meant to test, or whether it answered the question. Evidence that does not map to a stated assumption has not reduced uncertainty, however much of it there is. An earlier post in this series covered how to decide which assumptions to test first. This post is about testing them.

A different kind of project management

Most project management in a large company is about executing a plan. The scope is known, the tasks can be listed, and progress means completing them on time and on budget, not the resulting implications. Early-stage venture work is different. The plan is a set of experiments based on key questions to answer, and progress means learning the answers and new insights, including the ones that say the strategy hypothesis needs to change.

The activities also change from stage to stage. Early on, when the strategy is modeled at low fidelity and the investment at stake is small, the work is conversations, simple mock-ups and desk research. Later, as fidelity and investment rise, it becomes pilots, pre-sales and operational tests. A fixed task template, built for developing a product the company already understands, describes neither. What a venture needs is innovation-specific project management: planning and tracking the activities that have to happen in the real world, organized around what the venture needs to learn.

Plan each test from the assumption

A useful test starts with the assumption, not the method.

Name the assumption (or assumptions) the test is meant to answer, stated specifically enough to be wrong. “Customers want this” is too broad to produce a useful answer. “Operations managers at mid-sized manufacturers will pay a monthly fee for this without an integration into their existing systems” is specific enough that a result can show it is wrong.

Decide in advance what result would change the team’s mind. If no plausible result would, the test is not worth running.

Then choose the method: the fastest and cheapest one that can produce that result with enough confidence for the decision at hand.

Getting the order wrong is common. A team picks a method it knows, a survey or a prototype, and then looks for assumptions it might inform. The result is activity that feels productive and moves no specific belief.

The methods many large companies never consider

When large companies gather evidence for a new venture, they tend to reach for three methods: secondary research, large quantitative surveys and product prototypes. Each has its place. But secondary research describes markets that already exist, which can be helpful to bound or inform quantitative uncertainty ranges in models. Surveys ask people to predict their own behavior, which is not as reliable an indicator as their demonstrated behavior. And prototypes are slow and expensive and mostly answer whether the product can work. For many of the assumptions that matter most, there are faster and cheaper options.

They are grouped here by the question each mostly helps answer, and listed within each group from the fewest resources (time, money and people) to the most.

Desirability: do customers want it?

  • Landing-page tests: describe the offer and measure how many people sign up or ask for pricing.
  • Buying the competing alternative: use the product or workaround customers rely on today, to see what it does well and where it frustrates them.
  • Low-fidelity mock-ups: paper sketches, clickable screens, or a Wizard-of-Oz setup in which customers use what looks like a working product while people do the work behind it. They show how customers react before anything real is built.
  • Concierge tests: deliver the service by hand to a few customers, to learn whether they value the outcome.
  • A minimum viable product: the smallest version of the offering, or even just a critical differentiating feature, that lets the team learn whether its key assumptions hold. It is a learning tool, not the first or stripped down release of the product.

Feasibility: can it be delivered?

  • Structured expert interviews: a series of conversations with people who know the industry, the channel or the technology, run to a consistent guide.
  • Partner and supplier conversations: confirm that the parts of the whole solution the venture would not build itself can actually be sourced, and on what terms.
  • A focused proof of concept: test the one technical question that is most uncertain, rather than building a full prototype.

Viability: will customers pay, and can it make money?

  • Letters of intent: a written, non-binding statement that a customer would buy on stated terms, a stronger signal than interest expressed in an interview.
  • Pre-orders or deposits: asking customers to pay before the product exists, the strongest early signal of demand.

A paid pilot with a single customer is powerful evidence too, but it needs a nearly complete product. It belongs later, once cheaper tests have justified the investment.

Most of these can run in weeks on a small budget, and several produce evidence of what customers actually do rather than what they say they would do. That distinction matters. A commitment, however small, is stronger evidence than an opinion.

A caution on AI

AI tools can make much of this work faster. They are good at widening and speeding up desk research, drafting interview guides and survey questions, and summarizing what a set of interviews revealed. Our position on AI encourages exactly those uses.

They are not a substitute for the real world. Tools that generate synthetic customer personas or simulated survey responses produce answers that sound plausible, because they are drawn from what has already been written, and those can be helpful as a pre-flight check to make sure that the response you get from a survey will actually deliver the answers you’re looking for, but they are not a real market signal. A new venture is testing something that has not happened yet: whether these customers will change what they do for this offering. That can only be observed. Treating an AI-generated answer as market evidence gives a team the feeling of validation without the substance.

Record what the evidence did to the belief

Most tracking records that an activity happened: twelve interviews completed, pilot launched, survey closed. What matters most is what the activity did to the assumption it was meant to test. Did the evidence level rise? Did the range around an uncertain input narrow? Did the assumption turn out to be wrong, and does the strategy need to change as a result? Our post on what-if modeling and the pivot covers what happens when it does.

Recording the evidence level before a test and after it makes that visible. It also exposes the activities that produced a lot of material and moved no belief at all.

Tracking what the evidence did also means reviewing it as it arrives, not saving the analysis for the end. If the first few interviews expose a fatal flaw, running the rest of the planned activities may be pointless. It is better to stop, re-evaluate the hypothesis, explore pivots based on what was learned, and save the remaining resources for that or for other opportunities in the portfolio.

Why the usual tools leave gaps

Project management tools plan tasks, owners and dates. They can track that an experiment ran, but they have no record of which belief it was meant to test or whether it did.

Research repositories and survey platforms store interview notes, recordings and responses, sometimes very well. But the findings are organized by study, not by the assumptions in the strategy, so connecting them back to a decision is left to whoever remembers.

Experiment boards and canvases help a team frame tests around hypotheses, which is the right instinct. But they sit apart from the strategy model and the investment decision, so what is learned on the board does not change the numbers or the review.

What is missing is a plan of evidence-gathering activities tied directly to the prioritized assumptions, with the method, cost and outcome of each, and the results flowing back into the strategy they were meant to test.

There’s an app for that: how Growth Forge® Software addresses this

Evidence Gathering is where that plan lives. It works from the prioritized list in Assumptions Manager: each activity is linked to the assumptions it is meant to test, so every piece of planned work has a stated purpose before it starts.

The team chooses from a library of predefined evidence-gathering methods, from problem interviews and concierge tests to pricing studies and pilot programs, each described with its relative cost and time and the kinds of questions and stages it suits. Custom activities can be defined for anything the library does not cover. Each activity has an owner and an estimated budget, headcount and duration, and tracks what was actually spent against them, along with its progress.

When an assumption is linked to an activity, its evidence level at that point is recorded. As results come in, the team updates the evidence level and notes what it learned, so the change each activity produced is visible alongside what it cost. Those updates flow back to the assumptions, to the ranges in the financial models, and to Strategy Evaluation, so the next investment decision reflects what has actually been learned. For interviews, the problem and solution interview tools provide structured scripts and a place to capture what customers said.

The tool does not run a single experiment. What it contributes is the connection between the work done in the real world and the beliefs the strategy depends on.

What no software can settle

Evidence comes from the real world, and there is no substitute for engaging with it directly. The team needs to be in front of customers, partners and the functions it depends on, early and often, watching what people do as well as listening to what they say. Much of what matters most shows up only there: the workaround nobody mentioned in a survey, the approval step a buyer forgot to describe, the partner who is enthusiastic in a meeting and slow to follow up. Reports from others, however good, are a step removed.

Getting that engagement depends on things only the organization controls.

The first is access. Customer relationships in a large company usually belong to the sales organization, which has good reasons to protect them and little reason to let a venture team test an unproven offering on its best accounts. Legal may restrict pre-selling, letters of intent or pilots with outside customers. Procurement may have no way to buy a competitor’s product for evaluation. These are the same frictions our post on whether the organization will reject a new venture describes, and they can stop evidence gathering before it starts.

The second is honesty about the result. Naming in advance what would change the team’s mind, and then honoring it when the evidence arrives, is a cultural commitment. Where the answer is known before the experiment runs, the tooling is decoration.

Both are set by leadership: whether venture teams get access to customers and the freedom to test with them, and whether a result that contradicts the plan is treated as progress. An outside team with experience running this kind of work across many companies can help design tests that get useful answers quickly, and can help the innovation leader work through the internal barriers, but opening those doors has to be done from inside.

Putting it into practice

Before a venture’s next round of evidence gathering starts, a few steps keep it pointed at what matters. For each priority assumption, write down what result would change the team’s mind before choosing a method. Consider the faster, cheaper methods, especially those that measure what customers do rather than what they say. Record the evidence level before and after each activity, and look hard at the activities that moved nothing. These steps can be worked through with a framework, facilitated by an experienced consulting firm, or with purpose-built software such as Growth Forge, where Evidence Gathering provides the guidance and the tracking. The evidence itself still has to come from the real world. The rest is making sure it answers the questions the decision depends on.

If the harder question is whether your venture teams can get access to customers and permission to test with them, our Innovation Capability Assessment is a short scored read on where that stands.

Next in this series: making decisions that actually decide something.

BRI Associates helps companies grow by drawing on decades of practitioner experience in corporate innovation and new business development — practitioners, not pundits or academics — through direct consulting, training workshops, and Growth Forge® Software, built for the unique requirements of corporate innovation and growth organizations.

Curious where your organization's innovation capability actually stands? Take BRI's free Innovation Capability Assessment — a short diagnostic that names your capability gaps and where to focus."

New Business Growth & Innovation are Hard.
We Can Help!

Connect with us today. We're happy to spend some time with you to understand your needs and explore how we can help.