How to Make Business Decisions Without Enough Data

You can't eliminate uncertainty, but you can learn to make good decisions even when the picture is incomplete.

When information is missing, many businesses put decisions on hold and wait for more data. The trouble is that markets move faster than statistics accumulate. A competitor launches a new service, a platform changes its algorithm, customers drift to another channel, and meanwhile your decision is still "under review". In practice, postponing is also a decision, and sometimes it is the most expensive one of all.

In this guide we cover:

• the difference between risk and uncertainty, and why it changes the way you think;
• the situations where businesses most often run short of data;
• three frameworks that help you decide: one-way and two-way doors, the pre-mortem and the Three Horizons;
• how to turn a hypothesis into data with small, fast experiments;
• how to present a decision to your team and leadership when there is no hard proof yet;
• which mistakes to avoid.

Risk vs uncertainty: why the old methods no longer work

Uncertainty in business is not just a missing number in a report. Usually it is a combination of several things: there is no historical data, past experience doesn't fit the new situation, expertise in the area is lacking, and you can't estimate, even roughly, what the outcome of a decision will be.

Risk: when you can estimate the odds

Risk comes with predictable scenarios. You can ask how many customers cancel a subscription on average, what a lead costs in a campaign that has been running for six months, or what happens to your website during peak season. You have numbers from the past, and you can build a forecast and a fallback plan on top of them. Tools such as a risk matrix work perfectly in exactly these situations.

Uncertainty: when most of the factors are unknown

Under uncertainty you have nothing to compare against. It's unclear whether there is any demand at all, what the customer is willing to pay and how competitors will react. Standard forecasting models simply don't work, because they have nothing to lean on.

Why old data is losing its value

Uncertainty has always been part of doing business. But in recent years it has become the permanent background: regulatory change, economic swings, energy and supply chain shocks, and new technologies such as AI that reshape entire industries within months. Many companies know the feeling well: one quarter everything runs as usual, and the next one customers disappear or behave completely differently.

In that environment, historical data loses its predictive power. So the logic of decision-making changes too: instead of trying to remove uncertainty, you learn to operate inside it. You work with hypotheses, think in scenarios and move forward in short iterations that produce first results and first conclusions.

When and why businesses get stuck without data

A lack of data happens far more often than it seems. It isn't limited to rare strategic choices. Uncertainty shows up in day-to-day work as well: when you launch a new service, choose a marketing channel or decide whether to rebuild your website. These are the four most common situations.

Launching a new product, service or feature

You have no historical data to rely on. You don't know how big demand will be, whether customers will pay or whether the investment will pay back. It's also hard to predict how the new service will affect the business as a whole. So you work with assumptions: you build hypotheses about unit economics, track indirect signals and learn from what works in adjacent markets. We cover ways to test demand before you invest in our guide on how to validate product demand.

Entering a new market or reaching a new audience

Here almost everything changes: your go-to-market strategy, your pricing model and your customers' needs. For example, a firm that has worked with large enterprises for years and now targets small businesses is effectively entering a new market. The decision-makers are different, the priorities are different, and what one customer values the other doesn't care about. Expanding into another country or another language market also means rethinking your messaging, channels and pricing from scratch. The key tool here is thorough market and audience research that gathers data step by step and sharpens your hypotheses. If you haven't defined your audience yet, start with our guide to defining a target audience.

Sudden changes in the external environment

Sometimes information is missing because of factors outside your control: a regulatory reform, a platform policy change, economic turbulence or a technology leap. What worked yesterday no longer guarantees results today. Search is a familiar example: a growing share of users get answers directly from AI engines, and the rules of organic visibility are changing on the fly. In a situation like this you decide based on the current context and a handful of possible scenarios, not on last year's report.

Strategic crossroads

Strategic shifts are rare, but their impact is large: changing your business model, entering a new segment, a merger or an acquisition. Decisions like these require analysis and scenario planning, yet ready-made data almost never exists, precisely because every such situation is unique.

A well-known example is Discord. The service started as a communication tool for gamers, and over time the company saw potential in a much wider audience and began positioning itself as a home for communities of every kind. That move meant rethinking both the product and the business model, with no certainty that it would work.

In all four situations, waiting for "enough data" usually means not deciding at all. That's why the key skill is acting on the minimum sufficient evidence, and designing the path so that it generates the data you're missing.

Framework 1: one-way door or two-way door

When data is missing, relying on gut feeling alone is a bad idea. Simple decision-making models help you structure the situation and act deliberately. Let's start with the simplest tool, one you can use almost every day.

The one-way door / two-way door model became widely known thanks to Jeff Bezos, the founder of Amazon, who described it in his letters to shareholders. The idea: first you classify the decision, and only then do you evaluate it.

How to classify a decision

• A two-way door is a reversible decision. If you get it wrong, you can walk back through and try a different approach. Examples: changing a headline on a landing page, testing a new audience in a campaign, adding a WhatsApp button to your website.
• A one-way door is a decision that is almost impossible to reverse and where mistakes are costly. Examples: an annual contract with a supplier, moving the entire business to a new CRM, a full rebrand or a major development investment.

Classification tells you how much time and data you really need. For a reversible decision you can lower the bar for information and move fast. For an irreversible one it pays to slow down, test your assumptions and bring in other opinions. The common mistake is treating every decision with the same gravity: small things get stuck in endless discussions, and the big decisions still don't get the attention they deserve.

How to turn a one-way door into a two-way door

The most valuable trick in this model is not just classifying decisions but changing their structure. Many decisions that look irreversible can be broken into reversible steps:

• launch the new service as an extra option alongside the existing one, not as a replacement;
• start with a basic version that has the minimum set of features;
• test it on a limited audience: first employees or loyal customers, then a share of new customers;
• sign up with a supplier for a trial period instead of a full year;
• define success metrics and an exit point in advance.

Example: online booking instead of phone calls

Imagine a clinic or a fitness studio considering a move from phone bookings to booking through the website. If you move every client over at once and drop the phone line, that's a one-way door: long-standing clients may get confused, and the workload on your team spikes at exactly the wrong moment. If you add online booking as an extra option and spend a month or two watching how many clients choose it, how many cancellations there are and whether phone calls go down, the decision becomes reversible. At the end of the period you have real data, not just a hunch.

Framework 2: the pre-mortem, imagine it has already failed

Anyone who works on projects, especially in tech teams, knows the post-mortem: an analysis of an incident or failure after it has happened. The pre-mortem works the other way round. The method was described by psychologist Gary Klein: the team imagines in advance that the project has failed and tries to work out what led to it.

The central question is simple: "Six months have passed and the project has failed. What exactly went wrong?"

Why it works

When a team is excited about an idea and sure it will succeed, attention narrows to the positive scenario and some factors stay out of the picture. Framing it as "it has already failed" lets people voice concerns without looking as if they oppose the idea. That surfaces blind spots: organisational constraints, wrong assumptions about the customer, missing resources or a misreading of the market.

How to run a pre-mortem in one hour

1. Present the plan in a few minutes: what you're launching, for whom and when.

2. Declare the failure. "Six months have passed. The project has failed completely."

3. Everyone writes on their own every reason they can think of, with no discussion. Ten minutes is enough.

4. Collect the lists and go round the room reading one reason from each participant until the lists run out.

5. Group and rank: which reasons are most likely and which would hurt the most.

6. Define actions: for each of the top three or four risks, agree what you'll do right now to reduce it, and who owns it.

You can run the exercise alone, too, for example before launching a new website or a large campaign. In a team it becomes a structured brainstorm where quieter voices get heard.

Example: launching a new website

Typical reasons that come up in a pre-mortem before a website launch: 301 redirects from the old URLs weren't set up and Google traffic dropped; the form wasn't connected to the CRM and leads were lost; the mobile version was only checked on a desktop; the content wasn't ready in time. Each of these has a simple fix you can make before launch, and that is exactly the value of the exercise. You'll find a structured pre-launch checklist in our guide to website QA testing.

Framework 3: the Three Horizons for different levels of uncertainty

The Three Horizons framework is used for strategic planning and helps you handle decisions of different scale. The model was introduced by McKinsey consultants in the late 1990s, and it divides a company's development into three horizons.

Horizon 1: the present

These are the areas that generate today's revenue and keep the business stable: core services, regular customers, channels that work. Here you usually have enough data, so decisions are driven by analytics and operational metrics. The goal is to optimise, cut manual work and reduce errors.

Horizon 2: emerging areas

This is where changes to existing processes and new growth points live: an add-on service for existing customers, a new sales channel, automation that changes how you work. You have partial data, and you decide using a mix of numbers and hypotheses.

Horizon 3: the future

Experiments, new markets and entirely new business models. There is almost no data here, and you act on hypotheses and strategic assumptions. It's the riskiest horizon, but it is often where the next sources of growth are born.

Why it matters to keep the horizons separate

When you don't separate them, things get distorted. Some businesses focus only on the present and stop investing in the future, until the market shifts and they discover they have nothing new to offer. Others fall in love with innovation and put the stability of the core business at risk. Separating the horizons lets you decide upfront how much time and budget each one gets, and measure each with the right tools: in Horizon 1 you measure profitability, in Horizon 3 you measure how fast you learn.

Example: a small service business

A small insurance brokerage could split its work like this. Horizon 1: tidy up how website leads are handled and connect them to the CRM so that no enquiry gets lost. Horizon 2: add a WhatsApp bot that answers common questions and collects details before a call. Horizon 3: test whether existing clients would buy an entirely new digital service. Each horizon has different goals and a different level of risk. You can set measurable goals for each horizon with the OKR method.

Turning a hypothesis into data: small, fast experiments

The three frameworks help you decide what to test and how carefully. But the only way to genuinely reduce uncertainty is to generate your own data. The good news: in 2026 that is cheaper and faster than ever.

Write a hypothesis that can be proven wrong

A good hypothesis states who, what and how you'll know. Not "customers will want a subscription service", but "at least 5% of visitors to the service page will leave their details for a monthly subscription within four weeks". You set the threshold in advance, based on what makes sense for your business, and you don't adjust it afterwards to fit the result.

Tools for fast experiments

• A test landing page. Describe the new service, add a form or a "Tell me more" button and send traffic to it. A page like this can be built in Webflow within a few days.
• A small paid campaign. A limited budget on Meta or Google shows whether the message interests anyone at all, and which message performs better.
• A/B testing. Two versions of a headline, an offer or a price, and you check which one brings more enquiries. Wait for a reasonable amount of data and don't declare a winner after two days.
• Customer conversations. Five to ten short conversations often reveal more than any report. Ask about past behaviour, not future intentions.
• "Wizard of Oz". Before you build an automation, run the process manually behind the scenes and check whether customers actually use it.

Define metrics and a stopping point upfront

Before the experiment starts, agree on three things: what you're measuring, how long the experiment runs, and what counts as success or failure. Without that, every result will look "promising" and the experiment will run forever. To get data in without manual work, connect your forms, CRM and analytics from the start. That's exactly what we do in our automation and integration service. If your forms or analytics collect personal data from visitors in the EU, make sure your consent and privacy setup follow GDPR and the ePrivacy rules (this is a general note, not legal advice).

How to present a decision to your team and leadership without enough data

A lack of data makes every initiative vulnerable to criticism. When analytics are thin, ideas come across as risky or unfounded. So what you decide matters, but so does how you explain the decision.

Speak in hypotheses, not statements

When data is scarce, frame the initiative as an assumption you can test. It lowers resistance and moves the discussion somewhere constructive: instead of arguing whether the decision is right or wrong, the team discusses which data would confirm or disprove it.

Break big decisions into short iterations

Long initiatives with unclear outcomes breed distrust. It works better to break them into small steps: first test the core assumptions, then scale what worked. You get feedback faster and lower the risk of big mistakes. A project roadmap with clear milestones helps everyone see where you are.

Provide a strategic frame

In any situation, people expect clarity from a leader. The team needs to understand where the business is heading, what the goals are for the next year or two and how current decisions connect to the strategy. A clear frame helps people see individual experiments as part of a bigger logic rather than random jumps.

Document the decision

Write it down in a few lines: what you decided, based on what, what the hypothesis is, how you'll measure it and when you'll review it. A note like this saves arguments in hindsight and teaches the team to make better decisions next time, because you can see what was known at the moment of the decision.

Common mistakes when deciding under uncertainty

Waiting for data that will never arrive

Some decisions no report will make for you. If the missing information won't accumulate on its own, waiting only adds cost. The right question isn't "Do we have enough data?" but "Which small experiment will give us the most important data point?".

Treating every decision as a one-way door

When every headline change takes three meetings, the business loses speed. Save caution for the irreversible decisions, and give your team the authority to make the reversible ones on their own.

Moving the goalposts after the experiment

If you didn't set a threshold in advance, it's very easy to read any result as a success. That's how projects that should have been stopped long ago keep going.

Falling in love with the idea

Enthusiasm is fuel, but it also hides risks. This is exactly where the pre-mortem helps, along with the voice of someone who wasn't involved in the idea from the start.

Putting all your resources into one horizon

Only Horizon 1 leads to stagnation; only Horizon 3 leads to instability. Even a small business should set aside a little time for experiments while keeping the core stable.

How RJL Studio helps you make data-driven decisions

At RJL Studio we build the infrastructure that turns hypotheses into data: landing pages for testing ideas, forms connected to your CRM, automations and tracking that shows what actually works. When you're weighing a new service, a new website or a new marketing channel, we help you break the decision into small, reversible steps, set up a fast experiment and measure it properly.

Want to test an idea before you commit your whole budget to it? Get in touch and we'll build a testing plan that fits your business.

Frequently asked questions

What is the difference between risk and uncertainty?

With risk you can estimate the odds: there is historical data and there are predictable scenarios, so you can build a fallback plan. With uncertainty most of the factors are unknown and there is nothing to compare against. That's why risk is managed with numbers, while uncertainty is managed with hypotheses and small experiments.

How do I know whether a decision is a one-way door?

Ask how much it would cost to go back if you're wrong: in money, time, reputation and customer trust. If reversing is cheap and quick, it's a two-way door and you can decide fast. If reversing is expensive or impossible, slow down, test your assumptions and try to break the decision into reversible steps.

How long does a pre-mortem take?

For a small team one hour is enough: a few minutes to present the plan, ten minutes of individual writing, a round of sharing, ranking the risks and agreeing on actions. On your own you can run a shorter version of 20 to 30 minutes before any important launch.

Do these frameworks work for a small business?

Yes, perhaps even more so. A small business has no research department and no room for expensive mistakes. Classifying decisions as doors helps you move fast on small things, the pre-mortem prevents launch mistakes, and the Three Horizons remind you to reserve some time for the future without neglecting today's revenue.

What is the cheapest experiment for testing a new idea?

Usually it's a combination of a simple landing page and customer conversations. The page shows whether people are willing to take a step, such as leaving their details, and the conversations explain why. If you already have a customer list, you can start with an email or a message to them before spending anything on advertising, as long as you have consent or another lawful basis to contact them under GDPR.