How to Build a Growth Hypothesis Tree: From Evidence to Experiments

Growth teams rarely suffer from a shortage of ideas. There is always another campaign to launch, audience to target, landing page to improve or automation to build.

The harder question is whether those activities will contribute to the outcomes the business actually needs.

A growth hypothesis tree provides a structured way to connect business objectives with the assumptions, ideas and experiments that might help achieve them. Instead of collecting disconnected tactics, teams can see why each initiative exists, what it is expected to change and what evidence would justify continuing.

At NBT, we use this logic within our broader Growth-Driven Marketing approach. Strategic and annual objectives inform quarterly OKRs. Those quarterly outcomes shape growth roadmaps, where hypotheses are translated into specific tests and refined through a Test & Learn process.

The goal is not to run more experiments. It is to make better growth decisions.

1. Start with an objective, not a list of ideas

A hypothesis tree should begin with a meaningful commercial objective.

Consider an illustrative B2B software company with the strategic ambition to reach £10 million in annual recurring revenue.

That ambition may translate into an annual objective to accelerate recurring revenue growth. The leadership team then identifies the outcomes that need attention in the coming quarter.

For example:

• Acquire qualified opportunities from new target accounts.

• Increase expansion revenue from existing customers.

These are different growth challenges. The first might depend on reaching the right buyers and improving qualification. The second might depend on identifying unmet customer needs and communicating the value of additional capabilities.

Both contribute to revenue growth, but each requires different assumptions, evidence and tests.

This distinction matters because an activity should be justified by the outcome it is intended to influence, not by its popularity or ease of execution.

2. Translate quarterly OKRs into growth opportunities

Within NBT's approach, quarterly OKRs establish what the team wants to achieve and how progress will be measured.

The quarterly growth roadmap then addresses a different question:

What might we do to achieve those results?

Suppose one quarterly Key Result concerns generating more sales-accepted opportunities from a priority customer segment.

Potential growth opportunities might include reaching decision-makers who are currently underrepresented in the pipeline, improving the relevance of campaign messaging, or reducing friction between initial enquiry and sales qualification.

At this stage, these are not yet validated solutions. They are areas where the team believes intervention could improve the result.

A useful hypothesis tree makes this connection explicit.

The commercial objective and Key Result establish direction. The hypothesis and experiment establish what the team needs to learn.

3. Build the hypothesis tree from evidence

Once a quarterly outcome has been defined, examine the evidence that explains current performance.

For a qualified-pipeline objective, useful inputs might include CRM qualification records, sales rejection reasons, campaign performance by customer segment and interviews with recent buyers.

Imagine the evidence suggests that enquiries from certain campaigns rarely progress because they come from unsuitable companies.

There are several possible explanations:

Hypothesis A: Targeting problem. Campaigns are reaching companies outside the intended customer profile.

Hypothesis B: Messaging problem. The campaigns reach suitable companies but attract the wrong stakeholders or create inaccurate expectations.

Hypothesis C: Qualification problem. Suitable enquiries are being rejected because the handoff process does not capture the information Sales needs.

These branches require different responses. Increasing advertising spend could make the situation worse if the underlying issue is targeting. Rebuilding the landing page may achieve little if the actual problem lies in CRM qualification rules.

The tree should therefore distinguish observed evidence from interpretations and proposed interventions.

A team does not need perfect data to begin. It does need to be clear about what is known, what is assumed and what remains uncertain.

4. Turn assumptions into testable hypotheses

A useful growth hypothesis identifies a proposed cause-and-effect relationship.

For example:

We believe that showing industry-specific implementation evidence to operations directors in our priority segment will improve the quality of enquiries, because their main uncertainty concerns delivery feasibility.

This is stronger than:

Let's run a LinkedIn campaign.

The first statement explains the audience, proposed mechanism and expected business effect. The second only names an activity.

Before accepting a hypothesis into the quarterly roadmap, ask whether the assumption is supported by evidence, whether it matters to a Key Result, and whether a practical test can distinguish the proposed explanation from alternatives.

Some problems do not require an experiment. If tracking is broken or an agreed CRM handoff is not happening, the appropriate response may be to repair the process.

Test & Learn should address uncertainty, not turn every operational fix into an experiment.

5. Design experiments around a clear learning objective

In NBT's Growth-Driven Marketing methodology, Test & Learn includes three connected activities: defining the test, establishing key metrics, and using the evidence for iteration and optimisation.

The experiment should specify what will change and what the team expects to learn before execution begins.

Consider a hypothetical test comparing LinkedIn messaging for two buyer groups.

The example deliberately does not prescribe a budget, duration or minimum number of leads. Those choices depend on audience size, expected response volume, decision cycle and the reliability of the available data.

Clicks and enquiries can help explain performance, but they do not automatically establish commercial success. A campaign that generates fewer enquiries may still be more valuable if a substantially higher proportion is suitable for the business.

For experiments with limited sample sizes or imperfect comparisons, findings should be described with appropriate uncertainty rather than presented as definitive causal proof.

6. Prioritise experiments within the quarterly roadmap

A hypothesis tree can contain more potential tests than the team has the capacity to execute.

Prioritisation should consider how strongly each test connects to the quarterly objective, the importance of the underlying assumption, the quality of existing evidence, and the resources needed to obtain useful new information.

It should also account for dependencies.

If CRM definitions are inconsistent, testing a new source of qualified leads may produce ambiguous results. Fixing the measurement system first could be more valuable than launching another campaign.

For each selected experiment, the roadmap should identify an owner, the hypothesis being investigated, the intended audience or process, the evidence required and the next decision point.

The roadmap is not simply a calendar of activities. It is a plan for changing outcomes and reducing uncertainty.

7. Convert test results into decisions

A completed experiment is not automatically a successful experiment.

The relevant question is whether the evidence improves the team's understanding of what will achieve the quarterly objective.

A test may support the original hypothesis, contradict it or remain inconclusive.

If persona-specific messaging improves enquiry quality, the team may decide to extend it to other suitable segments. If it increases engagement but produces no useful change in sales acceptance, the initial explanation may need revision.

An inconclusive result should not be automatically treated as a failure. The test may have been too small, the measurement unreliable or the buying cycle too long for the selected observation window.

Record what happened, what can reasonably be concluded, what remains uncertain and what the team will do next.

This creates an evidence trail rather than a collection of disconnected campaign reports.

8. Feed learning back into the OKRs and roadmap

The hypothesis tree should change as evidence accumulates.

Some branches will be strengthened. Others will be rejected or replaced. New constraints may become visible as previous ones are addressed.

The feedback also needs to flow upwards.

For example, a quarterly campaign experiment might reveal that demand exists but the sales process cannot accommodate the complexity of the intended customer segment.

The right response may be to revise the roadmap, improve the commercial process or reconsider the assumptions behind the quarterly plan.

Not every test result justifies changing an OKR. The objective should provide enough stability to evaluate progress, while the initiatives beneath it remain adaptable.

However, when credible evidence shows that the chosen objective or measure no longer reflects the underlying business priority, leadership should explicitly review it rather than continue with a misleading target.

9. Connect Test & Learn to the full growth system

A growth hypothesis tree is most useful when it operates within a connected management framework.

NBT's Growth-Driven Marketing approach combines four elements:

Growth Loop provides the continuing process of diagnosis, strategy development, planning and optimisation.

Full-Funnel Management connects opportunities and performance constraints across Reach / Attract, Engage / Convert, Nurture / Close, and Retain / Grow.

OKR-Based Planning translates strategic direction into annual and quarterly outcomes.

Test & Learn-Based Iteration turns the prioritised opportunities within the quarterly roadmap into experiments and evidence-based decisions.

The hypothesis tree sits primarily within the Test & Learn component. It draws direction from the OKRs, evidence from the customer journey and operational insight from the Growth Loop.

This prevents experimentation from becoming detached from commercial strategy.

10. A practical hypothesis-tree checklist

Before an experiment enters the growth roadmap, the team should be able to answer:

1. Which quarterly Key Result is this intended to influence?

2. What evidence suggests there is an opportunity or constraint?

3. What exactly do we believe will change, and why?

4. What alternative explanation could also fit the evidence?

5. What is the smallest credible way to investigate the assumption?

6. Which metrics and observations will inform the decision?

7. Who owns execution, and who owns the commercial outcome?

8. What will we do if the hypothesis is supported, contradicted or remains unresolved?

The purpose is not to create a complicated testing bureaucracy. It is to ensure that every meaningful experiment has a reason, a learning objective and a clear connection to business priorities.

Conclusion: Make learning part of growth execution

A growth hypothesis tree connects the company's ambitions with the practical decisions teams make every week.

It starts with a desired outcome, identifies the assumptions that matter, translates those assumptions into testable ideas and uses new evidence to refine the quarterly roadmap.

Within NBT's Test & Learn-Based Iteration approach, the aim is not to maximise the number of experiments completed. It is to improve the quality of decisions that move the business towards its goals.

The strongest growth teams can explain not only what they are doing, but why they believe it will work, what would prove otherwise and how the result will change their next decision.