The Customer-Problem Separation Review
Separate customer, business, and efficiency problems before choosing a solution.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 97%
HubSpot's growth teams force specificity before discussing solutions. Instead of writing a generic problem statement, they label whether it is a customer problem, a business problem, or an efficiency problem. If it is a business problem, the team asks why it has not resolved itself and searches for the upstream customer problem producing the negative business result. That separation reduces the temptation to lift a short-term metric through a customer-hostile change. The review then makes assumptions explicit: if the team acts, what does it predict will happen? Repeated why questions test the justification, while repeated what questions expose the blast radius and domino effects. The result is a problem definition that connects customer value to durable business value before implementation begins.
Origin
Miller explained this as the structure used by the HubSpot growth teams he leads to turn customer obsession from a slogan into language embedded in feature and experiment documentation.
Core principles
- 01An unqualified problem statement hides whose outcome is being optimized.
- 02A business symptom often has an upstream customer cause.
- 03Specific language creates daylight between customer value and business pressure.
- 04Every proposed change has assumptions and a downstream blast radius.
How to run it
- 1
Classify the problem
State explicitly whether the issue belongs to the customer, the business, or operational efficiency. Split a mixed statement into separate problems rather than allowing one label to cover all three.
Pro tip Make the qualifier mandatory in briefs and experiment documents.
Watch out A generic 'problem' heading encourages teams to treat a business metric as the customer's need.
- 2
Trace the upstream cause
For a business problem, ask why it has not solved itself. Identify the customer friction, unmet need, or behavior that creates the downstream business symptom.
Pro tip Write the customer and business statements side by side to preserve the distinction.
Watch out Correlation between a customer behavior and a business outcome does not yet explain the cause.
- 3
Expose assumptions
List what must be true for the proposed intervention to help both the customer and the business. State the predicted first-order and derivative effects before building.
Pro tip Turn assumptions into observable signals wherever possible.
Watch out Hidden assumptions make a successful metric movement easy to misinterpret.
- 4
Interrogate why and what
Keep asking why to challenge the rationale and what to reveal second-order effects. Continue until the team can describe both the causal logic and the likely domino effects.
Pro tip Invite functions downstream from the change to test the blast-radius map.
Watch out Stopping at the immediate metric ignores effects on trust, support, retention, and later conversion.
- 5
Choose the durable solution
Select the intervention that addresses the customer cause while improving the business outcome over the relevant time horizon. Decline a quick lift that depends on lasting customer hostility.
Pro tip Evaluate the decision two to four years out when short-term pressure distorts the trade-off.
Watch out Customer-hostile gains eventually catch up unless the company faces no competition at all.
In the wild
A team could label low checkout conversion as a business problem and add pressure that lifts near-term revenue. The review separates that symptom from the customer problem, such as unclear value or checkout friction, then maps the assumptions and downstream effects of each remedy. The selected change must improve the customer's ability to decide, not merely force the decision.
→ The team pursues durable conversion rather than a temporary gain that damages trust or retention.
Common mistakes
Writing an unqualified problem
Without naming whose problem it is, the team can quietly substitute a business goal for a customer need.
Optimizing only the first-order metric
A change can lift immediate revenue while worsening trust, support load, retention, or future demand.
Leaving predictions implicit
If assumptions are not recorded before the intervention, the result cannot cleanly update the team's beliefs.
Is it for you?
Best for
Product and growth teams reviewing feature briefs, experiment plans, pricing changes, or funnel interventions.
Not ideal for
Emergency operational incidents where the immediate cause and required containment are already unambiguous.
From the episode
Relentless curiosity, radical accountability, and HubSpot’s winning growth formula
Christopher Miller (VP of Product, Growth and AI)