Dr. Jonah Tebaa evaluates customer-facing AI by asking what the customer is actually able to do after receiving a response.
That shifts the review away from surface qualities alone. An answer can be relevant, well written, and consistent with policy while still leaving the central customer decision unresolved. Jonah calls the stronger alternative a decision-ready answer: one that removes enough uncertainty for a person to take a useful next step, select another route, or make an informed choice not to continue.
The hidden decision inside an ordinary question
Consider a familiar ecommerce question: “Will this arrive before Friday?”
A general reply might state that standard delivery takes three to five business days. The information is related to the question, but the customer must still calculate whether the deadline is realistic. The answer does not reveal whether the item is available, whether an order cutoff applies, or whether another fulfilment choice would be safer.
A more useful response would assemble those decision-relevant details. For example, it might explain that an order placed before 3 p.m. can use express delivery for an expected Thursday arrival, that standard delivery is possible but not guaranteed by Friday, and that pickup is available earlier.
The point is not the wording itself, and the example is not a universal delivery promise. The point is that the customer can now distinguish among real choices. The response has been designed around the purchase decision rather than around the delivery-policy sentence.
Four details that change the usefulness of the answer
Jonah identifies four elements visible in this example.
First, the answer includes the current fact that affects the choice, such as availability, location, eligibility, or booking status.
Second, it states the condition that changes the outcome. A deadline, cutoff, requirement, or constraint can be more important than a broad policy range.
Third, it names an action the customer can take immediately. Depending on the context, that may be buying, booking, confirming, changing, comparing, or resolving.
Fourth, it presents a realistic alternative when the preferred result is uncertain. An alternative is useful when it gives the customer a safer route, not when it merely adds more information.
Jonah does not present these four elements as an exhaustive standard for every interaction. They are a practical way to inspect questions where a customer’s next decision is easy to overlook.
How leaders can apply the test
The method can be used across support, sales, ecommerce, and digital customer journeys.
A team can begin with ten common or commercially important questions and complete one sentence for each: “After this answer, the customer should be able to…”
The blank should end with a concrete action. If the intended result is vague, it becomes difficult to judge whether the answer is useful. If the result is clear, the team can check whether the response contains the facts, conditions, choices, and alternatives required to support it.
The same approach also sharpens measurement. Instead of counting only whether a response was delivered, leaders can examine whether customers completed the intended next action, abandoned the journey, returned with the same issue, or changed course because an important condition was omitted.
Jonah’s central point is simple: the purpose of customer-facing AI is not merely to return relevant information. It is to reduce uncertainty enough for responsible customer action.
Read Dr. Jonah Tebaa’s full first-person explanation and use the test on one high-volume customer journey this week.