How to read a naming study when the results are close

This walkthrough uses the InstaSights example report, a name testing study of three names for a ready-to-drink oat latte. Its 250 responses are simulated. It is an illustration of how to inspect a report, not evidence that one of these names will perform best in the market.

Start with the question and context

The first-choice question asks which name is most appealing: Daybreak, Oat & Co., or First Light. That answers a specific question about appeal in the study’s context. It does not tell us which name people will remember, which is legally available, or which will sell more cans.

Before interpreting a result, check the product description, audience definition, question wording, and answer choices. You can inspect these in the example questionnaire.

Read the size of the lead

Oat & Co. receives 36% of first choices, First Light 34%, and Daybreak 30%. The lead over First Light is two percentage points. A useful description is: “Oat & Co. leads narrowly in this simulated sample.”

That is different from saying the study has established the best name. These percentages alone do not establish statistical significance or guarantee that the ordering will hold with real customers.

Check whether related questions agree

The ranking question adds a different result: Daybreak and First Light each receive 41% of first-place rankings, compared with 18% for Oat & Co. The first-choice and ranking questions therefore point to different orders. That makes it especially important to inspect both questions rather than declare a winner from one chart.

There is also a response-quality issue in the attribute question: 12 of its 191 responses select four attributes despite an instruction to choose up to three. The attribute chart excludes these 12 responses and uses 179 responses as its base. The other questions keep their original responses. See the example data checks before using that chart. The demonstration shows why inspecting individual responses is part of reading a report.

Look at the audience groups

In the example, 52% of the 18–24 group choose Oat & Co., while 49% of the 55–64 group choose First Light. The overall result hides these differences.

Check the group sizes too: those groups contain 46 and 37 simulated responses respectively. Treat the differences as directions to investigate, and focus on the audience you intended to serve before seeing the results. Avoid picking a segment simply because it favours the option you already like.

Connect the finding to a next step

For this example, one reasonable next step is to keep Oat & Co. and First Light on the shortlist and investigate the trade-offs with intended customers. Another is to revisit whether the names communicate the product clearly before testing appeal again.

The choice depends on the decision and its consequences. Simulated preferences can help you form a hypothesis. They do not replace evidence of actual product experience or purchasing.

Keep the evidence with the conclusion

When sharing a result, include the question, audience, response distribution, and the fact that responses were simulated. Preserve the uncertainty in the summary so the next person can judge the conclusion for themselves.

Read our methodology and limitations, or explore the report to see the numbers behind this walkthrough.

Planning a similar study? Start with the brief and naming questionnaire template, or use the evaluation worksheet to examine the evidence behind a method.