A top box score is the percentage of survey respondents who chose the most favorable answer on a rating scale. It turns a spread of ratings into one easy number, such as the share of customers who were “very satisfied.” Teams use it for satisfaction, purchase intent, and likelihood to recommend.
That simplicity has a cost. Two brands can post the same top box score and still have very different customers behind it.
In this guide, we’ll cover how to calculate a top box score, which box to report, and how to avoid the mistakes that hide real differences.
What is a top box score?
A top box score is the percentage of respondents who select the highest, most favorable point on a rating scale. The “box” is one answer option on the scale. Analysts also report the top two or top three boxes, which add the most favorable options together.
The formulas are simple:
- Top box: Respondents who chose the highest option ÷ total respondents × 100
- Top two box: Respondents who chose either of the two highest options ÷ total respondents × 100
Say 200 customers answer a five-point satisfaction question. If 70 pick “very satisfied,” the top box score is 35%. If another 50 pick “satisfied,” the top two box score is 60%.
Researchers like this measure because executives read a percentage faster than a mean. It appears most often on a Likert scale, which is a rating scale with ordered answer options such as “strongly disagree” to “strongly agree.” Here are some Likert scale examples if you need to design one. A top box score is one of many quantitative metrics you can pull from a closed-ended question.
Top box, top two box, bottom box, and net box compared
These terms sound alike, but each answers a different question. Pick the one that matches the decision you need to support.
| Measure | What it counts | Example on a five-point scale | Best for |
|---|---|---|---|
| Top box | Share choosing the highest option | 35% pick “very satisfied” | Spotting strong advocates |
| Top two box | Share choosing the two highest options | 60% pick “satisfied” or “very satisfied” | Checking broad approval |
| Bottom two box | Share choosing the two lowest options | 12% pick “dissatisfied” or “very dissatisfied” | Finding unhappy customers |
| Net box | Top box minus bottom box | 35% minus 5% = 30 points | Comparing with one number |
| Mean score | Average of all ratings | 3.9 out of 5 | Tracking overall movement |
| Net Promoter Score (NPS) | Promoters minus detractors on a 0 to 10 scale | 55% minus 20% = 35 | Loyalty benchmarking |
NPS is closely related. Promoters (scores of 9 and 10) are effectively the top two boxes on a 0 to 10 scale, but NPS then subtracts detractors. This NPS calculation guide walks through the full formula.
How to calculate a top box score step by step
You can calculate a top box score by hand in a spreadsheet in about five steps. Follow the same rules every wave so your trend lines stay comparable.
- Check the scale.
Confirm which end is favorable. A scale where 1 is best needs recoding or careful labeling.
- Set your base.
Count total valid responses. Decide how to treat “not applicable” answers and blanks, and document that choice.
- Count the top box.
Add the responses in the favorable option or options.
- Divide and convert.
Divide the top box count by the base, then multiply by 100.
- Add the other end.
Calculate the bottom box too, so favorable and unfavorable results sit side by side.
Here is a worked example. Suppose 400 valid respondents answer a likelihood-to-recommend question. A total of 120 choose “definitely would recommend,” which gives a top box of 30%. Another 48 choose “probably would recommend,” so the top two box is (120 + 48) ÷ 400 = 42%.
Now 64 respondents choose the two lowest options, so the bottom two box is 16%. The ratio of favorable to unfavorable is 42 ÷ 16, or about 2.6 to 1.
Top box score example: Same score, different story
A top two box score can look identical for competitors whose customers feel very differently. This hypothetical example is adapted from a case discussed in a Decision Analyst white paper by Jerry Thomas. Three supermarkets all post a 45% top two box score on “likelihood to recommend.”
| Store | Top box | Second box | Top two box | Bottom two box |
|---|---|---|---|---|
| Alpha Beta | 35% | 10% | 45% | 5% |
| Super Store | 25% | 20% | 45% | 5% |
| Acme Foods | 15% | 30% | 45% | 20% |
The scores match, but the stories do not. Alpha Beta has the largest group of enthusiastic fans at 35%. Acme has few of them, and its respondents are four times as likely to say they would not recommend the store.
Collapsing the scale into one number would bury both findings. Show the full distribution, or at least pair the top box with the bottom box.
Top box score vs mean: Which should you report?
Report both. The mean summarizes the whole scale, while the top box captures the strength of positive opinion. They answer different questions, and they can disagree.
MeasuringU tested this on a seven-point ease-of-use question about seat selection on 12 airline websites. Top box scores ranged from 16% to 41%. Means ranged only from 5.0 to 6.1. The correlation between mean and top box was r = .76, which means they shared about 58% of their variance, according to Sauro and Lewis at MeasuringU.
The mean and the top two boxes moved together much more closely, at r = .97. That makes sense, because the top two boxes cover more of the scale.
Polarized audiences are where the gap hurts most. A group split between 5s and 1s can share a mean with a group that answers 3 across the board. The top box and bottom box expose the difference.
Pros and cons of top box scores
Top box scores earn their place in reports because they are quick to read. They also carry limits you should state up front.
Pros
- Executives understand a percentage without a statistics lesson.
- Waves, segments, and competitors are easy to compare.
- The score focuses attention on strongly positive customers.
- It works across satisfaction, intent, and loyalty scales.
Cons
- It discards the rest of the response distribution.
- Results shift with scale length and answer wording.
- Small changes can be noise rather than real movement.
- It shows what people chose, not why they chose it.
How to choose the right box for your survey
Choose the box that fits your scale length and the decision at hand. Work through these five questions in order.
- How long is the scale? On a five-point scale, the top two boxes cover about 40% of the options. On a seven-point scale, it covers only about 29%, so MeasuringU suggests considering the top three box there.
- What decision follows? Use the top box to predict advocacy or repeat behavior, because extreme responses tend to predict future behavior better than tepid ones. Use the top two boxes for broad approval and the bottom box for churn risk.
- Is the audience polarized? If yes, show both ends of the scale.
- Do stakeholders need one number? Use net box or NPS, and keep the distribution in an appendix.
- Is there a benchmark? Match the benchmark’s exact definition, scale, and wording before comparing.
How to tell if a top box change is real
A top box score is a sample estimate, so it moves with chance. Check the margin of error before you celebrate a two-point gain.
For a 95% confidence level, use this approximation:
Margin of error ≈ 1.96 × √(p × (1 − p) ÷ n)
Here, p is the top box share as a decimal and n is the number of respondents. With p = 0.42 and n = 400, the margin is about ±4.8 points. With n = 100, it widens to about ±9.7 points.
Comparing two waves needs a bigger gap. If wave one scored 42% and wave two scored 45%, each with 400 respondents, the difference must exceed roughly 7 points to stand out from noise. Here is a full explanation of margin of error and how to apply it.
Segments shrink n fast. A top box for 60 new customers is far shakier than one for 600 total customers.
Common mistakes when reporting top box scores
Most top box errors come from inconsistent rules, not from the math. Watch for these.
- Reporting a score without the base.
Always show n, or readers cannot judge reliability.
- Mixing scale lengths.
A 50% top box on a five-point scale does not equal 50% on a ten-point scale.
- Changing the denominator.
Dropping neutral answers one wave and keeping them the next breaks the trend.
- Skipping segments.
Heavy and light users, or new and existing customers, can hide opposite stories.
- Using only the top two boxes.
It can bury differences at the very top and bottom of the scale.
Top box score examples across common survey metrics
Top box logic shows up in many metrics, often under another name. These examples show how teams apply it.
- Customer satisfaction (CSAT): Teams usually report the share of respondents who chose “satisfied” or “very satisfied,” which is a top two box. This CSAT measurement guide covers the formula.
- Likelihood to recommend: The percentage who “definitely” or “probably” would recommend a brand, product, or store.
- Purchase intent: The share who say they “definitely” or “probably” will buy, common in ad and concept testing.
- Brand tracking: The share who rate a brand “very favorable” across quarterly waves.
- Employee engagement: The share who “strongly agree” with statements about their workplace.
Calculating top box scores with QuestionPro
QuestionPro’s Analytics dashboard shows the count and percentage for each answer option. You can read the top two box straight from those percentages by adding the two highest options.
There is also a shortcut. The weighted mean help document explains Customize Scale, which lets you assign a weight to each answer option. The customized mean divides weighted counts by total responses. In principle, giving the top options a weight of 1 and every other option 0 returns the top box share as a decimal. Test it on one question first to confirm it behaves as you expect.
For ongoing programs, QuestionPro Customer Experience supports NPS, CSAT, and CES surveys with real-time dashboards, so favorable scores sit beside the rest of your metrics. Its AskWhy follow-up asks respondents to explain a low rating, which supplies the context a top box score lacks.
Report the score and the story behind it
A top box score is a headline, not the full report. It tells you what share of people were delighted, and it does that quickly. It cannot tell you why they were delighted or how the rest of your audience feels.
Use it as the first line of your findings. A complete top box report also includes:
- The bottom box, so readers see the unfavorable side
- The mean, to show where the whole scale sits
- The base size and margin of error
- A segment cut that explains any surprise
When two numbers disagree, trust the distribution and look for the segment that explains it.
Frequently Asked Questions (FAQs)
There is no universal benchmark. Results depend on scale length, question wording, industry, and audience. Compare against your own earlier waves and direct competitors measured the same way. A 50% top box on a five-point scale is not comparable to 35% on a seven-point scale.
Yes. Neutral respondents are real respondents, so they belong in the denominator. Exclude only blanks and “not applicable” answers, and apply the same rule every wave. Dropping neutrals inflates the score and makes trends hard to trust.
Yes. Use weighted counts in both the numerator and the denominator. Many US studies weight samples to match U.S. Census Bureau demographics, so confirm your weighting scheme before comparing waves, and report whether the figures are weighted or unweighted.
Define the top box by meaning, not by number. If 1 is “very satisfied,” then 1 is your top box. Recode reversed scales before combining waves or surveys, so every question points in the same direction.
Only loosely. A top box needs an ordered scale. On a yes-or-no question, the “yes” share already works as the favorable rate. For open-ended answers, code the text into favorable and unfavorable groups first.
