Top-2-box score

Survey scales produce distributions — so many 1s, so many 2s, and so on — but managers need a single number to compare across branches and months. Box scores solve that by counting only the answers that clearly mean “satisfied”. The name comes from paper questionnaires, where each answer option was a box: the “top two boxes” are the two best options.

Three variants are common:

• Top-box — only the single best option (5 on a 1–5 scale). Strict; measures delight.
• Top-2-box — the two best options (4 and 5). The most widely used convention for CSAT.
• Bottom-2-box — the two worst options (1 and 2). Measures clear dissatisfaction and is useful for alerting.

Top-2-box score = (number of responses in the two highest options ÷ total responses) × 100

Example: 250 responses to “How satisfied were you with your visit?” on a 1–5 scale:

Answer	Responses	Share
5 — Very satisfied	110	44%
4 — Satisfied	80	32%
3 — Neutral	30	12%
2 — Dissatisfied	20	8%
1 — Very dissatisfied	10	4%

Top-2-box = (110 + 80) ÷ 250 × 100 = 76%. Top-box = 44%. Bottom-2-box = 12%. The mean is 4.04. All four numbers describe the same data, which is why every report should state which one it uses. The CSAT calculator runs the top-2-box calculation for you.

On a seven-point scale, top-2-box means 6 and 7; some teams use top-3-box on longer scales. On smiley scales, the two happiest faces are the top two boxes.

• It is easy to understand. “76% of customers were satisfied” needs no explanation; “average 4.04” does.
• It ignores lukewarm answers. A customer who picks the middle option is not actually satisfied, and T2B does not reward that answer.
• It is robust to scale quirks. Means can be pulled by a few extreme answers; box scores count people, not points.
• It maps to action. Pair T2B (who is satisfied) with bottom-2-box (who needs a call back) and you have both the headline and the to-do list.

The trade-off: T2B hides movement within the boxes. A shift from 5s to 4s leaves T2B unchanged while top-box falls — so track top-box too if delight matters to your brand.

• State the convention next to every number: “CSAT (top-2-box, 1–5 scale)”.
• Keep the scale constant over time; switching from 5 to 7 points breaks the trend.
• Report the base — 76% of 25 responses is not the same evidence as 76% of 2,500.
• Break it down by branch, service and touchpoint to find where the bottom boxes come from.
• Read the comments of 3s. Neutral answers often contain the most specific, fixable suggestions.

Qmeter’s survey builder offers smiley, star (3 to 10 stars) and NPS components, and the report builder calculates CSAT and other measures by branch, channel and period. See CSAT software and NPS vs CSAT vs CES.

Top-2-box and NPS both turn a rating scale into one number, but they work differently:

	Top-2-box (CSAT)	NPS
Question	Satisfaction with an interaction or service	Likelihood to recommend
Scale	Usually 1–5 or smileys	0–10
Calculation	% in the two best options	% 9–10 minus % 0–6
Range	0% to 100%	−100 to +100
Penalises dissatisfaction?	No — unhappy answers simply do not count as satisfied	Yes — detractors are subtracted

Because NPS subtracts detractors, it reacts more strongly to clearly unhappy customers; top-2-box is easier to explain and suits short transactional surveys. Many programs report top-2-box CSAT for interactions and NPS for the overall relationship.

• Comparing a top-box number with a top-2-box number from another team or vendor.
• Mixing scales — five smileys at the kiosk, ten stars by email — and blending them into one T2B.
• Unbalanced scales with more positive than negative options, which inflate T2B.
• Reporting T2B without volume or without the bottom-2-box share.