Customer Behavior & Satisfaction Analytics Flashcards

A hotel survey shows 55% promoters, 30% passives, and 15% detractors. What is the Net Promoter Score?

Answer: 40

NPS equals % promoters minus % detractors, so 55 - 15 = 40.

A hotel emails its satisfaction survey three weeks after checkout and gets very low response rates. What is the most effective first fix?

Answer: Send the survey within 24-48 hours of checkout

Surveys sent soon after the stay capture fresher recall and get higher response rates.

Only guests with very strong opinions respond to a hotel's survey, so the results misrepresent the full guest population. This is an example of:

Answer: Non-response bias

Non-response bias happens when the people who respond differ systematically from those who don't.

Which metric is best suited to measuring how easy guests found the mobile check-in process?

Answer: Customer Effort Score

Customer Effort Score measures how easy a specific interaction or task was for the customer.

On a 5-point satisfaction scale, a hotel reports a 'top-box score' of 62%. What does this mean?

Answer: 62% of respondents chose the highest rating

Top-box is the share of respondents who picked the single highest scale point.

Survey responses on a strictly ordinal Likert scale are best summarized with which measure of central tendency?

Answer: Median

Ordinal data has rank order but no guaranteed equal intervals, so the median is the most appropriate measure.

An analyst wants to cut the margin of error of a guest satisfaction estimate in half. Roughly how much must the sample size grow?

Answer: Quadruple it

Margin of error scales with 1/√n, so halving it takes four times the sample.

An analyst wants to know whether review sentiment about 'breakfast' differs from sentiment about 'staff' within the same reviews. Which technique fits best?

Answer: Aspect-based sentiment analysis

Aspect-based sentiment analysis assigns sentiment to specific attributes mentioned in the text.

Hotel A has a 5.0 average from 3 reviews; Hotel B has 4.6 from 900 reviews. Which approach ranks them most fairly?

Answer: Use a Bayesian (shrunk) average that weights toward the overall mean

A Bayesian average pulls ratings with few reviews toward the overall mean, which reduces small-sample noise.

Online hotel ratings often cluster at 5 stars with a smaller spike at 1 star. This J-shaped distribution mainly reflects:

Answer: Self-selection, where guests with extreme experiences are more likely to post

Reviews are voluntary, so guests with very good or very bad experiences are overrepresented.

In text mining guest comments, which method highlights words that are frequent in one hotel's reviews but rare across the whole portfolio?

Answer: TF-IDF weighting

TF-IDF raises the weight of terms that are common in one document set and lowers the weight of terms common everywhere.

Which metric aggregates review scores from multiple online sources into a single reputation benchmark?

Answer: Composite online reputation index (e.g., Global Review Index)

Composite reputation indexes normalize and combine scores from several review sites into one benchmark.

After a lobby renovation, a hotel's overall review score stays flat. What is the best next analytical step?

Answer: Break reviews down by topic to isolate mentions of the lobby before and after

A topic-level comparison shows whether lobby sentiment improved even if other topics pulled the overall score down.

An analyst finds that hotels with higher management response rates to reviews tend to have higher ratings. What is the most accurate conclusion?

Answer: The data shows an association, and causality needs further testing

Observational correlation does not prove causation, since confounders such as overall management quality could explain both.

An RFM model for loyalty members segments guests based on which three variables?

Answer: Recency, Frequency, Monetary value

RFM scores customers on how recently they stayed, how often they stay, and how much they spend.

A guest generates $200 annual margin, the retention rate is 60%, and the discount rate is 10%. Using CLV = m × r / (1 + d − r), what is the CLV?

Answer: $240

200 × 0.6 / (1.1 − 0.6) = 120 / 0.5 = $240.

To compare the repeat-stay behavior of guests acquired in Q1 with those acquired in Q3, which analysis is most appropriate?

Answer: Cohort analysis

Cohort analysis groups customers by acquisition period and tracks their behavior over time.

A guest whose complaint was resolved exceptionally well later reports higher loyalty than guests who had no problem. This is known as:

Answer: The service recovery paradox

The service recovery paradox describes cases where excellent recovery leaves satisfaction higher than it was before the failure.

Before running k-means clustering on guest spend ($) and nights stayed, an analyst should usually:

Answer: Standardize the variables to comparable scales

K-means uses distance, so unscaled variables with large ranges, such as dollars, would dominate the clusters.

A corporate traveler spends 30 nights a year in hotels, 12 of them at your brand. What is your brand's share of wallet?

Answer: 40%

Share of wallet = 12 / 30 = 40% of the guest's total category spend.