SERVQUAL Gap Analysis

The SERVQUAL model

SERVQUAL is the best-known instrument for measuring service quality. It was developed by A. Parasuraman, Valarie Zeithaml and Leonard Berry in the 1980s, based on the idea that customers judge service quality by comparing what they expect from a service with what they perceive they received. The difference, perception minus expectation, is the service quality gap: negative gaps mean the service falls short, positive gaps mean it exceeds expectations.

The standard instrument has 22 pairs of items grouped into five dimensions, remembered by the acronym RATER:

• Tangibles (items 1–4): physical facilities, equipment, appearance of staff and communication materials.
• Reliability (items 5–9): ability to perform the promised service dependably and accurately. It is usually the most important dimension.
• Responsiveness (items 10–13): willingness to help customers and provide prompt service.
• Assurance (items 14–17): knowledge and courtesy of employees and their ability to convey trust and confidence.
• Empathy (items 18–22): caring, individualised attention to customers.

Each item is asked twice, once about expectations of an excellent company in the sector and once about perceptions of the specific company, usually on a seven-point scale from strongly disagree to strongly agree.

What this tool calculates

• Item gaps. For each pair, the mean expectation, mean perception, mean gap (P − E), its standard deviation, a paired t-test of whether the gap differs from zero, and Cohen’s d_z as an effect size. Significant negative gaps are highlighted in red.
• Dimension gaps. Each respondent’s dimension score is the mean of its items; the tool reports mean E, P and gap per dimension with paired t-tests and effect sizes.
• Reliability. Cronbach’s alpha for the perception items and for the gap scores of each dimension, because difference scores can be less reliable than their parts.
• SERVQUAL scores. The unweighted score is the average of the five dimension gaps. If you enter importance weights, for example from a question asking customers to allocate 100 points among the five dimensions, the tool computes the weighted SERVQUAL score as Σ weight × dimension gap.

The dimension definitions box uses the standard 22-item structure by default. If you used a shortened or adapted questionnaire, edit the lines, for example Reliability: 1-4 or Access: 2, 5, 9, with the positions of the items in your selected lists. Leave the box empty to analyse all items as one dimension.

Worked example with the built-in data

The example contains 210 customers answering all 22 expectation and perception items on seven-point scales. Expectations are high and uniform (about 6.1 everywhere), while perceptions vary by dimension. Every dimension shows a significant negative gap. The largest shortfall is in Reliability (gap −1.26, d_z = −1.52), followed by Responsiveness (−0.88) and Empathy (−0.72); Tangibles has the smallest gap (−0.39). The unweighted SERVQUAL score is −0.76 and the weighted score, using importance weights in which reliability counts 32%, is −0.86: the weighting makes the overall picture worse because the biggest gap is in the most important dimension. All perception scales have alphas around .77–.83, and gap scales around .71–.77, which is acceptable.

The gaps model behind SERVQUAL

The measured gap between expectations and perceptions is “Gap 5” in the original gaps model of service quality. The model explains it as the result of four internal gaps: (1) management not knowing what customers expect, (2) service standards that do not reflect those expectations, (3) delivery that does not meet the standards, and (4) external communications that promise more than is delivered. When SERVQUAL shows a large customer gap, these internal gaps are where to look for causes. A reliability shortfall, for example, often points to Gap 3 (delivery not meeting standards), while high expectations driven by advertising point to Gap 4.

Designing and collecting SERVQUAL data

• Adapt the wording to your sector (banking, hospitals, hotels, universities, public services) but keep the dimensional structure if you want comparability.
• Ask expectations and perceptions with the same scale. Some studies ask them in separate sections, others side by side; be consistent.
• Use the same order of items in both sections so the columns pair correctly, for example E1–E22 and P1–P22.
• Aim for at least 100 respondents, and 200 or more for subgroup comparisons and factor analysis.
• Add an importance question (100-point allocation or a rating per dimension) if you want a weighted score.

Interpreting the results

Focus on the dimensions with the largest significant negative gaps, then drill down to the items within them to find specific problems. Effect sizes help judge practical importance: a gap of d_z around −0.2 is small, −0.5 moderate and −0.8 or more large. A positive gap means expectations are exceeded, which may indicate over-investment if the dimension is not important to customers. Compare gaps across branches, segments or over time to track improvement programmes.

Criticisms and alternatives

SERVQUAL has been widely debated. Cronin and Taylor (1992) argued that perceptions alone (SERVPERF) predict overall quality and satisfaction as well as or better than gaps, with half the questionnaire length. Difference scores can have lower reliability and inherit problems of the expectation measure, which tends to be high and compressed near the top of the scale. The five-dimension structure does not always replicate in confirmatory factor analysis. Good practice is therefore to report reliability, test the factor structure with CFA (available on this site), and, where possible, compare the gap results with a perceptions-only analysis.

Reporting

Report the number of respondents, the scale, the dimension structure, and for each dimension the mean expectation, perception and gap with the paired t-test, plus the overall (weighted) score and reliability. Example: “Perceived service fell significantly short of expectations on all five SERVQUAL dimensions (N = 210). The largest gap was for reliability (M = −1.26, t(209) = −22.07, p < .001, d_z = −1.52); the overall unweighted SERVQUAL score was −0.76.”

Glossary of service quality terms

• Expectation: what customers believe an excellent service provider in the sector should offer.
• Perception: how customers rate the actual service of the company being evaluated.
• Gap score: perception minus expectation; negative values show a shortfall.
• RATER: the five dimensions: reliability, assurance, tangibles, empathy and responsiveness.
• Zone of tolerance: an extension in which customers give both a desired and an adequate level of service, defining a range of acceptable performance.
• SERVPERF: a perceptions-only alternative that often predicts satisfaction as well as gap scores.
• Weighted SERVQUAL: the overall score obtained by weighting dimension gaps by their importance to customers.
• Gaps model: the framework explaining the customer gap by four internal gaps in knowledge, standards, delivery and communication.

Key references: Parasuraman, Zeithaml and Berry (1988), Journal of Retailing; Parasuraman, Zeithaml and Berry (1985), Journal of Marketing; Cronin and Taylor (1992), Journal of Marketing.