Importance-Performance Gap Analysis: How to Build and Read the IPA Grid

The technique comes from John Martilla and John James, "Importance-Performance Analysis", Journal of Marketing, vol. 41, no. 1, 1977, pp. 77 to 79. It compares attributes on a two-dimensional grid showing their relative strengths and weaknesses. Its popularity in service and tourism research is why Azzopardi and Nash wrote a methodological evaluation of the literature, listing validity concerns to mitigate (Tourism Management, vol. 35, 2013, pp. 222 to 233; Bond University record).

Each respondent rates every attribute twice: how important it is and how well the brand performs. You then average both across respondents, giving two numbers per attribute.

Koji's guide to the method names the quadrants and what each suggests (Koji). The quadrant labels are widely used in IPA writing.

The word "possible" in the last label matters. An attribute can be low in importance on average but vital to a segment, or can be a basic expectation that only matters when it fails.

Koji recommends data-centred crosshairs at the average of each axis across the attributes, not at the scale midpoint, and names crossing at the fixed midpoint as the most common mistake. Because the lines move with your data, the grid shows relative priorities. It cannot say an absolute score is good or bad.

State the rule you used. Medians, or the average across all attributes, will move borderline attributes between quadrants, so show the numbers next to the chart.

These scores are illustrative. Importance and performance are both averages on a 1 to 5 scale. The average importance across the six attributes is 3.88 and the average performance is 3.83.

Delivery and support are the two attributes to improve. Price is the instructive one. At 3.9 it only just clears the performance line of 3.83, so a small change in wording or sample could move it into the first quadrant. Do not present a borderline attribute as a clear strength.

The gap column is our arithmetic, not a standard from the original paper. Some analysts rank by the gap instead of using the grid. A gap only means something if both ratings use the same scale, and it ignores how far each attribute sits from the lines.

Grigoroudis and Spyridaki define stated importance as the straightforward customer preferences for the weight of a satisfaction criterion, and derived importance as estimated statistically from customers' judgments of performance (Operational Research, 2003). Koji notes respondents tend to rate nearly everything as important, which compresses a stated axis, and that derived importance is usually more trustworthy where the two diverge.

Run both. If delivery is stated at 4.5 but derived importance ranks it fifth, find out why before moving budget. For how to estimate the derived measure, read our key driver analysis guide.

Azzopardi and Nash's evaluation is the standard critique, and it focuses on conceptual and measurement validity. Koji adds practical cautions: read quadrants as directional priorities, not precise coordinates, because importance and performance can be correlated across attributes; split vague or double-barrelled attributes; and re-run the exercise regularly. It also says IPA is a prioritisation lens, not a discovery method, so choose attributes through earlier qualitative work.

A practical sequence is to list attributes from earlier interviews, collect importance and performance for each on the same scale, add one overall satisfaction question, and compute the derived importance from that question. Then draw the stated grid and the derived grid side by side. Attributes that stay in the same quadrant on both are safe to act on. Attributes that jump quadrants need a closer look at the open-ended answers before any budget moves.

Add intervals. Each average has sampling error, so show confidence bars on both axes, using the approach in our significance testing guide. Attributes whose bars cross a cut line are borderline.

Poseidon, our analytics engine, lists an importance-performance gap among the formulas it selects by question type. Whichever tool you use, check how it sets the cut lines and whether importance is stated or derived.

Hercules surveys are app-based and reach smartphone users only. The Free plan includes 30 responses, enough to test an attribute list and rating wording.