Cloud4Feed
Request a Demo
Customers Pricing
Request a Demo

SEM: find what really drives NPS, and how

VoC data tells you how customers feel. Structural equation modelling shows the hidden routes from each touchpoint to that feeling, so you can invest in the levers that move NPS and isolate the risks that pull it down.

Cloud4Feed Research Team Updated 7 min read
Quick answer

Structural equation modelling (SEM) estimates how touchpoints (X) affect an outcome such as NPS (Y), directly and through mediating perceptions (M). A touchpoint’s real contribution is the product of the coefficients along its path, plus any direct effect.

XTouchpoint
MPerception
YNPS
Indirect effect = β(X → M) × β(M → Y)Try the calculator ↓

At a glance

Method
Structural equation modelling with mediation
Inputs
Survey ratings of touchpoints, perceptions and NPS
Output
Direct, indirect and total effects per touchpoint
Typical tools
lavaan, AMOS, Mplus
Use it for
Prioritising CX investments

What is structural equation modelling (SEM)?

Structural equation modelling is a statistical method that tests how a set of variables influence one another, not just whether they move together. In VoC analysis it maps the routes by which a touchpoint, such as staff attentiveness or checkout waiting time, reaches an outcome such as NPS, satisfaction or loyalty.

The key idea is mediation. Touchpoints rarely change recommendation directly; they change a perception first, such as perceived service quality or emotional attachment, and that perception moves the score. Knowing how customers feel (the NPS) is only half the diagnosis. To act, you need to map the structural paths that create that feeling.

Why not just use correlation?

AspectCorrelation or simple key driver analysisSEM
VisibilitySees only direct, surface relationships; leaves the “how” unanswered.Maps indirect paths and the perception mechanisms behind them.
Risk detectionWeak or mixed relationships are easy to overlook.Separates negative direct effects from the positive paths around them.
MeasurementTreats each survey item as error-free.Can model perceptions as latent constructs measured by several items, correcting for measurement error.
Action planPriorities rest on intuition and raw correlations.Priorities rest on estimated effect sizes along each path.

How to read a SEM

XIndependent variable (X)A customer touchpoint, for example store layout or staff attentiveness.
MMediator (M)The hidden perception it works through, for example perceived service quality.
YDependent variable (Y)The final goal, for example NPS or overall satisfaction.

Each arrow carries a standardised coefficient (β). The effect of X through M is the product of the two coefficients on the way:

The lever formulaIndirect effect = β(X → M) × β(M → Y)Total effect = indirect effect + direct effect (X → Y)Example: staff attentiveness → service satisfaction (β = .42) × service satisfaction → NPS (β = .54) = .227.

A four-step diagnostic framework

A mediation analysis tests four research hypotheses in sequence.

  1. H1, direct effect (X → Y). Does the touchpoint affect satisfaction or NPS directly?
  2. H2, mediator formation (X → M). Does the touchpoint significantly predict the perception?
  3. H3, mediator effect (M → Y). Does the perception have a significant effect on the outcome?
  4. H4, indirect effect (X → M → Y). How large is the real effect that travels through the mediating path?

Case study: what really drives NPS in retail

A retailer measured five touchpoints and asked one question: which investments really increase the intention to recommend? The model grouped the five touchpoints into two perception mechanisms. Staff attentiveness and knowledge work through service satisfaction; store layout, checkout waiting time and discounts work through the shopping experience.

Illustrative coefficients. Real coefficients must be estimated with SEM software such as lavaan (R), AMOS or Mplus.
TouchpointPathCalculationTotal effect
Staff attentiveness→ Service satisfaction → NPS.42 × .54.227
Staff knowledge→ Service satisfaction → NPS.38 × .54.205
Store layout→ Shopping experience → NPS.35 × .41.144
Discounts→ Shopping experience → NPS.31 × .41.127
Checkout waiting time→ Shopping experience → NPS + direct.29 × .41 + (−.18)−.061

What the model shows

People are the biggest lever. Staff attentiveness (.227) and knowledge (.205) together add up to .432, about half of the total indirect effect of all five touchpoints (.822). Investment in staff, through service satisfaction, is the strongest route to a higher NPS in this model.

Waiting time is a hidden risk. Checkout waiting time has a weak positive path through the shopping experience (.29 × .41 = .119) and a negative direct path to NPS (−.18). Its total effect is negative (−.061). When the direct and indirect paths point in opposite directions, the touchpoint is affecting recommendation through something the mediator does not capture. A simple correlation would show a small, harmless-looking number and hide both effects.

Path effect calculator

Enter the standardised coefficients from your model for each driver: X → M, M → Y and, if there is one, the direct path X → Y. The ranking updates as you type. It starts with the case-study values.

Ranked by total effect Total effect = β(X→M) × β(M→Y) + β direct
1Staff attentivenessUrgent.227
Indirect .227 · Direct —
2Staff knowledgeUrgent.205
Indirect .205 · Direct —
3Store layoutUrgent.144
Indirect .144 · Direct —
4DiscountsUrgent.127
Indirect .127 · Direct —
5Checkout waiting timeUrgent−.061
Indirect .119 · Direct −.180

From model to action

Rank the touchpoints by total effect and act in that order, but treat any touchpoint with a negative direct effect as urgent whatever its rank. In the case study:

  • Staff attentiveness (.227). Coaching, staffing levels and service rituals that make customers feel noticed.
  • Staff knowledge (.205). Product training and quick access to information on the shop floor.
  • Store layout (.144). Wayfinding and tidiness in the areas customers use most.
  • Discounts (.127). Clear, credible offers rather than more of them.
Urgent: checkout waiting time (−.061). Improving the experience around it is not enough; remove the friction itself, for example with more open tills at peak times or self-checkout.

Then measure again. Track the touchpoints with transactional surveys, and check the overall effect with relational NPS; use MaxDiff when you also need customers’ stated priorities.

Data and method checklist

  • Start from a theory. Draw the paths you expect before you look at the data. SEM tests a model; it does not discover causes on its own, and the arrows are only as causal as your design and theory allow.
  • Measure perceptions with several items. Latent constructs such as service satisfaction are more reliable with three or more survey items each. Check reliability and the measurement model first.
  • Collect enough responses. A few hundred complete responses is a common minimum for a model of this size; larger models need more.
  • Check model fit. Report fit indices such as CFI, RMSEA and SRMR, and do not interpret coefficients from a poorly fitting model.
  • Test indirect effects properly. Use bootstrapped confidence intervals for indirect effects rather than judging each path separately.

SEM FAQ

SEM is a more complete form of it. Key driver analysis usually regresses the outcome on all drivers at once and gives one importance number per driver. SEM adds the mediating perceptions, separates direct from indirect effects and can correct for measurement error.

Cloud4Feed Research Team

Our research, analytics and CX team writes about what we learn running feedback programmes across 15+ industries.

About us →

New guides and research, once a month.

No sales emails. Unsubscribe anytime.

We use your details to respond to your request. See our Privacy Policy.

Ready to put this into practice?

A 45-minute walkthrough set up for your industry and your team.