Nº 0 · Advanced Analytics

Key Driver Analysis

The factors that drive your business outcomes aren’t always the ones you’d expect.

Nº 1 · Benefits

Isolates what genuinely moves the needle.

Prioritises investment decisions

Ranks each factor by how much it actually moves the outcome, so spend goes to what matters most.

Identify area hurting your customer experience

Isolates the weak points dragging satisfaction down so you know exactly where to intervene.

Build on your brand’s strengths

Reveals the drivers you already lead on so you can protect and extend the advantages that win customers.

Nº 2 · The method

What is key driver analysis?

When everything seems to matter, nothing is truly a priority. Key driver analysis cuts through the noise to identify which factors genuinely drive your outcomes and ranks them by the size of their impact, so investment decisions have a clear, evidence-based basis.

The technique draws on multiple sophisticated approaches: regression analysis, Shapley value decomposition, relative importance analysis, dominance analysis, and machine learning. Each controls for confounding factors and isolates the independent contribution of each driver. The output is not a list of associations but a ranked picture of what actually moves the needle, and by how much.

Key driver analysis (KDA) goes further than standard correlation to identify which factors are genuinely driving your critical business outcomes, whether that’s satisfaction, NPS, loyalty, purchase intent, or brand preference.

Combined with kano analysis, KDA often surfaces non-linear relationships that standard modelling misses: must-have attributes that damage outcomes when absent but deliver no uplift when present, alongside genuine differentiators that create significant competitive advantage when delivered well.

When to use it

When you need to know where to focus. Key driver analysis is valuable across a range of business decisions.

Customer satisfaction improvement and NPS driver identification

Brand health – identifying which attributes drive consideration and preference

Customer experience investment and touchpoint prioritisation

Churn reduction – understanding what drives attrition

Budget allocation across products, services, and channels

Employee engagement driver identification

Nº 3 · Decisions

Decisions it supports.

Key driver analysis is most valuable when you need to move beyond instinct and allocate resources with confidence.

Proposition

Which features drive purchase intent?
How do we optimise our offering for maximum appeal?

Brand

Which brand attributes drive consideration and equity?
Where should brand-building investment be concentrated?

Customer

Which touchpoints have the most impact on satisfaction?
What genuinely drives loyalty and reduces churn?

Communications

Which messages shift brand perception?
What content drives engagement and action?

Nº 4 · How it works

Five steps from signal to actionable drivers.

The most appropriate modelling approach is selected based on the data and objectives, drawing on techniques such as regression, Shapley values, dominance analysis, and machine learning. Every model is rigorously validated before any findings are shared.

STEP · 01

Define objectives

Identify the outcome to model and the key business question it needs to answer.

STEP · 02

Prepare data

Structure and refine the dataset, whether from existing trackers or new research, ensuring it is ready for modelling.

STEP · 03

Select approach

Apply the most appropriate techniques, including regression, Shapley value decomposition, relative importance analysis, dominance analysis, and machine learning.

STEP · 04

Build model

Estimate the independent contribution of each driver, isolating true impact from simple correlation.

STEP · 05

Validate results

Test the model using hold-out samples to confirm accuracy and ensure findings are robust and reliable.

Nº 5 · Getting started

What you need to begin

A clearly defined business outcome to model against
An attribute list — or a working hypothesis we can refine together
Existing tracking or survey data, if you have it
A view of how the answer will translate into investment

Nº 6 · FAQ

Frequently asked questions.

Correlation measures association. When one variable changes, another tends to change. KDA employs techniques that control for confounding factors and isolate the independent impact of each driver. That distinction matters. Correlation tells you what moves together. KDA tells you what genuinely influences what.

Machine learning exhaustively tests thousands of model specifications, variable combinations, and approaches to find the strongest possible model. It removes analyst bias, never misses interaction effects, and consistently achieves higher predictive accuracy. Explainable machine learning makes these models transparent and actionable.

We set aside 20–30% of the data before building the model, then test its predictions on that hold-out group. Strong models predict hold-out outcomes accurately. Where possible, we also validate against actual behavioural data, such as purchases, renewals, and referrals, to confirm that attitudinal drivers translate into actions.

Yes. KDA works well with existing brand trackers, satisfaction programmes, or employee engagement surveys. We need the outcome measure and the potential driver questions. This is often the most efficient approach, leveraging existing research investment rather than commissioning new fieldwork.

Nº 7 · Get in touch

Ready to run a key driver analysis?

If you need to understand what’s truly driving your outcomes, we can help. Get in touch to discuss your brief.

Nº 8 · Related