Transactional vs Relational NPS: When to Use Each (2026 Guide)

The Bottom Line

Relational NPS measures overall loyalty to your brand on a slow cadence (quarterly or semi-annual). Transactional NPS measures satisfaction with a specific moment - right after onboarding, a support ticket, a purchase, or a renewal. You need both: relational NPS is your loyalty trend line, transactional NPS tells you which moments in the journey are bending that line up or down. The mistake most teams make is treating one number as if it does both jobs.

The deeper problem with either type is the same: a 0-10 score without the why is a thermometer, not a diagnosis. That is why leading teams in 2026 pair the NPS scale question with an AI follow-up interview - so every score arrives with a clear, quotable reason attached.

Transactional vs Relational NPS at a Glance

What Is Relational NPS?

Relational NPS is the classic Net Promoter Score: a periodic pulse on how customers feel about your company as a whole. You send it on a schedule, independent of any single interaction, to a broad sample of your base.

Use relational NPS to:

• Track loyalty as a trend line over quarters and years
• Benchmark against your industry (SaaS averages around 36, ecommerce around 45, financial services around 22)
• Report a single north-star loyalty number to leadership
• Segment loyalty by plan, region, and cohort

Because it is not tied to an event, relational NPS answers "are we earning loyalty over time?" - but on its own it rarely tells you what specifically to fix.

What Is Transactional NPS?

Transactional NPS fires immediately after a defined moment in the customer journey, so the rating reflects that experience while it is fresh. Common trigger points:

• Post-onboarding - did the first-run experience deliver value?
• Post-support - did we resolve the issue well?
• Post-purchase / post-checkout - was buying frictionless?
• Post-renewal or post-cancellation - what drove the decision?
• Post-feature-adoption - did a key feature land?

Transactional NPS is your early-warning system. A dip in post-onboarding NPS shows up weeks before it drags down your relational score or your retention numbers. The catch: trigger too many transactional surveys and you cause survey fatigue, which depresses response rates and distorts the data.

How to Decide Which to Run

Run both, with clear roles:

1. Relational NPS quarterly to a representative sample for your loyalty trend and benchmarking.
2. Transactional NPS event-triggered at the 3-5 journey moments that matter most, with frequency caps so no customer is surveyed too often.
3. Reconcile them. When relational NPS dips, drill into transactional scores to find the moment responsible. When a transactional moment scores low, you know exactly where to invest.

A practical rule: relational NPS is for direction, transactional NPS is for diagnosis.

The Missing Half: The Why Behind Both Scores

Whether relational or transactional, a bare 0-10 score has the same fatal gap - it tells you the temperature but not the cause. The traditional fix is an open-text "Why did you give that score?" box, and it fails for the same reason every time: most respondents skip it or write one vague line.

This is where an AI-native platform changes the economics. With tools like Koji, the NPS scale question is followed by an AI-moderated conversation that adapts to each answer:

• A Detractor who says "onboarding was confusing" is automatically asked which step, what they expected, and what would have helped.
• A Promoter is probed for the specific outcome that earned the 9 - gold for your marketing and product teams.
• Every response is auto-analyzed into themes, sentiment, and pull quotes, so you are not hand-coding hundreds of verbatims.

Koji supports six structured question types - open_ended, scale, single_choice, multiple_choice, ranking, and yes_no - so a single flow can capture the NPS scale, a ranked list of drivers, and an open-ended follow-up. See the structured questions guide. Because Koji runs by text or voice with no human moderator and applies a quality gate so only genuine, complete conversations count, you can attach this depth to both your relational and transactional NPS without adding research headcount.

Benchmarks and Sample Size for Both

• Relational NPS: aim for 100+ responses per period for a stable score; 300+ to reliably detect a 5-point change.
• Transactional NPS: because it is continuous, you accumulate volume quickly - but watch for selection bias (only happy or only angry customers responding).
• Benchmark relational, not transactional, against industry averages - transactional scores tend to run higher because the experience is fresh and positive moments over-respond.

Common Mistakes

• Using one survey for both jobs. A quarterly relational survey cannot diagnose a broken support flow.
• Over-triggering transactional surveys. Cap frequency to avoid fatigue.
• Collecting scores without the why. A number with no reason cannot be acted on - add AI follow-up.
• Comparing transactional scores to relational benchmarks. They are not the same baseline.
• Ignoring Passives. The 7-8 crowd is where churn risk and upside both hide; probe them too.

A Practical 90-Day Plan for Running Both

You do not have to launch relational and transactional NPS at once. Sequence them:

Days 1-30: Establish your relational baseline. Send the classic relational NPS question to a representative sample of your customer base. Aim for 100+ responses so the score is stable. Add an AI follow-up so even this first wave arrives with reasons, not just a number. This becomes your loyalty trend line and your benchmark against industry averages.

Days 31-60: Instrument your key journey moments. Map the customer journey and pick the 3-5 moments that most influence loyalty - typically onboarding completion, first support interaction, and renewal. Set transactional NPS to fire immediately after each, with a frequency cap so no customer is surveyed more than once in a set window.

Days 61-90: Reconcile and act. Now you can connect the two. When relational NPS moves, drill into the transactional scores to find the moment responsible. When a transactional moment scores low, the AI follow-up tells you exactly what went wrong and for whom. Route Detractor themes to the owning team and close the loop with affected customers.

How Koji Makes Both Types Effortless

Running two NPS programs by hand means two survey setups, two analysis pipelines, and a lot of manual tagging. An AI-native platform collapses that work. With Koji you configure the relational pulse and the transactional triggers once, and every response - regardless of type - flows into the same automatic analysis: themes, sentiment, and pull quotes with no manual coding. Because interviews run by text or voice with no human moderator, you can scale transactional NPS across every key moment without adding research headcount, and the quality gate keeps low-effort responses from polluting either dataset.

Related Resources

• NPS Survey Guide
• NPS Benchmarks by Industry 2026
• NPS Follow-Up Interviews
• CSAT vs NPS vs CES
• Structured Questions Guide
• Voice of Customer Research Program
• Customer Effort Score Guide