Nine neutral questions for marketing automation customer interviews

The best marketing automation customer interviews reconstruct what happened. They do not ask a prospect to endorse a roadmap or predict what they might buy. They are also the first step of any platform review: before comparing HubSpot, Marketo, Salesforce Marketing Cloud, Klaviyo, Braze, ActiveCampaign, Adobe Campaign or Customer.io on features and licence terms, you need to know what work the team actually does. The nine questions below are designed to expose that work, the hand-offs and the constraints, while leaving room for an unexpected answer.

Method and limitations: this non-ranked list was compiled on 5 September 2026 from current UK government research and ICO guidance. It covers business workflow discovery, not clinical, academic or vulnerable-participant research. No interview was conducted, no answer was supplied and no product was tested. A qualified reviewer must approve the protocol and data handling.

What to take away

1. What was the last campaign or customer journey you worked on?

Ask for one recent instance, then establish its start and end. Specific events are easier to check than general opinions. GOV.UK's in-depth interview guidance recommends open, neutral questions and real examples rather than accounts of how work should happen.

Ask which platform the journey ran in — HubSpot, Marketo, Salesforce Marketing Cloud, Klaviyo, Braze, ActiveCampaign, Adobe Campaign or Customer.io — and whether the record lived in the CRM, a CDP or the email service provider. Naming the system early turns a generic process story into evidence you can carry into a platform review.

2. What triggered the first action?

The trigger may be a form, booking, renewal date, transaction or manual request. Learn who recognised it, where it was recorded and what happened when information was incomplete.

Translate the trigger into platform terms: a form or landing-page submission in HubSpot, an ecommerce event in Klaviyo, a lead-score threshold in Marketo, a lifecycle stage change in the CRM, a renewal date in a billing system. Ask whether the automation platform detected the trigger itself or received it through an integration.

3. Can you walk me through each hand-off?

Follow the artefact, not the organisation chart. Ask which person or system received the item, how they knew it was ready, and how exceptions returned to the previous step.

Map the hand-offs by system rather than by team: CRM to automation platform, automation platform to sales engagement, ESP to subscription centre. Ask whether each connection was native, built in middleware or maintained by hand, and who owns it when it breaks.

4. Where did somebody have to correct or re-enter information?

Request an example without collecting unnecessary customer details. A correction may reveal field definitions, permissions or ownership problems that automation alone will not solve.

Ask which fields are re-keyed between the CRM and the automation platform. Lead-scoring values, lifecycle stages, consent and subscription fields, and campaign attribution are the usual casualties, and each points to a field-mapping, integration or permission problem that belongs in the platform review.

5. How did you know the journey had succeeded or failed?

Listen for an observable outcome and its recording point. "More engagement" needs translation into a defined event, denominator, time window and accountable system.

Ask which system holds the number they trust: attribution in the automation platform, pipeline or revenue in the CRM, deliverability and inbox placement in the sending platform. Record the metric, the report it comes from and the system that owns it, because those definitions differ between vendors.

6. Who could change this process, and who could stop a change?

Map operators, process owners, data or security reviewers, procurement and finance. Then ask what evidence each role expects before approval.

In platform terms, ask who owns the instance, who approves new workflows, and who controls user permissions and API connections. On the commercial side, ask who signs the licence, who approves a seat or tier change, and who can veto a migration.

7. What information would you refuse to put into this workflow?

This brings sensitivity, permission and trust into the discussion without assuming a legal conclusion. Follow up on why the boundary exists and how the team handles the work today.

This is where consent, special-category data and suppression lists surface. Ask which fields the team keeps out of the automation platform altogether, what it will not sync to a CRM or CDP, and how that decision was recorded.

8. What happened the last time a tool or supplier changed?

Explore migration, training, contract exit, lost history and integration maintenance. A recent story is more useful than asking whether switching is "difficult".

Migration stories are platform-specific, so ask how contact history and suppression lists transferred, whether lead-scoring models and nurture flows were rebuilt from scratch, how long the old and new platforms ran in parallel, and what the contract exit actually required.

9. Which part should remain a human decision?

Invite the participant to identify judgement, exception handling or accountability that should not be hidden in an automated rule. Ask what evidence would make them revisit that view.

Human decision points tend to sit around pricing exceptions, high-value accounts, complaint handling and anything with a regulatory consequence. Ask how the platform escalates those cases, who is named as accountable and what the audit trail shows.

Before the session starts

Send a plain-language information sheet covering purpose, activities, data, observers, recordings, sharing, retention and withdrawal. The Service Manual explains how to obtain and retain evidence of informed participation. Separately document the organisation's UK GDPR lawful basis and privacy information.

Keep the invitation free of promotional material. The ICO distinguishes genuine research from activity that includes a sales message or is designed to enable later marketing in its Direct marketing guidance. Do not add participants to a nurture flow merely because they agreed to an interview.

Pilot the guide with a colleague for clarity, then recruit actual or likely users. During analysis, label notes as observation, participant explanation or researcher interpretation. The output should be a set of supported findings and open questions, never invented quotations.

Before you act

Common questions

What kind of interview questions does the article recommend?

It recommends open, neutral questions that reconstruct what happened, asking for one recent instance with a clear start and end. Specific events are easier to check than general opinions, and the guide should leave room for an unexpected answer rather than seeking endorsement of a roadmap.

Why should you ask about corrections or re-entered information?

A correction may reveal field definitions, permissions or ownership problems that automation alone will not solve. Ask for an example without collecting unnecessary customer details, then use it to understand where the workflow breaks down and who is accountable for fixing it.

What must happen before the session starts?

Send a plain-language information sheet covering purpose, activities, data, observers, recordings, sharing, retention and withdrawal. Separately document the organisation's UK GDPR lawful basis and privacy information. The invitation must stay free of promotional material, and agreeing to an interview must not add anyone to a nurture flow.

Which platforms should a marketing automation customer interview cover?

Cover whichever platform the participant actually uses — HubSpot, Marketo, Salesforce Marketing Cloud, Klaviyo, Braze, ActiveCampaign, Adobe Campaign or Customer.io, for example — and the systems around it: CRM, CDP and email service provider. Ask about lead scoring, nurture flows, lifecycle stages, deliverability and attribution, because those are the capabilities a platform review compares.