Kelly Addington

Qualitative UX researcher for regulated enterprise sectors.

I run AI-moderated research, moderated interviews and usability testing for financial services, telecommunications, health and retail clients.

Ten years in product design and design operations before moving into research full time. Based on the Sunshine Coast, Queensland, working remotely across Australia and New Zealand.

Services

AI-moderated studies and Insight Streams

Weekly or one-off AI-moderated interview streams. I write the discussion guide, set the screener, review participant quality and produce the report. Suited to fast-turnaround questions where a client needs 40 to 60 responses in days rather than weeks.

Usability testing

Moderated testing of live products, prototypes and communications. Participants work through real tasks while I observe, probe and record. Clients get a prioritised list of issues with a clip against each one.

Moderated interviews

One-on-one sessions with customers, including people with access requirements and participants in sensitive situations. I handle recruitment, moderation, synthesis and reporting.

Longitudinal and diary studies

Multi-week digital diaries with follow-up interviews, for behaviour that only shows up over time.

AI quality review

Review of AI-generated participant responses and AI-drafted research outputs. I check for fabricated quotes, low-quality participants and findings that don't hold up against the transcripts.

Work by sector

Financial services: Retail banking, lending, payments, superannuation and investment products
Telecommunications: Consumer mobile, broadband and device customers
Health: Blood and plasma donation
Retail and property: Grocery and general merchandise, residential property platforms

How I work

Every study starts with a short conversation to confirm the research question before I write a guide.
Reports are written for the stakeholder who has to act on them. Findings first, evidence second, method in an appendix.
I come from product design, so findings are written as decisions a team can act on, not observations left for someone else to interpret.
Every quote in a report is traceable to a transcript.
I say no to AI moderation when the topic calls for a human. Vulnerable participants and emotionally sensitive subjects get a moderated study.

Selected work

Weekly customer insight stream

Telecommunications

An ongoing AI-moderated stream for a national telco, with a new topic each week: fibre upgrades, device upgrades, early churn, wearables. Each week the client receives a detailed report covering behavioural findings, demographic data and opportunities for further exploration.

Voice AI accessibility study

Financial services

Research with customers with access requirements into an automated voice system. Identified where the system failed users with hearing, speech and cognitive access needs, and what the client needed to change.

Usability testing for a lending product

Banking

Two rounds of moderated testing across an email, a web page and a booking tool. Delivered a prioritised list of usability issues with a video clip for each.

Health and beauty shopper research

Retail

A mixed-method study combining moderated interviews with an AI-moderated survey of the wider customer base. Findings were cut by shopper segment and delivered as a single report with tagged video clips for each theme.

Proposition testing for a new investment product

Financial services

Moderated interviews with existing customers to test a draft proposition for a new investment product before build. Tested the product name, the plain-language description, a set of behavioural hypotheses and a candidate feature list. Delivered a prioritised feature set and guidance on the language customers used to describe their goals.

Category page conversion research

Banking

Two-phase study on the acquisition page for a bank's everyday transaction accounts. Phase one paired discovery interviews with mobile-first usability testing of the live page to find where customers hesitated before applying. Phase two ran unmoderated testing of prototype variants with 50 participants to pick the design direction. Two reports, one per phase, with design recommendations.

Discovery interviews for a consumer app

Early-stage startup

Twenty-five short interviews with neurodivergent and perimenopausal women, the app's target audience, covering pain points, the fit of the current feature set, pricing and the free-trial sign-up. Reported back on messaging clarity, feature priorities and the barriers to paid conversion.

Contact

kelly@kellyaddington.com

Available for project work and ongoing engagements.