When new users struggle to get started, a redesigned product tour is an attractive fix. It may also be the wrong fix. The obstacle could be permissions, missing data, unclear terminology or a first task nobody understands. Here is how to hire a UX partner to diagnose the problem before commissioning a redesign.
Define the first useful outcome, not the welcome screen
Identify what a new user must accomplish to experience real product value: complete a workflow, import meaningful data or collaborate with a colleague. Clicking through a tutorial is not necessarily evidence of success.
Separate self-serve trial users from enterprise administrators configuring SSO, permissions and integrations. Their goals and constraints differ. A credible UX agency should propose research segments rather than one universal onboarding flow.
- User roles: end user, administrator, implementer and decision-maker.
- First meaningful task and dependencies that block it.
- Known friction from support tickets, product events and interviews, with data limitations noted.
Buy a diagnostic study before buying new screens
Ask candidates how they would combine product analytics, support signals and observed task performance. Funnels may identify where users drop out, but rarely explain the cause by themselves. Interviews without tasks can overrepresent what users say rather than what they do.
The UK Government Service Manual recommends realistic tasks that do not reveal the intended solution. This is especially useful when a product team is attached to a particular onboarding design.
- Realistic tasks based on user goals rather than menu labels.
- Representative recruitment across roles and accessibility needs.
- Safe test environment, informed consent and appropriate data handling.
Separate research, design and engineering responsibilities
A research consultancy can identify causes and prioritise hypotheses. A product-design team can prototype alternative flows. An implementation partner may be needed for API, permission or backend changes. Some agencies cover all three, but the proposal should separate the work.
If your internal team cannot build the recommended changes, budget for implementation or select a supplier with named engineering responsibility. A slide deck of insights does not fix a broken data import.
- Research: moderated tests, observation, analysis and prioritised findings.
- Design: flows, prototypes, microcopy and accessibility considerations.
- Engineering: instrumentation, implementation, QA and release support.
Choose measures that reflect real user progress
Consider first-task success, time to meaningful value, support dependence and return usage. Segment by role and account type. An uplift after release may also reflect changes in acquisition mix or sales-assisted onboarding; ask how the agency will account for these confounders.
Nielsen Norman Group’s guidance on task scenarios advocates realistic, actionable tasks without hints. This makes the observation more meaningful than simply asking participants whether they liked a screen.
- Primary outcome defined before testing: task success.
- Supporting observations: time, errors, help requests and confidence.
- Baseline and follow-up plan, including what can and cannot be attributed to the redesign.
Scope the budget in phases
A useful first engagement can include a research plan, participant recruitment, moderated sessions, synthesis and a prioritised decision backlog. Cost depends on recruitment difficulty, number of user roles and markets, product complexity and whether design work is included.
Request separate line items for discovery, prototyping, experiment design and implementation. A full visual redesign may be unnecessary if users are blocked by an account permission or integration step.
- Who recruits participants and pays incentives?
- How many review cycles and design iterations are included?
- Who owns research materials, prototypes and the final prioritisation decision?
Watch for these warning signs
Be cautious when an agency promises a specific activation lift before understanding the product or proposes a standard product tour without seeing real tasks. Ask how each recommendation connects to observed evidence and which assumptions still need testing.
Check how the study will include users with different levels of experience, accessibility needs and technical environments. A product that works only for the test moderator is not ready.
- UI mockups without a research or measurement plan.
- Findings presented as facts when they are hypotheses.
- No ownership of measurement after implementation.
Write a brief that lets agencies demonstrate fit
If you primarily need UI production, explore a UX/UI design agency, but verify its research capability. For formal accessibility concerns, see the accessibility agency selection guide.
Describe your roles, first-use task, available evidence, decision deadline and engineering capacity. You can compare specialist categories in the agency directory.
- What exact task should the study explain?
- Which user groups are essential to recruit?
- What decision must be possible after phase one?
Use Mosai to compare agencies on research method, SaaS product understanding and implementation fit rather than a polished portfolio alone.