Blog › ICP guides

UX writer retainer: microcopy audit, onboarding advisory, and content style guide governance on monthly retainer

July 24, 2026 · ~18 min read

A B2B SaaS product has a 7-step first-run activation sequence. The product team has instrumented the sequence with analytics and knows the completion rate at each step. Step 1, entering a workspace name, completes at 91%. Step 2, inviting team members, completes at 84%. Step 3, connecting the first integration, drops to 43%. The product team has hypothesized three causes: the integration requires OAuth, which adds friction; the integrations list displays 47 options with no filtering; and the tooltip guiding users through the step says “Connect your tools to get started.” The product team has run two redesign sprints trying to improve step 3 completion. Neither sprint changed the completion rate meaningfully.

The UX writer reviewing the activation flow identifies the problem in the tooltip copy within the first read-through. “Connect your tools to get started” fails three of the four criteria for effective instructional microcopy. It is not specific: it does not tell users which tool to connect or why. It is not motivating: “to get started” implies the user is not started yet — the message is implicitly telling the user that everything they did in steps 1 and 2 was preamble, not progress. And it provides no recovery path: if the user does not have admin access to connect an OAuth integration, the tooltip gives no indication of what to do instead. The tooltip that the product team has been shipping through two redesign sprints is actively working against step 3 completion by framing the step as a prerequisite rather than a capability unlock.

The revised tooltip: “Connect Slack, Jira, or Salesforce to pull your team’s activity into [Product]. Skip for now — you can add integrations any time from Settings.” Completion rate at step 3 increases from 43% to 67% in the week following the copy change. No redesign. No engineering work. Two sentences. The product team attributes the improvement to the redesign sprint work; the UX writer’s copy change is the only variable that changed between the previous week and the measurement week. This is the specific dynamic that makes UX writing retainer hours systematically undervalued: the output is two sentences; the analytical work that produced those two sentences — the heuristic audit, the completion rate analysis, the four competing drafts, the A/B annotation in Figma — is invisible.

A UX writer or content designer on monthly retainer operates in this analytical layer continuously. Between the visible copy changes that ship in product releases lies a continuous cycle of interface copy auditing, onboarding flow review, style guide governance, and cross-functional alignment work that the client never directly observes — and that is exactly the work that prevents the interface copy from accumulating the compounding inconsistencies and usability failures that require expensive redesign sprints to correct.

Microcopy audit advisory

Microcopy audit advisory is the systematic review of interface copy against established heuristics and product-specific criteria. It is the UX writing equivalent of a code review: the interface copy ships continuously, and without an ongoing audit function, inconsistencies, deprecated terminology, and usability failures accumulate in the product at the rate of every sprint. The UX writing retainer’s audit function catches those failures at the copy layer before they require engineering investment to fix.

Error message heuristic review

Error messages are the most consistently audited category of interface copy because error states represent the highest-stakes user moments and are the most likely to be written by engineers rather than writers. Nielsen Norman Group’s error message heuristics provide a practical audit framework: error messages should be expressed in plain language (no error codes, no jargon), precisely describe the problem (not “something went wrong” but “we couldn’t save your changes because the file name already exists”), and constructively suggest a solution (tell the user what to do, not just what happened).

In a typical SaaS product audit, a UX writer reviewing error messages against these three criteria finds a consistent pattern of failures in the same categories. Validation errors in form fields frequently use technical field names from the database schema rather than the label names displayed in the UI: “organization_id cannot be null” where the UI shows “Company” as the field label. API error states surface HTTP status codes without interpretation: “Error 403” where the user experience requires “You don’t have permission to edit this project. Ask your workspace admin to grant you Editor access.” Authentication errors frequently use overly cautious generic language that obscures the actual problem for legitimate users: “Invalid credentials” where the account exists but the password has expired, and a message that says “Your password expired 14 days ago. Reset it here.” would be both more helpful and no less secure.

A comprehensive error message audit for a product with multiple authenticated surfaces, multiple user roles, and multiple integration points typically covers 60 to 120 distinct error states. Auditing each state against the heuristic criteria, noting the failure mode, drafting a replacement message, and documenting the copy decision in a format that the engineering team can implement takes approximately 15 to 25 minutes per error state. An audit covering 80 error states is 20 to 33 hours of focused work that produces an audit document and a copywriting spec. The client sees the spec delivered; the hours of methodical review that produced it are not directly visible.

Empty state and zero-data copy review

Empty states are the interface moments where a user has landed in a view that has no content to display, either because they have not yet created any data (the new user seeing an empty list for the first time) or because their current filter or search returned no results (a user who has filtered a list to a combination that matches nothing). These states are frequently the first experience a new user has with a core product surface, and they are routinely under-specified in design because the team’s design process focuses on the populated state.

In one content design engagement, a UX writer auditing empty states across a project management product found four distinct voice tones in use across the same product surface. The empty state for a task list with no tasks said “No tasks yet. Create your first task to get started.” The empty state for a project with no members said “This project has no members.” The empty state for a search result with no matches said “No results found for [query].” The empty state for a reporting view with no data said “There’s nothing here yet. Add some projects and come back.” The four strings were written by four different people at four different times with no shared pattern. The reader moving between product surfaces encounters a product that sounds like four different writers — one encouraging, one declarative, one neutral, and one conversational — with no consistent persona.

The content design advisory task is establishing an empty state copy pattern that produces a consistent voice across all surfaces. The typical framework identifies three empty state types: first-time empty states for users who have not yet created content (action-oriented, specific about what to create first, motivating about the value of creating it); filtered empty states for users whose current filters produce no results (explain the no-results condition, suggest how to modify the filter to find what they are looking for); and error-condition empty states where a technical problem prevents content from loading (acknowledge the problem, give a recovery action, avoid implying user fault). Auditing all empty states against this three-type framework, drafting consistent replacements, and adding the pattern to the style guide is approximately 6 to 10 hours of work that produces one section of style guide documentation and a set of copy replacements. The client delivers the copy replacements to engineers and may never see the style guide section. Log both.

CTA and button label precision review

Button and call-to-action labels are the interface copy elements with the most direct, measurable impact on user behavior and the most consistent pattern of failure across products. The failure pattern is specific: labels that describe the system state rather than the user’s action, labels that use the same word for actions with materially different consequences, and labels that omit the specificity that would distinguish a destructive action from a reversible one.

In one CTA audit, a UX writer reviewing a multi-step form submission flow identified a button labeled “Continue” on the final step of a cancellation flow. The button appeared visually identical to the “Continue” button on steps 1 through 4 of the same flow — same size, same color, same position. But where the previous four “Continue” buttons advanced the user to the next step in the cancellation request, the final “Continue” button submitted the cancellation irreversibly and immediately terminated the user’s subscription. Users who clicked “Continue” at the end of step 5 expecting to see a confirmation screen — because every other multi-step form in the product showed a confirmation screen at the final step — instead found themselves logged out with a “Your subscription has been cancelled” confirmation email in their inbox.

The copy recommendation was to change the step 5 button from “Continue” to “Cancel my subscription” and to add a consequence statement immediately below the button: “You’ll lose access to [Product] immediately. This cannot be undone.” The label change and consequence statement together transformed an ambiguous navigation action into an explicit commitment action. The copy change required zero engineering work — it was a string replacement in the codebase. The advisory work that identified the problem, diagnosed the failure mode, drafted and evaluated multiple label candidates, and documented the recommendation took 3 hours. The string replacement took 15 minutes.

Onboarding flow copy advisory

Onboarding flow copy advisory is the UX writing retainer’s highest-impact ongoing function. The onboarding flow is the interface sequence where the largest percentage of user churn occurs, where the product’s value proposition must be demonstrated rather than described, and where the copy that guides users through first-time setup is most likely to drift out of sync with the actual product behavior as features change and the onboarding flow is not correspondingly updated.

Tooltip and coach mark copy precision

Tooltips and coach marks — the overlays, popovers, and highlight callouts that guide users through a product’s first-run experience — are the most common place where onboarding copy fails to match the actual product behavior. The failure mechanism is systematic: the onboarding copy is written during the feature development cycle, when the feature is in its designed state. After launch, the feature evolves through bug fixes, UX iterations, and scope adjustments. The tooltips are rarely updated to reflect each iteration. Over time, a significant percentage of onboarding tooltip copy is describing a version of the feature that no longer exists.

In one onboarding audit, a UX writer reviewing a 7-step activation sequence found that step 3’s tooltip said “Click Add Member to invite your team.” The button in the current product was labeled “Invite People.” The button had been relabeled six weeks earlier as part of a nomenclature consistency effort. The tooltip had not been updated. A new user following the tooltip instruction would look for a button labeled “Add Member”, not find it, and potentially abandon the step — because the copy told them to do something the current interface did not support in the described form. A similar audit of steps 4 through 7 found two additional instances of copy referencing deprecated UI labels. All three were a string replacement in the codebase; identifying them required reading the tooltip against the live product interface for each step. That systematic comparison is the audit work, and it takes approximately 20 to 40 minutes per step depending on the complexity of the feature being described.

The ongoing maintenance function for onboarding tooltip copy is reviewing the activation sequence against the live product interface on a regular cadence — monthly at minimum, more frequently during periods of active product development. The review produces either a no-change confirmation (the copy matches the current product; log the review) or a list of deprecated references requiring updates. Neither output is a visible deliverable in the way a new feature launch is — but the second outcome, if not caught, degrades the activation completion rate silently until someone notices the step 3 drop-off in analytics and initiates a redesign sprint.

First-run experience scripting

First-run experience scripting covers the complete sequence of interface copy that a new user encounters from signup through first value — the signup confirmation email, the welcome screen, the activation checklist, the contextual tooltips, the empty state copy in the main product surfaces, the first-session nudge email if the user does not complete activation within a defined window, and the “you’re set up” milestone confirmation when activation is complete. The combined copy across these touchpoints constitutes the user’s first impression of the product voice, and inconsistencies across the sequence create a fragmented experience even when the individual pieces are well-written.

In a first-run experience audit for a project management product, the welcome screen said “Welcome to [Product]. Let’s build something great.” The activation checklist said “Complete these steps to get your workspace ready.” The first tooltip in the main app said “You’re almost there — just a few more things to set up.” The milestone confirmation said “You’re all set! Your workspace is ready.” The first-session nudge email said “Hey, you haven’t finished setting up your account yet. Complete your setup here.” Each piece was grammatically correct and individually reasonable. Together, they communicated three different things about the user’s progress state, used four different terms for the same concept (workspace, account, setup, configuration), and oscillated between encouraging (“let’s build something great”) and mildly accusatory (“you haven’t finished”).

The scripting advisory task is auditing the full first-run sequence as a single experience, identifying the copy inconsistencies, and producing a revised sequence draft where the voice is consistent, the progress framing is consistent, and the product terminology is consistent. For a first-run sequence covering 12 to 18 distinct interface moments (screens, emails, tooltips, notifications), the audit and revision takes 8 to 14 hours. The client receives the revised copy spec. The analytical work that identified the inconsistencies across 18 touchpoints — and the judgment calls about which inconsistencies mattered most for the user experience — are not visible in the spec.

Conversational UI and chatbot response advisory

Conversational UI copy advisory covers the design of chat interface responses, automated message scripts, and in-product notification copy for products that use chat-native interaction models. As more B2B SaaS products incorporate AI assistants, onboarding chatbots, and support automation, the quality of the conversational copy becomes as important as the quality of the visual interface copy — and it requires a distinct skill set from traditional microcopy writing.

Chatbot response pattern design

Chatbot and AI assistant response design requires decisions about response length, tone calibration, error handling language, and recovery patterns when the system cannot fulfill a request. The most common failure modes in chatbot copy are over-apologizing (“I’m so sorry, I’m not able to help with that right now” as the response to every out-of-scope request), under-explaining (responses that describe what the system did but not why, producing confusion when the result differs from the user’s expectation), and false precision (responses that use confidently specific language for outputs that carry significant uncertainty, training users to over-trust outputs that should be reviewed).

In one conversational UI advisory engagement, a UX writer reviewing a B2B assistant’s error handling responses found that 23 of the 31 distinct error conditions produced the same response: “I wasn’t able to complete that. Please try again or contact support.” The response pattern meant that a user who asked the assistant to do something outside its capability range received the same message as a user who encountered a transient API timeout. Both received “I wasn’t able to complete that” with no indication of whether the problem was temporary (retry immediately) or permanent (the feature does not exist, contact support to request it). The advisory work was designing a three-tier error response pattern: transient errors that warrant an immediate retry (“There was a temporary issue. Try again in a moment.”), scope-limitation responses for requests the system is not designed to handle (“I can’t [specific action] yet. Here’s what I can help with: [list of relevant capabilities].”), and escalation responses for issues requiring human support (“This needs a look from our support team. Here’s how to reach them: [link].”). Designing the three-tier pattern, writing the response templates, and documenting the error routing logic for the engineering team was 6 hours of advisory work.

In-product notification copy calibration

In-product notification copy — the banners, toasts, alerts, and push notifications that communicate system events to users — is one of the categories most frequently written by product managers or engineers rather than content designers, and one of the most likely to violate voice consistency guidelines and notification fatigue principles simultaneously. Products with active development cycles tend to accumulate notifications because each new feature ships with at least one notification informing users of the new capability, and the notifications are rarely pruned as the feature matures from new to established.

In one notification audit, a UX writer catalogued all in-product notification copy for a SaaS platform and found 47 distinct notification strings active in the product. Eleven of the 47 notifications were for features that had been generally available for more than six months. Three notifications referred to features that had been deprecated and removed from the product. Four notifications used a tone that was more urgent than the underlying event warranted (“Action required” for an optional profile completion prompt). The advisory recommendation was to archive 11 mature-feature notifications, remove 3 deprecated-feature notifications, reclassify 4 over-urgent notifications as informational rather than action-required, and establish a notification lifecycle policy: every new feature notification is tagged with an expiration condition (ship immediately; archive when feature reaches X% of user adoption). The audit and recommendation document was 4 hours of work. Implementation was a series of configuration changes.

Content style guide governance

Content style guide governance is the UX writing retainer function that produces the least visible work per hour and the most compounding value over time. The style guide is the reference document that allows product, engineering, and design contributors to make interface copy decisions consistently without consulting the UX writer for every string. A well-maintained style guide means the UX writer’s judgment is distributed across the team; a poorly maintained one means the team defaults to individual judgment, producing the four-tone empty state problem and the deprecated button label problems described above.

Voice and tone documentation

Voice and tone documentation is the foundational style guide component: a description of how the product communicates that is specific enough to produce consistent copy decisions but flexible enough to accommodate the different registers required by different user moments. A product that delivers a confident, direct voice in its navigation labels and onboarding tooltips needs to be able to shift to a more empathetic, less directive tone in error states and cancellation flows. The style guide needs to specify not just the product’s general voice but the tone calibration for different emotional contexts.

The common failure mode in voice documentation is defining the product voice at a level of abstraction that is not decision-guiding in practice. A style guide that says the product voice is “friendly, professional, and clear” does not help a product manager decide whether the empty state should say “No tasks yet” or “You don’t have any tasks yet” or “Looks like you’re all caught up!” All three are friendly, professional, and clear. The effective style guide entry specifies the rule at the copy-decision level: “Use second-person (‘you’) when addressing the user directly. Avoid first-person from the product (‘we’, ‘our’) in interface copy — reserve it for email. In empty states, state the condition declaratively before the action call: ‘No tasks yet. Create your first task.’ not ‘You haven’t created any tasks yet.’” That level of specificity can guide a copy decision by someone who is not a professional writer.

Writing the voice and tone documentation to that level of specificity requires deriving the rules from the existing product copy, not from abstract brand principles. The UX writing advisor audits the existing interface copy, identifies the implicit rules the product is already following (and violating), and makes those rules explicit in the style guide. For a product with two years of shipped copy spread across multiple product surfaces, this derivation process takes 8 to 16 hours of analytical work and produces a 10 to 20 page style guide. The document is the visible output. The hundreds of copy examples reviewed and the pattern extraction work that produced the rules are not visible in the document.

Terminology and glossary management

Terminology management is the style guide component that requires the most active ongoing maintenance and produces the most visible impact when neglected. The terminology problem in SaaS products is consistent: the product team uses one term, the marketing team uses a different term, the engineering team uses the database column name, and customer support uses the term that customers use when they file tickets. The four terms refer to the same concept. Users who read the marketing site, onboard through the product, encounter an error message, and then contact support may encounter four different names for the same thing without recognizing them as synonymous.

In one terminology audit, a UX writer catalogued every term used across the marketing site, product interface, help documentation, and support email templates for a core product concept — the primary organizational unit in a B2B project management product. The marketing site called it a “workspace.” The product interface used “workspace” in the navigation but “organization” in the settings panel. The help documentation used “account” throughout. The support email templates used “environment” when referring to the same entity. Users who searched for help with “workspace settings” found help articles that described “account settings” without indicating that “account” and “workspace” referred to the same thing. The recommendation was to standardize on “workspace” as the canonical term across all surfaces, update the 47 help articles that used “account” for this concept, and add “workspace” as the controlled vocabulary term for this concept in the style guide with a note that “organization” and “account” are deprecated synonyms. The audit took 5 hours. The help article updates were a separate project.

Frequently asked questions

Tracking UX writing retainer hours with HourTab

UX writers and content designers on monthly retainer face the most acute version of the invisible-work billing problem. The visible outputs of interface writing work are small: a revised button label, a rewritten error message, a two-sentence tooltip update. The analytical work that produced those small outputs is large: an hour of heuristic audit to identify the failure, comparative analysis of four draft options, alignment work with the product manager and designer, and a style guide update to prevent the same class of error from recurring. On a monthly retainer, the ratio of analytical hours to visible output strings can run 10:1.

When the monthly invoice arrives, clients who evaluate the retainer against visible copy deliverables apply a calculation that systematically undervalues interface writing advisory: “what copy did we receive this month?” If the answer is “14 string replacements and one style guide section,” the invoice feels expensive relative to the apparent volume of output — even though the 14 string replacements emerged from 40 hours of audit, design file review, draft iteration, and cross-functional alignment. The analytical hours are the majority of the retainer; the string replacements are the output of the analysis.

HourTab is built for exactly this billing challenge. Import your time-tracker CSV, and HourTab generates a public retainer-hours URL that your client can bookmark. The URL shows a live view of hours logged against the monthly retainer allocation, with the work log entries visible in chronological order. The client does not need a login or a portal to see where the retainer hours stand. When the invoice arrives, the client has already seen the microcopy audit session, the Figma review, the empty state framework research, and the tooltip copy maintenance review. The hours are not a surprise; they are a record of the content design advisory engagement the client has been following in real time.

The Free plan handles one active retainer: a public share URL, CSV import, and a work log with a progress bar showing hours consumed against the monthly allocation. The Solo plan at $9 per month supports up to 10 active retainers with a custom URL slug, no HourTab branding, CSV export, and email-a-summary for month-end reporting. The Studio plan at $19 per month supports unlimited retainers, a branded subdomain, two team seats, per-client headers, and rollover rules for engagements where unused hours carry forward.

Try HourTab free →