How-To Guide

How do you explain that a feature is not available to every customer?

Separate deployed from released from announced, keep the condition in the claim sentence, label plan and region unknowns, and give each reader a next step.

9 min read

By the Superdirector editorial team. The pages cited here, including Apple's App Store Connect Help, Stripe billing documentation, Shopify pricing, Leed's product documentation, and the practitioner guides named in the text, were read November 14, 2026. Updated November 14, 2026.

At a glance

How do you explain that a feature is not available to every customer?

Write one availability note per feature before any copy goes out, and put four named facts in it: the task a reader can complete, the audience that has the feature now, the boundary, which is who must not expect it yet, and the next step for each side. Keep the availability condition inside the sentence that carries the claim, because a footnote loses to a headline. A deployment, a release and an announcement are three different events, so a feature that is live for some accounts is not available to readers who cannot reach it. Where you cannot see the reader's plan, region or release phase, name the unknown instead of guessing it, and pair each unknown with the reader's real next step: check a setting, ask the account admin, or watch a dated report. If the availability state changes, widen or pause the wording in the same place the claim lives.

  • Three events, three words: deployed (code runs), released (accounts open it), announced (readers hear it). Copy that merges them generates the ticket pile the wording was meant to prevent.
  • The percentage rule: a rollout figure carries its population. Three of twelve invited pilot workspaces says something honest; a quarter of customers claims more than a small pilot shows.
  • Four entitlement unknowns get named, not guessed: plan, region, release phase, and who holds billing access. Each has its own wording, because a reached quota, a missing role, an unsupported region and a higher plan are different problems with different fixes.
A matrix of three record columns (eligibility, enabled access and observed use) beside a dashed unknown column.

Eligibility, enabled access and observed use stay in separate columns, with a fourth for what you could not check. Original diagram based on the article's Buildside three-record split.

Superdirector

Make it work for your brand

Find it. Remake it. Post it. Superdirector turns the viral formats in your niche into your brand's carousels, X posts and video scripts, and keeps the calendar full.

What the availability note owes each reader

A feature that not every customer can reach creates one written risk: a reader who believes the promise and then cannot complete it. The remedy is a note per feature carrying four named facts: the task a reader can complete, the audience that has the feature now, the boundary, which is who must not expect it yet, and the next step for each side. Everything else, percentages, phases, rollbacks, is an extension of those four facts.

  • The task: what an eligible reader can do today, in the environment they will use.
  • The audience: which accounts, roles, plans or regions the check verified.
  • The boundary: who stays outside the current availability, stated as information rather than as fault.
  • The next step: what an enabled reader does, what an eligible-but-unenabled reader does, and what a reader outside the boundary does instead.
  • The unknowns: plan, region, release phase or billing access the writer cannot verify, named as unknown rather than rounded into a claim.

The rest of this article builds that note in order: separation of the three events behind one word, the three-record check that produces the four facts, the sentence rules that keep the condition attached to the claim, the labelled copy example, the next steps per reader state, and the maintenance duty that keeps the note true after the announcement.

Separate deployed, released and announced

The one word available hides three events, and writers who merge them inherit tickets. Coded Female's separation is the cleanest: deployment puts code into the environment where it runs, release makes a capability accessible to its intended users, and an announcement tells an audience about the change. A feature can be fully deployed, partially released, and loudly announced at the same moment, which is precisely the state this question's wording has to survive.

Featurebase draws the same line with a practical consequence: users may or may not be able to see code that reached production, so a changelog that says the feature is live is accurate for some readers and misleading for the rest. The copy you publish speaks to accounts, not to infrastructure: write against who can reach the feature today. Apple's own phased release is a worked example of the separation: the update ships gradually over seven days on Apple's published schedule, reaching a sample of devices automatically, while anyone can still fetch the update by hand, so percentage copy that ignores manual downloads overstates the boundary.

Record eligibility, enabled access and observed use

Before any availability sentence, Buildside's three-record split keeps the facts separate: eligibility, which describes the group allowed to receive the feature under the current plan; enabled access, which describes who has received it; and use, which describes who completed the task during a defined period. The records stay distinct because they answer different questions, and a claim that mixes them inflates the release: enabled access is not use, and use without the period or the population cannot settle anything.

The check that fills the records is task-level, not button-level: seeing a button is weaker evidence than completing the action it promises, and a screenshot from a demonstration account can miss a difference in customer access. Whatever the check does not cover, record as unknown, in a column of the note, so the writer's remaining claims can narrow to what was verified. A note with named unknowns is honest copy inventory; a note without one is a draft waiting to be wrong.

Put the availability condition in the claim sentence

The condition travels with the claim or it does not travel at all. Doc Holiday's release-note guidance rejects both phrases that read naturally in a partial state: now available, which implies universal access the record does not show, and coming soon, which implies the code never shipped. The replacement set is operational rather than promotional: rolling out, piloting on named accounts, beginning a phased release to the plan that holds it.

The first-sentence rule exists because readers and engines quote openings, not footnotes: availability stated in the last paragraph is availability nobody sees. Percentages carry their population and unit in the same sentence, because a share of invited pilot workspaces and a share of signed-in users describe different access patterns, and the same number printed against the wrong denominator overstates the release. Migma's email rule compresses the whole habit: an available-now statement is a statement about access, so the material restriction sits beside the promise its limits belong to.

Label the entitlement unknowns explicitly

An entitlement is a customer's access to a feature, which is a record the writer usually cannot see at print time: the reader's plan, region, release phase and billing authority sit on the vendor side of the boundary. The honest label names the unknown and routes the reader instead of rounding it into a claim the record does not carry.

  • Plan unknown. Name the plan the feature sits on, and name the route that settles it: the billing screen or the account admin. Leed's own docs keep the two failure families apart precisely so a reader who cannot use a surface knows whether their role or their plan needs to change.
  • Region unknown. Print the supported regions, or borrow the vendor's own wording: Shopify's pricing page tells its customers plainly that some features vary by country.
  • Release phase unknown. Describe the phase, the named audiences it was verified on, and the observation window that decides expansion, rather than a date nobody owns.
  • Billing authority unknown. Say who holds the upgrade decision, because a reader without billing access needs a different sentence than one with it.

The unknowns do not share one sentence. ScreensDesign's entitlement study pairs each gated action with its own reason, because a reached monthly limit, a missing organisation role, an unsupported region and a higher plan are different problems that need different wording, and Leed's upgrade prompt stops shy of checkout, which keeps the availability line an answer rather than a funnel. Where the gate is visual, tone still matters: the wall reads as an available upgrade, never as a scolding.

Write the availability-labelled copy example

The example below is invented and labelled as invented: a fictional scheduling product, called Standby here, is piloting one feature, a queue-hours report, with named accounts. Nothing in it is a measured result of any real product.

  • Availability-labelled post (plan and phase unknowns named). Now piloting the queue-hours report: workspace owners with reporting on Growth plans, in the pilot group we confirmed this week in the web app. If your workspace has not received it yet, your admin can check Settings, then Reports, or you can join the dated review on November 28. Other plans and other regions are outside this pilot; the queue views you already have stay there.
  • Why it carries what it carries. The audience sentence names the verified plan and role; the unknown slots (regions beyond the pilot, billing authority) are named as outside-the-pilot and as an admin route rather than guessed; the population behind any count stays out because this example claims no count; and each reader state gets its own next step, all in the claim sentence.
  • The real-world pattern the example follows. Products print availability as a label beside the feature, not as a story about it: Shopify's plan table attaches a launch-window feature to its tier rows, and Superdirector's pricing page marks AI video with its upcoming, not available yet marker on both plans.

The same facts print differently per channel, and the note keeps them consistent: the changelog line gets the claim sentence and its boundary; the caption gets the claim sentence trimmed to the audience and the next step; the update email gets the boundary beside the promise, because a restriction left to a footer while the headline promises everyone is the support ticket the note exists to prevent. Where the availability line needs testing, send the destination through the role's own path, not merely through the URL.

One availability sentence split into four labelled parts, with an unknown marker under the boundary.

One availability sentence carries four parts (task, audience, boundary and next step) with an unknown marker where access cannot be verified. Original diagram based on the article's invented Standby example.

Give each reader state a next step

The note sorts readers by what it leaves them with, and each state's next step is written rather than implied. Enabled readers receive the task and the help route. Eligible-but-unenabled readers receive the enablement route, and if enablement is automatic they receive the state they can verify (the setting, the badge, the account flag) rather than a promise. Readers outside the boundary receive the plain boundary and the closest alternative; no wording needs to punish them for holding the wrong plan or the wrong role. Readers on the wrong plan receive the upgrade named as the plan name or the capability it unlocks, not as a guilt line.

The next step is also the honesty instrument: SupportMe's reply structure puts the concrete next option in every correction, up to and including an honest statement that no workaround exists, and the same bar applies to the availability line before it ever reaches a reader. A next step nobody can take is a claim with its condition removed.

Keep the note current: owner, review trigger and pause wording

The note is a record to maintain, which is the part the templates usually leave to nobody: StackPractices' release template prints availability as a named field on every external note, and its rollout FAQ keeps the two phases apart, with the deployment told to internal teams and the enablement told to customers. Extend that field so it survives contact with change: an owner who maintains it, a review trigger (a plan change, a region change, a phase change, an audience expansion), the date of the last check, and the check that stands behind the claims.

When the facts change, the wording moves in the same place the claim lives: widen it with the new fact and the new population, or pause it with a dated follow-up that says what happened and what to do now. A paused or rolled-back feature keeps its note; removing the claim without removing the interest is how a quiet rollback manufactures its own tickets, since the announcement's audience learned the feature exists and will go looking for it. Sequencing stays the rule after launch as before it: widen the announcement as access widens, not before.

A four-node loop linking owner, review trigger, check date and the widen-or-pause move.

The note is a record to maintain: owner, review trigger and check date, then widen or pause when the facts change. Original diagram based on the article's StackPractices release-template field.

Where Superdirector fits and where it does not

Superdirector (ours) runs social media work as an operated service, and its own pages practise the labelling this article teaches: the pricing page carries one marker for what is not shipping yet, an upcoming, not available yet note beside AI video on both plans, and its product notes list fully autonomous channels and AI video as upcoming rather than as shipped.

The plan boundaries are printed rather than implied: the Starter plan is $29 a month, MCP only, for hands-on work from Claude or ChatGPT; the Growth plan is $199 a month for one social account its AI runs. Publishing on Growth runs under the account-level standing posting permission the account sets: once posting is turned on for that account, eligible Instagram carousels and single X text posts publish by themselves at their next free time, and video scripts stay off the calendar. What the service does not take over is the record keeping above: the availability note, its check date, its review trigger and its pause decision belong to the team that owns the feature, and a copy discipline is the writer's job wherever the posts are published.

Frequently asked questions

How do you explain that a feature is not available to every customer?

Write the availability into the claim itself and name the boundary: which audience has the feature now, what task they can complete, who should not expect it yet, and what each side should do next. A deployment, a release and an announcement are three events, so say what is true of the readers you are addressing now. If you cannot see the reader's plan, region or release phase, name that unknown explicitly, and pair every percentage with its population. When the state changes, update the wording where the claim lives.

What is the difference between deployed, released and announced?

Deployment puts code where it runs; release makes a capability accessible to its intended users; announcement tells an audience about it. They can land at different times, and a feature missing from one account does not prove the release failed. The distinction matters for copy because we shipped it and everyone can see it are different claims: one describes your infrastructure, the other describes a reader's account. Write the customer-facing copy against the reader's account.

What words should copy for a partial rollout use?

Operational, progressive wording: we are rolling out, we have begun a staged release to named accounts, if you are part of the pilot program. Avoid now available, which implies universal access, and coming soon, which implies the code never shipped. Put the availability condition in the first sentence, not a footnote, and keep it there when the claim gets quoted or clipped. Where the writer only knows a configured percentage, print the percentage with its population and unit, for example a share of invited pilot workspaces rather than a share of all customers.

How do you write availability copy when you cannot see who has access?

Name the unknown and route the reader. If plan state is unknown, the availability line names the plan the feature sits on and how to check which plan applies (the account admin or the billing screen). If region is unknown, print the supported regions or the sentence the vendor itself uses, such as features varying by country. If the release phase is unknown, describe the phase, its named audiences and the observation window that decides expansion. The gate should be written from entitlement rules rather than marketing polish, and a reached limit, a missing role, an unsupported region and a higher plan should not share one sentence.

How do you reply to a customer who expected a feature their plan does not include?

Restate what the customer expected, acknowledge why the expectation was reasonable, explain what the product does under the current conditions, and give one concrete next step, which may be a workaround, the upgrade route, or an honest statement that no workaround exists. Blame phrases like that is just how the system works are out. The reply should name the plan boundary precisely and correct the wording that caused the expectation, because a pricing line that lets reasonable customers infer a feature repeatedly is a wording defect, not a reader defect.

Is it wrong to say a feature is coming soon?

Coming soon is an availability claim about a future state, so it needs a different job than a partially shipped feature: use it for capability that is announced and not built or not enabled for anyone, and never hide unshipped code behind it. Products that print their own labels set the pattern: Superdirector's pricing page marks AI video as upcoming, not available yet, and its product notes say the same of fully autonomous channels and AI video. Whatever the label, keep it honest and current, because a label nobody acts on trains readers to ignore availability lines altogether.

How does Superdirector explain which parts are available on which plan?

Superdirector (ours) publishes its two plans with their boundaries printed on the pricing page: the Starter plan at $29 a month is the route for hands-on work from Claude or ChatGPT, and the Growth plan at $199 a month carries one social account its AI runs, which finds and remakes ideas, fills the calendar, and, once posting is turned on for the account under the account-level standing posting permission that plan sets, publishes that account's eligible Instagram carousels and single X text posts by itself. Its own labels mark what is not shipping yet: AI video carries the upcoming, not available yet marker on the pricing page. What the service does not take over is the availability note this article teaches: the record, its check date and its review decision belong to the team that owns the feature.

Related practical guides

Write the three records for one feature today

Pick the feature customers keep asking about, fill eligibility, enabled access and observed use with the unknown column named, and put the condition inside the claim sentence.