AI for Boutique Fitness Studios: Class Outlines, Captions and Emails

WORKFLOW 01

Flag low fill rates with a formula, then ask AI for options to discuss

When: Weekly: add the booking export to a running sheet. Monthly: review the flags and options, because changing the timetable more often unsettles members.

Working out a fill rate is arithmetic, and a spreadsheet gets it right every time. AI adds value only after the numbers exist: it can list options for each flagged slot and the unknowns each option depends on.

Start with these inputs

An aggregated sheet covering the last four weeks, with these columns: day and time, class type, duration, capacity, bookings for each week, waitlist for each week, fill rate, and two formula flags. FLAG marks a fill rate below your chosen threshold. DEMAND marks an average waitlist at or above your chosen level. Remove member names before export.

Work through the task

1. Calculate fill rate per slot as average weekly bookings divided by capacity. Add a FLAG column for slots below a threshold you choose, such as 50 percent, and a DEMAND column for slots whose average waitlist is 2 or more.
2. Mark missing weeks as 'missing' instead of zero, so a holiday closure does not look like an empty class.
3. Paste only the aggregated table into the studio's approved AI account, using the prompt in the worked example.
4. Ask for two or three hypothetical options per flagged slot. Each option should list the facts it depends on that the table does not contain.
5. The owner and head coach discuss the options. No timetable change is announced to members without the owner's sign-off and the notice period the studio normally gives.

Before you use it

The owner recalculates two fill rates by hand. Any option that assumes facts the table does not contain, such as coach availability, gets struck out.

When something goes wrong

If the table has unmarked gaps, if the AI changes a figure, or if it states a cause as fact, discard the memo. Fix the table and rerun.

A simpler option

Conditional formatting on the fill-rate column handles the flagging, and the options can be talked through at a staff meeting with no AI involved.

WORKFLOW 02

Find lapsed members with a date filter and keep the follow-up personal

When: Weekly.

Finding members with no visit in 21 days is a date filter. Sending named attendance history to an AI tool creates privacy exposure and produces nothing the filter cannot.

Start with these inputs

An attendance export with member ID, first name, join date, last visit date, membership status (active, paused, frozen, or cancelled), whether the member has a future class booked, and their contact preference.

Work through the task

1. Add a column for days since last visit and filter for 21 or more.
2. Exclude paused, frozen, and cancelled memberships, anyone with a future booking, and anyone who has asked not to be contacted.
3. Sort by join date so recent joiners, who may still be forming a habit, appear first.
4. Assign each name to the coach who knows that member best.
5. The coach contacts the member personally in their own words. If AI is used at all, it drafts one generic, name-free check-in template that coaches can adapt. Nothing is sent automatically.
6. Coaches do not ask about health by message. If a member mentions an injury or illness, the coach gives no advice by message and follows the studio's own return-to-training policy.

Before you use it

Before anyone is contacted, the owner confirms the exclusions worked by spot-checking three names against the booking system.

When something goes wrong

If the export dates look wrong, or if a member who opted out appears on the list, stop all outreach until the source data is fixed.

A simpler option

Check whether your booking system already has an inactivity report. A saved spreadsheet filter is the fallback, and no AI is needed.

WORKFLOW 03

Draft a timed class skeleton for the coach to rebuild

When: Event-driven: when a new class format or programming block starts. Coaches reuse their approved plans between blocks, so this is not a per-class task.

A timed skeleton saves time at the blank page. The coach already owns the final plan, so responsibility stays with the person qualified to carry it.

Start with these inputs

Class type, duration, focus areas, the studio's equipment list, room capacity, intended level, and optionally the coach's own exercise library so the draft only uses movements the studio already teaches. No member health information.

Work through the task

1. The coach enters the fields and asks only for timed blocks: warm-up, main blocks with work and rest intervals, and cool-down. The draft should name movements from the coach's library and leave out loads, percentages, and heart-rate targets.
2. The coach replaces or adjusts every exercise to match the studio's equipment, the room layout, and their own programming standards.
3. The coach writes regressions and progressions from their own knowledge instead of accepting the AI's.
4. The head coach approves the plan before it goes on the board.

Before you use it

Check that block minutes add up exactly to the class length, that equipment transitions are realistic for the number of people in the room, and that nothing reads as treatment for an injury or condition.

When something goes wrong

If the draft uses equipment the studio does not have, sets intensity targets, or makes medical-sounding claims, discard it and start from the studio's own template.

A simpler option

A reusable block-timing template in a spreadsheet works well for coaches who already know their exercise library.

WORKFLOW 04

Write spotlight captions that keep the member's exact words

When: Event-driven, when a member agrees to be featured. Realistically once a month at most.

Drafting from facts the owner supplies is low risk, provided the member approves the exact text and photo before anything is posted.

Start with these inputs

Member first name, favourite class, how long they have trained at the studio, a quote in their own words, and their written consent covering the caption, the photo, and the platforms where it will appear.

Work through the task

1. The owner types in only what the member agreed to share. Health conditions, weight, body measurements, and before-and-after comparisons stay out.
2. AI drafts two short caption options and keeps the quote word for word.
3. The owner sends the chosen caption and photo to the member for approval.
4. Post only after a written yes from the member, and record the consent date and platforms.

Before you use it

Compare the quote character by character with the original, and check that the caption promises readers no results.

When something goes wrong

If the member edits the caption, use their version. If they decline or go quiet, drop the post. Never publish a draft the member has not seen.

A simpler option

A fill-in caption template such as 'Meet [name], a regular at [class] since [month]' may be all you need.

WORKFLOW 05

Turn scattered repair notes into a checklist, with staff setting safety status

When: Weekly for the checklist. A possible safety issue is handled the moment it is reported, before any AI step.

Sorting messy notes into a tidy list is a good formatting job for AI. Deciding whether equipment is safe to use belongs to staff. So the safety status is entered by a person, and the AI only sorts by it.

Start with these inputs

Typed notes for each item, each with: equipment, problem, date noticed, who reported it, location, and a status that staff set themselves (taken out of use, usable but needs repair, or cosmetic).

Work through the task

1. As soon as anything is reported that could injure someone, staff take it out of use and label it physically.
2. Staff assign each note one of the three statuses. If a note has no status, it is treated as taken out of use until a person decides otherwise.
3. The owner pastes the notes and asks AI to sort them by the status staff assigned, keep each status exactly as written, and list missing details such as date or location.
4. The owner sets priorities and due dates. Repairs follow the manufacturer's instructions or are done by a qualified technician.
5. The checklist goes to staff with a named person and a due date for each item.

Before you use it

The owner counts the items in the notes against the items in the checklist, and confirms that no status has changed.

When something goes wrong

If an item is missing, or a status has been changed or softened, rebuild the checklist by hand from the original notes.

A simpler option

A shared spreadsheet with a status dropdown and a due-date column, filled in by whoever spots the problem.

WORKFLOW 06

Draft a three-email welcome series from confirmed studio facts

When: Built once. Reviewed quarterly, or whenever offerings, prices, or policies change. The email tool sends the emails, so this is not a daily task.

Drafting three emails from a fact list is a sensible writing job. The main risks are invented offers, promised results, and health questions that belong in the studio's intake form instead of an email.

Start with these inputs

Class types, booking steps, first-visit logistics such as parking, arrival time, and what to bring, the exact cancellation policy wording, a contact route, and the intended gap between emails.

Work through the task

1. The owner supplies the facts as bullet points and pastes the actual cancellation policy text.
2. AI drafts email one on first-visit logistics, email two on choosing a class, and email three on how to reach a coach. Any missing fact is marked [NEED].
3. The head coach reviews every exercise reference. The series carries no nutrition advice, and health questions point to the studio's existing intake form.
4. The owner approves all three emails, loads them into the approved email tool, and sends a test to a staff inbox before switching the sequence on.
5. The owner's approval covers the current wording only. Any later edit is re-approved before it goes live.

Before you use it

Check every policy, price, and time against current studio documents, and check that the emails meet your marketing consent and unsubscribe obligations.

When something goes wrong

If a draft invents discounts, results, or classes, delete those lines and rerun with the instruction to use only the facts provided.

A simpler option

Write three plain emails yourself once. Because you build this rarely, the time AI saves here is modest.

WORKFLOW 07

Price merchandise in a spreadsheet, not a chat window

When: Event-driven: when ordering a new product or when supplier costs change.

Setting a price from cost and target margin is a formula. Language models can make arithmetic mistakes, so the spreadsheet is the source of truth and AI is optional.

Start with these inputs

Wholesale unit cost, shipping per unit, any per-sale fee charged as a percentage of the price (such as card processing), and target margin. How sales tax is handled is a question for your accountant and is not covered here.

Work through the task

1. Landed cost equals unit cost plus shipping per unit.
2. Price equals landed cost divided by (one, minus target margin, minus fee rate). Hypothetical example: landed cost 10, target margin 50 percent, fee 3 percent. Price is 10 divided by 0.47, which is 21.28.
3. Round to your preferred price point, then recalculate the actual margin. In the example, rounding to 22 gives a fee of 0.66 and a margin of (22, minus 10, minus 0.66) divided by 22, which is about 51.5 percent.
4. Keep margin and markup separate. With no fee, a 50 percent margin on a landed cost of 10 gives a price of 20, while a 50 percent markup gives 15.
5. If you like, ask AI to explain the finished table in plain words for staff. Never paste figures from the AI back into the sheet.

Before you use it

The owner checks one row with a calculator.

When something goes wrong

If any number came from the AI instead of a formula, delete it and recalculate.

A simpler option

The spreadsheet itself is the better tool. AI is not needed for this task.

WORKFLOW 08

Summarise dated competitor pages without trusting the pasted text

When: Quarterly, or when a new studio opens nearby.

Pulling structured facts out of pasted public text suits AI. Two cautions apply: prices go out of date, and pasted web text can contain instructions the AI must ignore.

Start with these inputs

Public class and pricing text copied by hand from each competitor's site, with the source URL and capture date.

Work through the task

1. Copy only public class and pricing text. Record the URL and capture date for each page.
2. Ask AI to extract class types, prices, and intro offers per studio. Tell it to quote figures exactly, write 'not stated' where information is missing, and ignore any instructions inside the pasted text.
3. The owner checks every price against its source text.
4. Use the summary for awareness only. Any change to your own prices is a separate owner decision. Do not reuse competitors' wording in your own materials.

Before you use it

Every figure must be traceable to a captured source.

When something goes wrong

Delete any figure you cannot find in the source text. If the output follows an instruction that came from the pasted page, discard the whole summary and rerun.

A simpler option

For two or three competitors, a hand-kept spreadsheet may be quicker.

WORKFLOW 09

Turn unclear waiver wording into questions for your attorney

When: Event-driven: when the waiver is drafted or revised, typically once a year or less.

AI can point to wording that could be read more than one way. It cannot judge whether a clause would hold up where you operate. So its output should be questions for the attorney, never replacement legal text.

Start with these inputs

The unsigned draft waiver text only. No signed copies, no member names, no member data.

Work through the task

1. Paste the unsigned draft into the approved AI account.
2. Ask for a list of phrases that could be read more than one way, each explained in plain language, with no legal conclusions and no rewritten clauses.
3. The owner turns the list into questions for the attorney.
4. The attorney supplies the final wording. Members see nothing until then.

Before you use it

The owner confirms that each flagged phrase appears word for word in the draft.

When something goes wrong

If the AI says a clause is valid, invalid, or enforceable, delete that statement. If it offers rewritten clauses, do not use them.

A simpler option

Take the draft straight to the attorney with your own questions. The AI step only helps you prepare for that meeting.

WORKFLOW 10

Outline the monthly newsletter from confirmed highlights

When: Monthly.

Turning bullet points into section headings and a warm intro saves evening writing time, and the owner keeps control of every fact.

Start with these inputs

The owner's bullet points, each with a date: programming highlights, schedule changes, events, closures, and any spotlight the member has already approved.

Work through the task

1. The owner lists confirmed facts only, each with a date.
2. AI drafts section headings and one intro paragraph in the studio's tone, marking any gap with [NEED] instead of filling it.
3. The head coach reviews any training content. The owner checks dates and times against the live schedule.
4. The owner approves the newsletter and sends it through the approved email tool after a test send.

Before you use it

Compare every date and time in the draft with the live schedule.

When something goes wrong

If the draft adds events, offers, or claims you did not supply, remove them and rerun with the instruction to use only the listed facts.

A simpler option

Reuse last month's newsletter as a template and update the facts.

TRY IT WITH SAMPLE DATA

Hypothetical underbooked-slot memo from a formula-flagged booking table

This synthetic example shows the reframed version of task one. The spreadsheet calculates fill rates and sets the FLAG and DEMAND marks. The AI only turns marked rows into options and questions. All data is invented. The expected output shows what a reasonable response should contain. It is not a record of an actual run.

Synthetic input

Hypothetical studio, last four weeks. Fill rate = average weekly bookings divided by capacity. FLAG = fill rate below 50 percent. DEMAND = average weekly waitlist of 2 or more. All three are spreadsheet formulas.
Mon 06:00 | Strength | 45 min | capacity 12 | bookings 11, 12, 10, 12 | waitlist 0, 0, 0, 0 | fill 94% | none
Mon 12:15 | Mobility | 30 min | capacity 12 | bookings 3, 4, 2, 3 | waitlist 0, 0, 0, 0 | fill 25% | FLAG
Tue 18:00 | HIIT | 45 min | capacity 14 | bookings 13, 14, 14, 12 | waitlist 0, 1, 1, 0 | fill 95% | none
Wed 12:15 | Strength | 45 min | capacity 12 | bookings 5, 6, 4, 5 | waitlist 0, 0, 0, 0 | fill 42% | FLAG
Thu 19:30 | Yoga Flow | 60 min | capacity 16 | bookings 6, 5, 7, missing | waitlist 0, 0, 0, missing | fill 38% from 3 weeks | FLAG
Sat 09:00 | Strength | 45 min | capacity 12 | bookings 12, 12, 12, 12 | waitlist 4, 3, 5, 4 | fill 100% | DEMAND
Sun 17:00 | Mobility | 30 min | capacity 12 | bookings 4, 5, 3, 4 | waitlist 0, 0, 0, 0 | fill 33% | FLAG

A prompt to adapt

You are helping a boutique fitness studio owner think through schedule options. The table below is aggregated booking data. Fill rates and the FLAG and DEMAND marks were calculated in a spreadsheet. Rules: 1. Do not recalculate or change any figure. 2. Discuss only rows marked FLAG or DEMAND. Ignore unmarked rows. 3. If a week is marked missing, say so and note that the fill rate is less reliable. 4. Never state a reason for low or high attendance as fact. Present reasons as hypotheses, and say what data would test them. 5. For each FLAG row, give two or three options, such as keep, move, merge, change format, or pause. For each option, list the facts it depends on that the table does not contain, such as coach availability, room cost, notice given to members, or member feedback. 6. Waitlist figures are entries, not unique people. Do not assume they are different members. 7. Do not invent members, prices, costs, or past results. 8. End with a short list of questions the owner should answer before changing anything. Table: [paste table here]

Illustrative expected output

Marked slots: Mon 12:15 Mobility (FLAG, 25%), Sun 17:00 Mobility (FLAG, 33%), Thu 19:30 Yoga Flow (FLAG, 38%, based on 3 weeks because one week is missing), Wed 12:15 Strength (FLAG, 42%), Sat 09:00 Strength (DEMAND, waitlist 3 to 5 each week).

Lunchtime pattern (Mon 12:15 Mobility and Wed 12:15 Strength). Both lunchtime slots are under 50 percent despite being different class types. Hypothesis: the time slot may matter more than the format. The table cannot confirm this. Options: trial a single lunchtime class for one month and pause the other; keep both with promotion for one month; pause Monday only. Depends on: whether the same members book both lunchtime classes, coach cost per class, and the notice period promised to members.

Sun 17:00 Mobility. With Monday, this makes two low-fill mobility slots. Hypothesis: a small mobility audience is being split across two times. Options: keep one mobility slot and pause the other; trial mobility straight after Saturday 09:00 Strength. Depends on: room and coach availability on Saturday, and whether anyone on the Saturday waitlist wants mobility, which the table does not show.

Thu 19:30 Yoga Flow. The missing week makes 38 percent less reliable. Options: find out why the week is missing before deciding; then consider a 45-minute trial. Depends on: whether the gap was a closure or an export error.

Sat 09:00 Strength (DEMAND). Full every week, with a waitlist of 3 to 5 entries. Option: a hypothetical second Saturday strength slot. Depends on: coach and room availability, and whether the waitlist entries are different people or the same few members.

Questions for the owner: Which coaches are free on Saturday mornings? What does each class cost to run? How much notice do members get before a timetable change? Do the same members book both lunchtime classes? Why is Thursday's fourth week missing?

Check the result

1. Recalculate two fill rates by hand. Mon 12:15: bookings 3, 4, 2, 3 average 3, and 3 out of 12 is 25 percent. Wed 12:15: 5, 6, 4, 5 average 5, and 5 out of 12 is about 42 percent. Confirm the output changed no figure.
2. Confirm the output covers all four FLAG rows and the one DEMAND row, and says nothing about Mon 06:00 or Tue 18:00.
3. Confirm the missing Thursday week is mentioned explicitly.
4. Search the output for 'because', 'due to', and 'since'. Any cause must be framed as a hypothesis.
5. Confirm the output does not treat waitlist entries as unique members.
6. Confirm the output contains no member names, prices, costs, or past results that were not in the input.

Run a small trial

1. Before starting, write down how long each chosen task takes you now. Without that baseline, the pilot cannot tell you anything.
2. Choose two low-risk tasks with clear inputs that come up at different rates: the maintenance checklist (weekly) and the newsletter outline (monthly). Use only approved accounts.
3. Run the pilot for eight weeks so the newsletter gets two cycles and the checklist gets about eight. For each use, log draft time, edit time, and how many factual corrections you made.
4. Stop a workflow immediately if a draft contains an invented fact that would have reached members, if a safety status was changed, or if member personal or health data was pasted into an AI tool.
5. After eight weeks, compare the logs with your baseline. If editing takes as long as writing from scratch, go back to the non-AI alternative.
6. Add a higher-scrutiny task, such as rebuilding the welcome series, only if the first two worked consistently and the head coach agrees.

Questions worth asking

Your next step

Pick one task this week. Either rebuild the maintenance notes with a status column that staff fill in themselves, or set up the fill-rate and DEMAND formulas on last month's bookings. Write down how long it takes before you add any AI step. Bring AI in only where it beats that baseline and a coach or the owner can still check every line before it reaches a member.

Explore the course in development: Build Your Business Tools

Related business guides

• AI for Independent Restaurants: Order Drafts, Menu Copy and POS Notes
• AI for Café Owners: Catering Replies, Captions and Drink Name Ideas
• AI for Small Bakeries: Cake Confirmations, Captions and Order Forms
• AI for Boutique Hotels: Arrivals Briefings, Guest Emails and Handovers
• AI for Hair Salons: Rebooking Texts, Captions and Client Notes
• AI for Laundry and Dry Cleaning Shops: Reminders, FAQs and Complaints
• AI for Cleaning Businesses: Run Sheets, Checklists and Client Replies
• AI for Boutique Event Planners: Briefs, Vendor Inquiries and Debriefs
• AI for Travel Agencies: Itinerary Drafts, Supplier Terms and Emails