Plan the month around five content pillars, have AI draft captions from your own notes and photos, then run each post through a four-question claims check before scheduling: is it true, typical, provable, and consented to by the people in it? A post that fails the check is rewritten as method, not outcome.
AI will add results you never gave it. Request a caption about a client's progress and it may add "lost 8 kg in six weeks" or "pain-free in days", figures nobody supplied. So the order matters: your facts and proof go in first, the AI writes around them, and a human checks the output against the proof file before anything is scheduled.
Five pillars that fill a month without inventing results
Twenty posts is five a week for four weeks, which suits a solo trainer who is also coaching. Splitting them evenly across five pillars keeps the feed from turning into all offers or all client stories, and it shows where the claim risk sits: the pillars that talk about people and promises need most of your checking time.
| Pillar | Posts | What it covers | Claim risk and what to check |
|---|---|---|---|
| Teach | 4 | One technique, cue or programming idea per post | Low. Check for medical promises such as "fixes back pain" |
| Train with me | 4 | Session clips, warm-ups, how you write a block | Low. Check that no client is identifiable without consent |
| Q&A and myths | 4 | Answers to questions clients really ask | Medium. Check for absolutes and health claims |
| Client journeys | 4 | One client's progress, in your words or theirs, with permission | High. Needs consent, proof and a typicality statement |
| Offers and community | 4 | Open slots, group sessions, start dates | Medium. Check for guarantees and time-frame promises |
The month on one page: 20 posts, week by week
This is the main worked example, for a solo trainer with about 14 clients posting to Instagram and Facebook. Every line is a prompt for you to adapt, and none contains a result. The client journeys are built from things you can prove, such as a dated milestone or a client's own words, and never from a number on a scale.
- Week 1. Teach: three cues for a better hip hinge (carousel). Train: the 30-second warm-up I use before every session (reel). Q&A: "Do I need to train fasted?" answered in three plain lines. Client journey: what 12 weeks of two sessions a week looked like for one client, the schedule rather than the scale. Offer: two spaces open for early-morning sessions.
- Week 2. Teach: how to choose a starting weight for goblet squats. Train: behind the scenes of how I write a four-week block. Myth: what "toning" usually means, and what it does not. Client journey: a client in her own words on what felt different in her week. Offer: a free 20-minute intro call, with no promises attached.
- Week 3. Teach: three ways to make a push-up easier or harder. Train: a 20-minute no-equipment session at home. Q&A: how often a beginner might train, framed as what I usually start people on. Client journey: a client's first unassisted pull-up, with the date and the video. Community: a Saturday park session, guests welcome.
- Week 4. Teach: why I programme a lighter week. Train: the equipment I use and what I skip. Myth: why a hundred crunches will not target belly fat. Client journey: a verbatim client quote about the sessions, posted with permission. Offer: next month's start dates and how to book.
The fourth-week myth post is worth copying. It corrects a false claim that other accounts make, which is a safer way to say something strong than promising a result of your own.
Caption prompts that stay inside your facts
Gather the raw material first: your notes for each post (what you actually said, taught or saw), three past captions that sound like you, and the dates and records behind any client story. Put the notes in a ChatGPT or Claude project, or a Gemini Gem (becoming a skill from November 2026), so you are not pasting the voice samples every time. Then use one prompt for all five pillars and change only the post type and the facts.
You write Instagram and Facebook captions for a solo personal trainer.
Write in the voice of the SAMPLES. British spelling. No emoji unless the
samples use them.
Use ONLY the facts in FACTS. Do not add any number, timeframe, result,
injury, medical condition or quote that is not in FACTS. Do not promise
or imply an outcome. Never use: guaranteed, melt, burn fat, detox, fix,
cure, toned in, "results in X days".
If a post needs a fact you were not given, write "NEED FROM TRAINER:"
and the question instead of guessing.
POST TYPE: [Teach | Train with me | Q&A | Client journey | Offer]
FACTS: [paste your notes]
SAMPLES OF MY VOICE: [paste three past captions]
Return: a hook line under 12 words; a caption under 120 words; three
hashtags; and a list headed "CLAIMS I MADE" with every factual statement
in the caption, one per line, so I can check each against my proof.The last instruction does the heavy lifting. A list of claims turns a smooth paragraph into separate statements you can tick off one by one, and it exposes the additions the assistant slipped in. Here is a Teach post and what came back (illustrative).
Hook: Three cues for a better hip hinge. Caption: Most people hinge by bending at the back. Try this instead: 1) push your hips back as if closing a car door behind you; 2) keep a soft bend in the knees; 3) keep your spine long, not rigid. This is the cue I use most with beginners, and it makes every deadlift safer. CLAIMS I MADE: three cues listed (from your notes); "most people hinge by bending at the back"; "the cue I use most with beginners"; "makes every deadlift safer".
Three of those four claims were not in the notes. "Most people" is a statistic about strangers, "the cue I use most" is a statement about your own coaching that may or may not be true, and "makes every deadlift safer" is a safety promise. The fix is to cut all three and keep the cues: "Three cues I use for the hip hinge", then the list. A Teach post is strongest as pure method.
Client journeys need more care because the facts are about a person. Suppose your notes say: signed consent on 3 September; trains twice a week since June; the pull-up was her goal from the first session; first unassisted pull-up on 14 October; her words, "I never thought I would manage one". The first draft comes back like this (illustrative).
From zero to pull-up in four months! Her strength has completely transformed, and she has lost inches along the way. CLAIMS I MADE: "from zero"; "four months"; "completely transformed"; "lost inches".
Run the claims against the notes. "From zero" implies she could not hang from a bar at the start, which nobody recorded. "Four months" is a guess at June to October. "Completely transformed" is puffery that no record could support. "Lost inches" is invented outright. The rewrite keeps only what the records carry: "A client's first unassisted pull-up, on 14 October. She has trained with me twice a week since June, and this was the goal we set at the start. In her words: 'I never thought I would manage one.' Posted with her permission." It is shorter and more believable, and every sentence has a record behind it.
The no-false-claims check: four questions before anything is scheduled
Run every post through four questions, in this order. A "no" at any step stops the post until it is fixed or rewritten.
- Is it true? Did it happen as the caption says, with the dates and numbers the caption gives? If the caption says "since June", a booking record should say June.
- Is it typical? If the result is exceptional, say so. One client's first pull-up is a story about one client. A caption that lets readers assume it is what most clients get is a claim about the service.
- Can you prove it? Keep a short proof folder per post: dated photos or videos, session or measurement logs, booking records, saved messages. If you cannot produce it in five minutes, you cannot publish it.
- Is everyone in it consenting? Written, specific permission from each identifiable person, covering what you are posting, where, and how they can withdraw.
The check works best on wording as well as on facts. These are phrases to watch for in AI drafts, with safer versions (illustrative rewrites).
| Draft phrase | Problem | Safer version |
|---|---|---|
| "Lose 10 kg in 30 days, guaranteed" | A guarantee plus a time-frame promise about a body | "What the first four weeks of training look like: three sessions, one simple eating habit, one weekly check-in" |
| "Her back pain is gone thanks to my programme" | A medical outcome you cannot prove or attribute | "She has trained with me for 12 weeks and says her sessions feel easier" (with consent, as her words) |
| "Burn belly fat with these five moves" | Spot reduction is a myth, and "burn fat" promises an outcome | "Five core exercises I use with clients, and how I cue each one" |
| "Transformation: 12 kg down in 8 weeks" | Exceptional, unproven and possibly misleading about what else changed | One client's story, with dated records, other changes she made, her consent, and "this is one client's result, not an average" |
| "The best trainer around" | A comparative claim nobody can prove | "Training clients since [year]; here is how I work" |
| "Detox your January" | A health claim with no meaning you can evidence | "Starting again after a break: a four-week return plan" |
A line such as "results may vary" does not repair any of these. If a headline implies a result, a footnote will not fix it, so change the headline. For more on the evidence habit behind the second and third questions, backing up every claim AI writes in your marketing goes through a full evidence log, and the same ideas apply to testimonials in editing a trainer's testimonials with AI.
A claims log that takes two minutes per post
The proof folder is easier to keep if each client post gets one row in a shared sheet. The columns are the four questions plus a decision. Here are two rows from the month above, one approved and one blocked (illustrative).
| Post | Claim | Evidence | Typical? | Consent | Decision |
|---|---|---|---|---|---|
| Week 3, client journey | First unassisted pull-up on 14 October; trained twice a week since June | Session log entry for 14 October; booking records June to October; saved message with her quote | No: one client's own goal, and the caption says so | Signed 3 September: name, video, quote, both platforms | Approved |
| Week 1, client journey (first draft) | "Lost 6 kg in the 12 weeks" | None: she did not weigh in | Unknown | Signed, but it covers photos only | Blocked: rewritten as the schedule only |
The blocked row shows the check earning its keep. The weight figure came from the assistant's guess at what a 12-week post "should" say, and the consent form turned out to cover photos but not figures. Two separate failures were caught by two minutes of looking.
Client consent: what to get in writing before anyone appears
Treat a client's progress as personal information, because training details can reveal health and lifestyle. Keep the consent separate from the training agreement and ask for it in specific parts, so that agreeing to one use does not stretch to another.
- What you will post. Photos, video, first name, a quote, a milestone, any numbers. Each is a separate tick.
- Where and for how long. Which accounts, whether the post can be boosted, and how long it stays up.
- How they can withdraw. A named contact, and a plain statement that you will remove it from your own accounts but cannot recall copies others have saved.
- Anything given in return. If a client receives a discount or free sessions for appearing, note it in the record and ask your adviser how it must be disclosed.
- Under-18s. A parent or guardian gives consent, and the post should show nothing you would not show their parent.
An AI draft of the core wording, for your adviser to check (illustrative):
I, [client name], agree that [business name] may post the following about my training on its Instagram and Facebook accounts: [ ] my first name [ ] photos [ ] video [ ] a quote I have approved [ ] milestones such as a lift or a date. I have seen the post before it goes live. I understand I can ask for it to be removed from [business name]'s accounts at any time by contacting [contact route]. I have not been promised anything in return unless stated here: [ ].
Never let the assistant write a client's words. It will happily draft a glowing testimonial "in the client's voice" from your notes, and that is a statement the client never made. Use verbatim messages, approved by the client, and edit only for typos. Another trap is image editing: do not use AI tools to change a client's body or to generate a "before" image, and do not post generated pictures of fit people as though they were clients. Basic exposure and cropping are fine; anything that changes how a body looks is a results claim that needs the client's agreement. For parallel consent wording in another trade, a dental practice's social media consent form shows photo permission laid out step by step.
Q&A and offer posts: the two medium-risk pillars
Question-and-answer posts attract health questions, and a trainer's lane is training, not diet therapy or medical advice. Offer posts attract the opposite problem: urgency and promises. One example of each shows the pattern (illustrative).
The question is "Do I need to train fasted?" and your notes say: there is no single rule; some clients prefer it, some feel flat; you ask what they have tried; anyone with a medical condition or on medication should check with their doctor first. The first draft says: "Training fasted burns more fat and speeds up results, so it is worth trying." Neither claim was in your notes, and both are contested. The version to publish follows a three-part shape that works for most Q&A posts: what I do with clients ("I ask what they have tried and how it felt"), the "it depends" factor ("some people train better after breakfast, others feel flat"), and when to ask a professional ("if you have a medical condition or take medication, check with your doctor first").
For an offer, your notes say: two spaces on Tuesday at 7 am from 3 November; the first session is a 30-minute chat about goals plus a trial of the warm-up. The draft says: "Only 2 spaces left! Transform your body this winter. Guaranteed results or your money back!" You never offered a guarantee, "transform your body" promises an outcome, and "only 2 spaces" is true only if the diary says so, which is a check against your own records. The rewrite is "Two Tuesday 7 am spaces from 3 November. The first session is a 30-minute chat about your goals and a trial of the warm-up. Book through the link." Scarcity claims are fine when they are accurate. Take them out of the post the moment the spaces go.
When you boost a post, the advertising rules apply as well
A boosted post is an ad, and Meta's own standards then apply. Its Health and Wellness standard (updated July 2026) bars clickbait in a health or weight-loss context, including "promises of specific outcomes within a set timeframe without disclaimers or qualifiers". It requires ads for dietary, health and weight-loss or weight-gain products and services to be targeted to people aged 18 or over, and it exempts offers marketed for general well-being, such as fitness services and health clubs, from that age rule. The same page allows before-and-after images for general cosmetic procedures aimed at adults, but it says nothing specific about before-and-after images in weight-loss ads. It does allow ads that show a dietary product's effect when the time taken to see results is stated clearly, and it bans close-ups that pinch fat on a body area. Read the current wording before you boost any transformation post.
A practical rule falls out of this: never boost a client journey on the strength of its numbers. Boost the teach, train and offer pillars, where the claims are method and availability, and let client stories stay organic until the page's wording and your records are both clear.
Scheduling the month and what it costs
With the claims log approved, load the month into a scheduler so that posting stops competing with sessions. Buffer's free plan, on its pricing page of 28 October 2026, connects up to three channels and holds 10 scheduled posts per channel, refillable at any time. Twenty posts a channel is more than that queue holds, so on the free plan you would top it up every week or so. Loading the whole month in one sitting points to the paid Essentials plan, which the page shows at $5 a month per channel billed yearly ($60 a channel a year), so Instagram and Facebook together come to $120 a year. Check the month-to-month price on the page before you commit. Other schedulers price differently, and a comparison of Buffer and Later for Instagram sets the plans side by side.
Three small settings matter more than the choice of tool. Ask the AI for three hashtags, not a list, because Instagram announced a five-hashtag cap in December 2025. Keep a human approval step between the AI draft and the schedule, so nothing reaches the queue without a tick in the claims log. And keep portrait photos at the 3:4 ratio Instagram now uses (1,080 by 1,440 pixels) so a carousel does not crop your cues off the edge.
Reviewing the month: rewrites, proof and enquiries
After four weeks, look at four things. First, how many posts the claims check changed, because a count of zero means either the prompt is excellent or the check was skipped, and you should expect a handful of rewrites in the first month. Second, spot-check five posts from the log: open the evidence for each and confirm it is where the log says. Third, read the replies and direct messages, because the best signal is enquiries that mention a specific post, not likes. Fourth, note which pillar produced enquiries and which produced silence, and shift two posts next month from the quiet pillar to the busy one.
If you want captions that sound more like you before the claims step, writing Instagram captions with AI that sound like you covers the voice samples in detail. For the client-facing side of the same business, session recaps and renewal texts uses the same rule of keeping the facts in your hands and the wording in the assistant's.
Questions trainers ask about posting results
Is a results may vary line enough to cover a transformation post?
No. A footnote does not repair a headline that overstates what is typical or promises an outcome. Make the post itself accurate: say it is one client, give the timeframe, the training and any other changes they made, keep the proof, and have written consent. If you are unsure, ask your adviser or insurer before posting.
Can I use AI-generated images of fit people in my posts?
For decoration such as a quote card or background, yes, with care. Never use generated or edited images to imply a real client, a real transformation or your own physique. If you retouch a photo of a client beyond basic brightness, tell them first and keep the original, because an altered body is a results claim.
How many hashtags should an AI caption include?
Instagram announced a cap of five hashtags per post in December 2025, so ask the AI for three and choose them yourself. A few specific tags about the topic, such as the exercise name, are enough. Long lists of generic fitness tags look automated and add little to what your caption already says.
Do I need consent to post a client's first-name story?
Yes. Get specific written permission before you share a client's progress, even without a photo, because training details can reveal health and lifestyle information. Say what you will post, where, for how long and how they can withdraw. Posts about under-18s need a parent or guardian's consent.
Further reads
- AI Marketing Best Practices for a Local Business: A Pre-Post Checklist — A pre-post checklist you can run on any business account.
- Can You Use AI to Write Testimonials? Where the Legal Line Is — Where the line sits between editing a testimonial and inventing one.
- Health Food Store Marketing Ideas: AI Posts Without Health Claims — Writing posts in a health-adjacent trade without health claims.
- How to Plan a Month of Marketing Content With AI in One Hour — A general one-hour method for planning a month of marketing.
- AI for Fitness Trainers: Client Check-Ins Without Medical Advice — Client check-ins that stop short of medical advice.
- AI Video for Salons: Turn Before-and-After Photos Into Short Reels — Before-and-after content in another trade, and the consent it needs.
- How to Plan a Month of Barbershop Instagram Content With AI — A month of Instagram content planned for a different local trade.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Meta's Health and Wellness advertising standard (read 28 October 2026, updated July 2026); Buffer's pricing page (read 28 October 2026); news reports of Instagram's December 2025 announcement capping hashtags at five per post.