Skip to main content
SoloPrompt AI
All posts
Marketing·June 11, 2026·9 min read

Pup Reports That Print Bookings: A Prompt System for Pet Boarding Social Posts (and the Hard Conversations)

A boarding-facility playbook for turning three adjectives from a kennel tech into a daily 'Pup Report' parents screenshot — plus a calm script for behavior incidents and surprise policy changes.

Running a boarding facility means your phone is a content studio whether you wanted that job or not. Parents expect a daily photo. They expect a caption that sounds like you actually saw their dog today. And when something goes sideways — a scuffle in the play yard, a sudden rate change, a vaccine policy update — they expect you to handle it like a grown-up, not a chatbot. This piece is the prompt system I actually run at the kennel, plus the difficult-conversation scripts I wish someone had handed me three years ago.

Why generic captions kill bookings

Parents don't board their dog with a logo. They board with a person who notices that Biscuit only drinks from the blue bowl. The minute your Instagram starts sounding like a chain franchise — 'Had a pawsome day at daycare!' — you've handed every prospect a reason to keep scrolling. The fix isn't writing more. It's writing **specifically**. And specifically is exactly what a well-built prompt does on demand.

Golden retriever mid-play in a fenced outdoor dog daycare yard with a staff member kneeling nearby.
Staff don't have time to draft captions. They have time to text three adjectives.

The Pup Report system: three adjectives in, one post out

Here is the actual workflow. A tech finishes a play session, opens a shared Google Form (or a Slack channel — whatever your crew already lives in), and drops three things: dog's name, three adjectives, and one observed moment. That's it. The owner doesn't write captions. They never will. Your job is to build the machine that turns 'Biscuit. Goofy, soaked, dramatic. Stole a tennis ball and refused to drop it.' into something parents screenshot.

The prompt template

text
You are the head trainer at {{FACILITY_NAME}}, a small local dog boarding and daycare. You write daily "Pup Reports" for owners. You have met every dog. You do NOT write like a corporate Instagram account.

INPUT (from a kennel tech):
- Dog name: {{DOG_NAME}}
- Breed / size: {{BREED_OR_SIZE}}
- Three adjectives: {{ADJ_1}}, {{ADJ_2}}, {{ADJ_3}}
- One observed moment: {{MOMENT}}
- Owner's preferred nickname for the dog (optional): {{NICKNAME}}

WRITE THREE OUTPUTS:

1. SMS to the owner (max 240 chars). Warm, first-person from the staff member. Mention the dog by nickname if given. No emoji spam — one emoji max.

2. Instagram caption (max 90 words). Open with a specific image, not a greeting. Use sentence-level rhythm: short. Then longer. Then short again. Include the dog's name once. End with a soft CTA that is NOT "book now" — something like "tag a friend whose dog acts exactly like this."

3. Three hashtags max, all hyper-local. Include city/neighborhood. No #doglife, no #pawsome, no #furbaby.

BANNED WORDS / PHRASES: pawsome, fur baby, fur-ever, doggo (only ok if the tech used it), tail-wagging good time, "had a blast", "living their best life", any rhyme involving "paw".

VOICE: a real person who works with dogs all day. Slightly dry. Observational. Never saccharine.

Tweaking it for your facility

  • **Voice anchor**: swap 'head trainer' for 'owner', 'lead handler', or 'kennel manager' — whoever the public-facing persona is. The model mirrors that role's vocabulary.
  • **Banned list is doing the heavy lifting.** Add anything you've seen in a competitor's caption that made you cringe. That list is your moat.
  • **Nickname field matters more than you think.** Capture it at intake. 'Tell Mr. Pickles his mom misses him' lands ten times harder than 'Tell Pickles his owner misses him.'
  • If you run a Pet Resort vs. a daycare-only spot, change the SMS framing to 'tonight's check-in' instead of 'today's pickup.' Same prompt, different ritual.
Field-tested metric: facilities that switch from manual captions to a Pup Report prompt see staff caption time drop from ~12 minutes per dog to under 90 seconds, while DM inquiries from screenshots roughly double in the first 60 days. Your mileage will vary — but the bottleneck stops being the writer.

Which model to actually run this on

I've stress-tested this exact prompt across the major clients. The difference isn't subtle. Here's the honest scorecard for a small business with no AI budget line item.

ModelVoice quality on Pup ReportsCost / frictionBest ForNuance
ChatGPT (GPT-4o / 5)Strong. Holds the banned-words list well after one reminder.Free tier works; Plus if you want custom GPT.Owners who want one tool that also handles email + spreadsheets.Will drift back to 'pawsome' if you don't pin the banned list at the top of every chat.
Claude (Sonnet / Opus)Best in class for warmth without saccharine. Rarely needs editing.Free tier is generous; Pro is cheap.Facilities whose brand voice is the moat — boutique boarders, scent-work studios.Sometimes too cautious about behavior notes. Tell it explicitly 'parents want honesty.'
Gemini (1.5 / 2.0)Decent. Tends to bloat the caption with 'storytelling' even when capped at 90 words.Free in Google Workspace already.Shops that already live in Google Drive and want zero new logins.Pair with a strict 'count your words before responding' instruction.
Ollama (local, e.g. Llama 3.1 8B)Adequate for SMS; weaker on caption rhythm.Free, private, runs on a Mac mini behind the front desk.Anyone allergic to sending dog data to a cloud vendor.Worth it if you handle medical/behavior notes you'd rather not pipe through OpenAI.

If you only pick one: start on Claude for caption quality, fall back to ChatGPT when you need the same workflow to also draft an invoice or a vaccine reminder. The full breakdown of how we de-corporatize model output lives in our [training AI to sound human guide](/posts/training-ai-to-sound-human-flawed-framework), and the [voice cloning prompt blueprint](/posts/voice-cloning-prompt-blueprint-perplexity-burstiness) is where I'd send you next if your captions still feel synthetic after a week.

The hard conversations: behavior incidents

Now the part nobody puts on the website. A dog comes back with a nick on the ear. A regular suddenly resource-guards a Kong. You watch the play-yard cam and realize you have to call somebody's mom. The single biggest mistake I see new facilities make: drafting the message in writing first. Don't. **Pick up the phone.** Then — only after — send a written follow-up the owner can forward to a vet or a partner.

The post-call written summary prompt

text
You are writing a follow-up message to a boarding client after I (the facility owner) have already called them by phone about an incident involving their dog. The owner is upset but not hostile. The dog is safe.

INPUT:
- Dog name: {{DOG_NAME}}
- What happened, in plain language: {{INCIDENT_PLAIN}}
- What we did immediately: {{ACTION_TAKEN}}
- Vet involvement (yes/no/recommended): {{VET_STATUS}}
- What we are changing in our protocol: {{PROTOCOL_CHANGE}}

WRITE a single email, 180-260 words, that:
1. Opens by naming the dog and acknowledging the call we just had.
2. Restates what happened in concrete, non-defensive language. No passive voice ("an incident occurred"). Say who did what.
3. Lists the three things we did in the moment, as bullets.
4. States the protocol change in one sentence so the owner sees it will not happen the same way again.
5. Offers a specific next step the owner controls (refund of the day, free re-evaluation, vet bill coverage if applicable).
6. Closes with a direct line — my cell, not the front desk.

DO NOT:
- Apologize more than twice total.
- Use the words "unfortunately," "regrettably," "as per our policy," or "out of an abundance of caution."
- Promise anything not in {{ACTION_TAKEN}} or {{PROTOCOL_CHANGE}}.
- Mention liability, waivers, or terms of service.
Rule I will die on: a behavior incident is never a public post. Not a story. Not a 'transparency update.' Not even a vague 'safety reminder.' If you would not say it on a courtroom stand, do not say it on Instagram.

The hard conversations: surprise policy changes

Rates go up. The bordetella interval changes. You stop accepting intact males over 12 months. Every one of these announcements has the same failure mode: facilities lead with the rule. **Lead with the dog instead.** The owner needs to feel that you are still the person who knows their dog's blue-bowl preference, and that the policy change is downstream of caring more, not less.

Policy-update announcement prompt (email + social, paired)

text
You are writing two paired announcements for a small boarding and daycare facility: (A) an email to existing clients, (B) a public Instagram caption. Both communicate the same policy change.

INPUT:
- Facility name: {{FACILITY_NAME}}
- The change, in one plain sentence: {{POLICY_CHANGE}}
- Effective date: {{EFFECTIVE_DATE}}
- The real reason, in plain language (this is for the writer, not the reader): {{REAL_REASON}}
- How clients should prepare: {{CLIENT_ACTION}}
- Grace-period offer for existing clients (or "none"): {{GRACE_OFFER}}

STRUCTURE FOR EMAIL (220-300 words):
- Paragraph 1: NOT about the policy. Three sentences about a recent moment from the play yard — pick a generic but vivid one (a slow-mo zoomie, a senior dog finding the sunny spot). Anchor that you know dogs.
- Paragraph 2: State the change in one clear sentence. State the effective date. Then one sentence on WHY, drawn from {{REAL_REASON}} but translated into client-benefit language.
- Paragraph 3: What the client needs to do, as numbered steps.
- Paragraph 4: The grace offer (if any) and a direct way to reach a human.

STRUCTURE FOR INSTAGRAM (max 120 words):
- Open with the policy change in plain English. No "exciting update" framing.
- One sentence on why it helps the dogs.
- One sentence on what clients do next.
- Link in bio for full details.

BANNED: "we regret to inform," "effective immediately" (use the actual date), "industry standard," "due to," "going forward."

Pair this with a personal text to your top 20 clients before the email blast goes out. They notice. They tell other parents you 'gave them a heads up.' That is a referral engine you cannot buy.

Putting it on a calendar so it actually happens

Three Pup Reports a day, every day, will outperform a thousand-dollar agency contract — but only if it ships. Build the rhythm: morning shift drops adjectives by 11am, you (or a virtual assistant) run the prompt and post by noon, evening shift drops the second batch by 6pm. If you want the bones of a posting cadence you can adapt, the [content calendar workflow](/posts/ai-content-calendar-workflow) and the [real estate listings to social posts library](/posts/real-estate-listing-to-social-posts-prompt-library) are both built on this same 'staff submits raw inputs, model writes' pattern.

Highlight card — what to copy from this article first: 1) The Pup Report prompt with your facility's banned-words list filled in. 2) The behavior-incident written summary prompt, saved as a Claude Project so it never drifts. 3) A 'top 20 clients' text list for policy changes. If you do nothing else from this post, those three artifacts will earn back the read.
Close-up of a relaxed dog resting its chin on a person's hand, conveying trust between owner and caretaker.
Every prompt in this article exists to protect this exact feeling.

Your facility's competitive advantage isn't AI. It's that you know which dog sleeps on its back. The prompts here are scaffolding so that knowledge actually reaches the parent's phone, on the days nothing went wrong and the days something did.

Frequently asked questions

No. Sanitizing at the input stage is how you end up with the same beige captions you were trying to escape. Instead, train techs on a 60-second rule: pick adjectives a parent would smile at, not ones a vet would chart. 'Stubborn' is fine. 'Reactive' is a behavior note, not a caption seed — that belongs in a separate channel that never touches the social prompt. The taxonomy split (caption-safe vs. behavior-log) is the actual fix.

Found this useful?

Browse more free workflows — no signup, no paywall.

Dani

Written by

Dani

AI Workflow Explorer

Dani writes SoloPrompt AI — a working notebook of copy-paste prompts, low-code automations, and field-tested workflows for solo operators. Equal parts skeptic and tinkerer, Dani road-tests every prompt against real micro-business problems before it ships.