AI Marketing Campaign Examples: Five Worked Ones for Local Businesses

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for AI Marketing Campaign Examples: Five Worked Ones for Local Businesses.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for AI Marketing Campaign Examples: Five Worked Ones for Local Businesses.

AI-assisted local campaigns are small and specific: one offer, one audience, a few weeks. The AI drafts copy, variants and messages; you set the offer, approve every word and count results. A cafe courts nearby offices, a salon launches one service, a gym runs an attendance challenge, a car wash texts on dry days, a dog daycare writes to vets.

Many published AI campaign examples are big-brand case studies with agency budgets, so their results can't be copied by one site and a few hours a week. The figures below are not results. They are illustrative budgets, break-even counts and checks, with every assumption flagged so you can swap in your own numbers.

Follow me on Instagram@sagnikteaches

What all five share, and the prompt skeleton behind them

Each campaign has the same six parts: one offer you can afford to honour, one audience you can actually reach, a handful of assets rather than twenty, a schedule with dates, a cash cost, and a count that shows whether it paid back. The AI does the drafting. Everything else is yours. Here is the skeleton every prompt below is built on:

Connect on LinkedInSagnik Bhattacharya
You are helping the owner of a [business type] write a small local campaign.
Business facts: [what you sell, prices, opening hours, capacity limits]
Offer: [the exact offer, dates, limits, how to claim it]
Audience: [who they are, how you reach them, how many]
Tone: [three words]. Plain British English, no exclamation marks, no emojis.
Rules: use only the facts above. Do not invent awards, reviews, ingredients,
guarantees, qualifications or numbers. No health or results claims.
Output: [the assets, each with a word or character limit]
Finish with a list of every claim in your draft that I need to verify.

The last line earns its place. It makes the AI surface the places where it filled a gap with something plausible, which is exactly where a local campaign gets into trouble. The line before it, "use only the facts above", does the other half of the job.

Subscribe on YouTube@sagnikteaches
JobToolList cost (checked October 2026)
Drafting and variantsA chat assistant; free plans cope, and paid plans are about $20 a month at list price$0 to about $20 a month
Flyers, signs, QR cardsCanvaFree plan includes up to 20 AI uses a month; Pro is $144 a year
Scheduling postsBuffer's free plan (three channels, a 10-post queue on each), or the scheduling built into Google Business Profile$0
Email sequencesMailerLite's free plan (250 subscribers, 2,500 emails a month, three automations of up to five steps); Mailchimp's free plan has no automation flows$0
TextsThe text feature in your booking, loyalty or till systemCheck your per-text price
TrackingA code word, a QR link, a "how did you hear about us?" field, a tally sheet$0

Tracking is the part owners skip. Decide before launch how you will recognise a customer the campaign brought in. Paid social makes this harder: since March 2026, Meta's click-through attribution has counted link clicks only, so someone who liked or shared an ad and then bought no longer shows up as a click-through sale. For a local offer, a code word at the till is the more dependable count.

Cafe: a tasting-box campaign aimed at nearby offices

A 28-seat cafe is quiet after 9:30 on weekdays, and the kitchen can prepare 15 boxes a day without another hire. The goal is three repeat office accounts in six weeks, selling breakfast and lunch boxes at $9.50 a head for teams of 6 to 20. The offer is a free tasting box for eight people at one office meeting, then 10% off the first paid order with the code OFFICE10. The owner walks the surrounding streets and writes down 40 offices, with an office manager's details taken from each company's own public page. (Quiet-hour offers for people who are already in the habit of visiting are a separate idea; the tutorial on filling cafe quiet hours with targeted offers covers that.)

You are helping the owner of a 28-seat cafe write a small local campaign.
Business facts: breakfast and lunch boxes for 6 to 20 people, $9.50 a head, order by
3 pm for next-day delivery, delivery on foot or by bike to offices within a 10-minute
walk, vegetarian and dairy-free options, an allergen sheet sent with every order.
Offer: a free tasting box (8 portions) for one office meeting this month, then 10% off
the first paid order with code OFFICE10.
Audience: office managers at 40 nearby offices; first contact by email, then a visit.
Tone: friendly, practical, brief. Plain British English, no exclamation marks, no emojis.
Rules: use only the facts above. Do not invent awards, reviews, ingredients, guarantees
or delivery times.
Output: three emails (under 110 words each, for days 1, 5 and 12), a one-page flyer in
plain text, two Google Business Profile posts under 120 words, and a counter sign under 25 words.
Finish with a list of every claim in your draft that I need to verify.

A sample of the first email that comes back (illustrative):

Subject: A free tasting box for your next team meeting
Hello [first name],
We run a small cafe a short walk from your office, and we are starting breakfast and lunch boxes for teams of 6 to 20. If you have a meeting coming up, we will bring a free tasting box for eight people so you can see whether it suits. Boxes are $9.50 a head, ordered by 3 pm for next-day delivery, with vegetarian and dairy-free choices and an allergen sheet with every order.
Reply with a date and I will confirm. If you would rather not hear from us again, just say so.
[your name]

That is close to publishable, but a first draft often isn't. A typical one says "freshly baked by our award-winning bakers" and "ready in 15 minutes", and lists "gluten-free options". None of the three was in the facts. The closing list of claims to verify should flag them, and all three come out. This is the usual pattern: the AI reaches for the most common cafe claims when the brief leaves a gap.

The assets are the three emails (days 1, 5 and 12), a one-page flyer, two Google Business Profile posts scheduled for weeks 2 and 4, a counter sign with a QR code to the order form, and the form itself, with a "How did you hear about us?" field and a box for the code.

WeekWhat happens
1Build the list of 40 offices; write and check the assets
2Email 1 goes out; flyers delivered by hand; first Google Business Profile post
3Email 2; tasting boxes delivered to offices that said yes
4Email 3; second profile post; a call or visit to each office that tasted
5 to 6First paid orders; ask "same time next week?"; review at the end of week 6

The cash cost, with assumptions marked: if 10 offices accept a tasting box, each costs $34.60 (eight portions at 35% food cost on a $9.50 head is $26.60, plus $8 for packaging and delivery), so $346. Add $12 for 60 flyers at an assumed 20 cents each, and Canva's free plan covers the design. Cash is $358, plus about eight hours of the owner's time. A typical repeat order of 10 people brings in $95, less $33.25 of food and $8 of packaging and delivery, so it contributes $53.75. A first order carries the 10% code, so it brings in $85.50 and contributes $44.25. Three first orders and five repeats come to $401.50, so about eight paid orders cover the campaign.

To check it, keep a tally with five columns: emails sent, replies, tasting boxes delivered, first orders and repeat orders. Decide at week 6. If at least two offices have ordered a second time, keep going; if not, look at the tasting box and the follow-up call before blaming the emails. Business-to-business email rules differ between places, so keep a plain opt-out line in every message, stop the moment someone asks, and ask a data-protection adviser whether named and generic addresses are treated differently where you trade.

Salon: launching one new service to existing clients

A four-chair salon has 620 clients in its booking system, and 410 of them agreed to texts and email. It is adding grey blending, a 90-minute colour service at $85 (illustrative), and wants 12 bookings in the first four weeks, filling Tuesday and Wednesday gaps. Two house rules: the launch goes to existing clients first, and everyone has a free 15-minute consultation before a first colour appointment. Anyone new to the colour needs a patch test first, following the manufacturer's instructions for the range you use; one range's guidance is a test 48 hours before each use, so check yours.

This campaign is a message sequence more than a set of posts, so here it is as a schedule:

DayChannelMessage and purpose
1Instagram postTeaser for the first silver hairs: a new blending service is coming; join the waitlist
3Text and email to opted-in clientsInvitation to the waitlist, first pick of Tuesday and Wednesday slots
7Instagram storyThree questions answered: how long it takes, whether it covers grey, why a consultation comes first
10Text to the waitlistBooking link for the first slots
17Email to the waitlist and non-bookersReminder of the slots left
After each bookingTextConsultation and patch-test reminder, with the date
Write the Day 3 text message and email for a salon launching a grey blending service.
Facts: 90 minutes, $85, Tuesday and Wednesday slots, consultation before a first colour
appointment, patch test where the colour range requires one.
Text under 160 characters including [first name] and the link, plain characters only.
Email under 120 words. No promises about coverage, "no more roots", or results.
Finish with a list of claims I need to check with my stylists.

A text that fits (illustrative, 126 characters): "Hi [first name], grey blending starts Tue and Wed from [date]. Waitlist gets first pick: [link] Reply STOP to opt out. [Salon]". Count the length again with a real name, date and short link in place of the placeholders. Keep texts plain: messages over 160 characters are split and billed in parts, and a single emoji or curly quotation mark can cut the limit to 70 characters a part.

A typical first draft claims "natural-looking results, guaranteed" and "no more roots, ever". A stylist cannot promise either. Replace claims like these with your stylists' own description of what the service does and doesn't do, and let them read the final copy. The same rule covers consent: message only clients who opted in, and if the booking system doesn't record that, don't send. An opt-in is not permanent either, so honour "stop" at once.

Cash and effort: 410 clients at two texts each is 820 texts, about $32.80 at an assumed 4 cents each, plus $120 for three stylists at two hours each to train and rehearse the consultation at $20 an hour. That is $152.80. A booking in an otherwise empty slot contributes about $41: the $85 price less $14 of colour and $30 of stylist pay for 90 minutes. Four bookings cover it.

To check it, count waitlist sign-ups, bookings from the waitlist, the split between new and existing clients, no-shows, rebookings within your usual colour interval, and any complaints about the messages. At week 4, if fewer than four bookings have come in, the offer or the audience is the problem, not the wording. The tutorial on a four-week AI marketing plan for salons has a broader diary-filling approach for the weeks between launches.

Gym: a six-week attendance challenge with no body promises

A 300-member independent gym has room on weekday evenings and a group of members who visit about three times a month. The challenge is 18 sessions in six weeks, three a week, with one session a day counting at most. Finishers get two guest passes and a draw entry for one free month ($45), and every 12 sessions earns a bring-a-friend pass. Friends claim passes through a short lead form.

The risk in gym campaigns sits in the claims, and loosely prompted AI drifts straight into them. Here is what a first draft tends to produce (illustrative) and what I would publish instead:

AI first draftWhy it can't runWhat I would publish
"Torch fat and transform your body in 42 days"A promised result in a set time; nothing in the offer delivers it"Turn up three times a week for six weeks"
"Burn 600 calories every class"Unverifiable and different for every personDelete it; describe the class instead: "45-minute strength circuit"
"Join hundreds who lost weight"An invented claim, and a weight-loss one"Join [real number] members already signed up"
"Before and after photos from last year's winners"Body-change imagery can breach ad rules and harms some viewersAttendance milestones, and member quotes with written consent
"Guaranteed results"You can't guarantee them"Your goals are yours; we make it easy to turn up"

A prompt that keeps the draft on turning up rather than on bodies:

Write the launch email and a one-page rules sheet for a six-week attendance challenge
at an independent gym of about 300 members.
Facts: 18 sessions in six weeks, at most one counted per day; finishers get two guest
passes and one entry to a draw for a free month; every 12 sessions earns a
bring-a-friend pass, claimed through [form link].
Tone: encouraging, plain, brief. Plain British English, no exclamation marks, no emojis.
Rules: use only the facts above. No weight, fat, calorie, body-shape or health claims,
no promised results, no time-bound outcomes, no before-and-after language. Praise
turning up, never how anyone looks.
Output: an email under 150 words, a rules sheet under 200 words, and subject lines
for the six weekly update emails.
Finish with a list of every claim in your draft that I need to verify.

Even with those rules, read the weekly emails before they go. A shout-out such as "[first name] is flying, down a dress size already" can slip in when a coach adds a personal note, and it breaks the same rules as the weight-loss and before-and-after rows of the table.

Meta's advertising standard for health and wellness, updated in July 2026, says ads shouldn't promise specific outcomes within a set time frame without qualifiers, so the first row is also an ad-policy problem. Check the standard's current text before using any body-change imagery. If your gym uses the PAR-Q+ readiness questionnaire at sign-up, point challenge entrants to it. Its publishers say the form must remain unaltered, so don't ask an AI to reword it. The tutorial on gym referral programmes and reward maths goes deeper on the friend-pass side.

The assets are a one-page rules sheet, a poster, six weekly emails (progress and shout-outs, attendance only), the lead form for friend passes asking for name, email, mobile and preferred session time, and a tracker sheet fed from the check-in system. The schedule is simple: set-up in week 0, launch email and an in-gym sign in week 1, weekly updates in weeks 2 to 5, finish in week 6, and a follow-up offer to finishers.

Cash: a $45 prize, $25 of printing and an assumed $5 a day for 14 days of paid ads, $70, so $140. At $45 a month in dues, that is 3.1 member-months, so four cover it: two friends who join and stay two months would do it. Give friend passes a code so you can credit joins to them, since the ad dashboard won't.

To check it, compare sessions per participating member in the six weeks of the challenge with the same members' six weeks before. Then look at weeks 7 to 10 for the cliff, when attendance can fall back below the old level, and at cancellations during the challenge. Count friend passes issued, used and converted. A decision date two weeks after the finish gives the passes time to convert.

Car wash: texts timed to the first dry day after rain

A hand-wash and automatic site is busy at weekends and quiet on Tuesdays and Wednesdays. The owner has 900 customers who opted in to texts through the loyalty app. The rule: when the evening forecast shows two or more dry days coming after three or more wet ones, text a segment that evening with $3 off any wash before 2 pm on Tuesday or Wednesday. A two-minute forecast check each evening is all the trigger needs, so no automation tool is involved.

TriggerMessage themeValid
First dry day after three wet days"Dry day tomorrow after a wet week"Tue or Wed, before 2 pm
Pollen or blossom season, dry forecast"Pollen on the bonnet again?"Tue or Wed, before 2 pm
Roads salted after icy daysFree underbody rinse with any washTue or Wed, before 2 pm
Dry weekend forecast"Beat the queue"Tue or Wed, before 2 pm
Write four text messages for a car wash, one per trigger in the table above.
Each under 150 characters including "Reply STOP to opt out. [Site name]".
Offer: $3 off any wash before 2 pm on Tuesday or Wednesday (the underbody one: free rinse
with any wash). Plain characters only, no emojis. Do not mention weather you cannot see in
the forecast, and make no claims about paint protection or results.

Two of the four variants you can expect back (illustrative, 128 and 129 characters) read: "Dry day tomorrow after a wet week. $3 off any wash before 2 pm on Tue or Wed. Show this text. Reply STOP to opt out. [Site name]" and "Pollen on the bonnet again? $3 off any wash before 2 pm, Tue or Wed this week. Show this text. Reply STOP to opt out. [Site name]". If a third says "your car will sparkle all week", rewrite it, because a wash can't promise that in changeable weather. Each message should describe the offer and nothing else.

Redemptions are easy to count: staff tally "show this text" at the till. Whether the campaign worked is harder, because some redeemers would have come anyway and rain moves volumes more than any text does. Compare dry days with dry days. Say the last four dry Tuesdays before 2 pm saw 38, 41, 36 and 43 washes, an average of 39.5. On the offer day the count is 56, so 16.5 extra washes. Staff tallied 24 redemptions, so 7.5 of the people who got $3 off would have come regardless.

With an average $14 wash, a $3 discount and about $4 of variable cost (water, chemicals and finishing time, an assumption), each extra wash contributes $7. The net for the send is 16.5 times $7, which is $115.50, minus 7.5 times $3 of discount given away, $22.50, minus the send cost of $12 (300 texts at an assumed 4 cents), so $81. Break-even with 24 redemptions needs nine extra washes. Four baseline days wobble, so look for the same pattern across several sends before you change the offer, and never text the same person more than once a week. For membership offers rather than one-day discounts, see the tutorial on car wash membership marketing and churn messages.

Dog daycare: letters to vets and groomers, plus an assessment offer

A daycare has room for ten more dogs on Tuesdays to Thursdays and wants 12 new regulars in eight weeks. Vets and groomers get asked every week where an owner could send a dog for the day, so the campaign writes to 12 of them. The offer is a free two-hour assessment (normally $25) with a $15 credit towards a first full day. Partners are paid nothing: some professionals may not be allowed to accept referral fees, and the letter works without one. They get an information sheet for the front desk and a promise to point owners back to them.

The letter comes from the same skeleton, with the referral-fee question settled inside the rules:

You are helping the owner of a dog daycare write to nearby vets and groomers.
Business facts: daycare Tuesday to Thursday, [opening hours], room for ten more dogs;
a free two-hour assessment (normally $25) with a $15 credit towards a first full day;
an information sheet explaining what we check before a dog joins.
Audience: practice managers and owners at 12 vets and groomers, who get the letter with
the information sheet, then a visit.
Tone: courteous, brief, practical. Plain British English, no exclamation marks.
Rules: use only the facts above. Offer no referral fee, commission or gift. Do not invent
qualifications, insurance cover, awards or reviews.
Output: a letter under 150 words and a front-desk card under 40 words.
Finish with a list of every claim in your draft that I need to verify.

A sample partner letter from the AI (illustrative), after editing:

Dear [first name],
We run a dog daycare a short drive from your practice. When owners ask where their dog could spend a day while they work, we would like you to have something to hand. We offer a free two-hour assessment, with a $15 credit towards a first full day, and an information sheet explaining what we check before a dog joins. We would be glad to send copies for your front desk, and to point any owner who asks about [veterinary or grooming service] back to you.
[your name]

A typical first draft adds lines such as "all our staff are certified canine behaviourists" and "we are fully insured for every situation". Neither was in the brief. Replace such lines with the qualifications, insurance and vaccination policy you actually hold, stated as facts you could show on request.

  1. Week 1. List 12 vets and groomers and the right person at each; write and check the letter, card and information sheet.
  2. Week 2. Send the letter with the one-page information sheet.
  3. Week 3. Visit with ten referral cards each, with a "referred by" line to be filled in at the assessment.
  4. Weeks 3 to 8. Hold assessments on Tuesdays and Wednesdays, using the score sheet from the tutorial on a dog daycare temperament test.
  5. After each assessment. Email attendees who haven't booked on day 2 and day 7.
  6. Weekly. Post a "meet the pack" photo only where the owner has agreed in writing, with no tags that reveal where anyone lives.
  7. Week 8. Review against the break-even count below.

The cash cost: 12 assessments at 2 hours of staff time at $16 is $384, credits for the eight who take them up are $120, and printing is $35, so $539. A regular booking two days a week at a $38 day rate brings in about $326.80 a month, and at an assumed 55% contribution that is $179.70. Three client-months cover the campaign. If 7 of the 12 assessed dogs become regulars, each new regular cost $77 to win.

To check it, count letters sent, cards handed over, assessments by referrer (use the "referred by" field on the intake form), bookings within seven days of an assessment, and regulars still booked at day 30. The number that matters most is the share of assessed dogs who book. If fewer than 4 of the 12 do, look at the assessment experience before sending more letters.

One results sheet for all five campaigns

CampaignCash cost (illustrative)Break-evenNumber to trackDecide at
Cafe tasting boxes$358About 8 paid ordersOffices that reorderWeek 6
Salon service launch$152.804 bookingsBookings in Tue and Wed gapsWeek 4
Gym challenge$1404 member-monthsFriend passes that become membersTwo weeks after the finish
Car wash dry-day text$12 per send9 extra washes at 24 redemptionsWashes above the dry-day baselineAfter three sends
Dog daycare letters$5393 client-monthsAssessed dogs who book within 7 daysWeek 8

Keep a weekly log of a few lines, written the same day. A filled-in example:

Campaign: office tasting boxes   Week: 3
Sent: email 2 to 40 offices.   Replies: 9.   Tasting boxes delivered: 6.   Orders: 0.
Notes: two offices asked for a vegan-only option; one wants invoice billing.
Decision this week: add a vegan line to the flyer. Keep going.

Small numbers need care. With 40 emails and 12 replies, one extra reply moves the rate by 2.5 points, so don't read a gap between 25% and 30% as meaningful. Use the counts against your break-even number, change one thing at a time, and let each campaign run its planned length before you edit it. Seasonality and weather affect all five, so note anything unusual in the log, such as a road closure or a local event.

Choosing which campaign to run first

Start from the asset you already have. A consented customer list suits the salon or car wash pattern: the cheapest to run and the quickest to read, because you are messaging people who know you. Capacity but no list suits the outreach pattern of the cafe or dog daycare, which is slower but builds a list as it goes. If your problem is members drifting rather than a shortage of new customers, the gym pattern is the closest fit.

Avoid running two campaigns on the same audience at once, or you won't know which one worked. And keep the AI in its place in all five: it drafts, varies and tidies, while the offer, the facts, the consent and the counting stay with you. The tutorial on a pre-post checklist for local business marketing gives a quick review to run on every asset before it goes out.

Questions about running AI-assisted local campaigns

Do I need a paid AI plan to run these campaigns?

No. The free plans of the main assistants cope with drafting, and Canva's free plan includes a small monthly AI allowance. A paid plan at about $20 a month list price helps only if you hit usage limits or want larger uploads. Spend your effort on the offer and the follow-up, which decide the result, not on tools.

How long should a local campaign run before I judge it?

Run it for the full period you planned, usually four to six weeks, and judge it against the break-even count you wrote down at the start. Weekly changes to the offer make the result unreadable. With small numbers, one booking moves a percentage a long way, so count orders and bookings instead of comparing rates.

Can AI choose my offer?

It can suggest options and do the arithmetic, but the offer is a business decision: only you know your margins, capacity and what you can honour. Give the AI the numbers, ask for three options with the cost of each, and pick the one you could afford to be wrong about.

What should I never let the AI claim?

Anything you can't prove or don't offer: awards, reviews, ingredients, results, guarantees, delivery times or staff qualifications. Tell it to use only the facts you provide, and end every prompt by asking for a list of claims to verify. Then check each one yourself before anything goes out.

Further reads

Sources: Canva, Buffer, MailerLite and Mailchimp pricing pages and Google Business Profile help as checked in October 2026; Meta's advertising standard for health and wellness (updated July 2026). All campaign figures are illustrative assumptions.

Want one of these campaigns built around your business?

On a 1:1 call we can choose the campaign that fits your customer list, capacity and margins, write the prompt around your real offer, and set the break-even count you will check at the end.

Book a 1:1 call with me