Choose one channel (Instagram, an email list of past clients, or local search), have AI draft that channel's content each week, and log every enquiry on a tally sheet with a "paid" column. A paid enquiry is one where a deposit or booking fee has landed. At week six, scale the channel only if paid enquiries cover your time and costs.
Most photographers do the opposite: a bit of Instagram, a newsletter in a burst of enthusiasm, a refreshed website, then a vague feeling that none of it works. Spread thinly, no channel gets enough weeks to show a result, and nothing is recorded, so the feeling can never be checked. Editing and culling are a separate job with their own tools; for those, see how photographers cut editing time and keep their style.
Why one channel and six weeks
A single channel gives you a clean read. If three things change at once (new Instagram captions, a newsletter and a rewritten website page) and bookings rise, you will not know which one did it, so you cannot repeat it. If bookings do not rise, you cannot tell which one to drop. One channel, one change, one tally.
Six weeks is long enough for the habit to settle and for slow-moving channels to produce a first enquiry, and short enough that you can walk away without having sunk a season into a dud. The first two weeks are warm-up: audiences notice a new rhythm slowly, and local search listings take time to be crawled. Weeks three to six are the readable part. Write that on the first page of the tally sheet so you do not panic in week two.
AI earns its place by removing the weekly blank page. A photographer who spends four hours a week writing captions will quit the test in week three; one who spends 45 minutes reviewing drafts will finish it. The AI does the first draft; you supply the facts, the photographs and the judgement about what sounds like you.
Choosing the channel: a three-way decision
Decide by where your past enquiries already came from and by what you can sustain, not by what is fashionable. The table sets out what each option asks of you and what AI drafts for it.
| Channel | Weekly effort | What AI drafts | Best when |
|---|---|---|---|
| 2 to 3 posts, 20 minutes of replies | Captions, a story script, a pinned "how to book" post | Visual genres such as weddings, families and portraits, with a following of at least a few hundred locals | |
| Email to past clients | One email a week | Subject lines, the email body, a re-booking offer | You have 150 or more past clients who agreed to hear from you |
| Local search | One profile update, one review reply batch | Business profile posts, service descriptions, review replies | Clients search by service and area: school, pet, headshot and family photographers |
Three rules break ties. If you have a list of 150 or more past clients, start with email, because those people already know you and the cost of reaching them is close to nothing. If your best clients came through search or a business listing, start with local search. If you have neither a list nor a listing worth anything, Instagram is the default, because it is the one channel that costs only time to begin.
Email is cheaper than it looks. A mailing tool such as Mailchimp lists paid plans from about $13 a month for 500 contacts at list price, and its free tier (up to 250 contacts and 500 sends a month as of October 2026) may be enough for a small list; check the current plan page for your own account, since the regional price shown can differ. Local search profiles are free to maintain. Instagram is free apart from your time.
The tally sheet that makes the test honest
This is the part most marketing advice skips. Create a sheet (a spreadsheet is fine) before week one, with one row per enquiry, whatever its source. Every enquiry gets logged, including the ones you suspect came from elsewhere, because the sheet must also show what the other sources are doing.
| Date | Name or initials | Source (as they said it) | Shoot type | Quoted | Paid? | Deposit date |
|---|---|---|---|---|---|---|
| Week 2, Tue | S. | Instagram post about autumn sessions | Family | $450 | No | |
| Week 3, Fri | R. | Friend told me | Headshots | $280 | Yes | Week 3, Sun |
| Week 4, Mon | J. | Newsletter | Family | $450 | Yes | Week 4, Thu |
| Week 5, Wed | M. | Searched for "family photographer" | Newborn | $600 | No |
Two columns carry the weight. "Source (as they said it)" is the client's own words, captured by one standing question in your enquiry form or first reply: "How did you find me?" Do not clean the answers up while you log them. "Paid?" turns on a rule you write down now: a deposit, a booking fee or a signed contract with money received. A friendly message that says "I'd love a session sometime" is an enquiry; it is not a paid enquiry, and the six-week decision rests on the paid ones.
Add a second small tab with the weekly cost of the test: your hours multiplied by what you would charge for them, any tool subscription, and any boosted-post spend (keep that at zero for this test so the result is organic). Without a cost line you will not be able to apply the week-six rule.
Main example: a family photographer picks email
An illustration, with made-up figures. A family and portrait photographer shoots about four sessions a month at an average of $450. She has 210 past clients on a list, all of whom ticked a box to receive updates when they booked, and an Instagram account with 380 followers. She picks email, because her list is larger than her following and because every person on it has already paid her once.
She sets the test rules before she starts: six weekly emails, one a week, sent on the same weekday; one call to action each week, the same booking link every time; no discounts for the first three weeks, so she can see whether plain reminders work; and a "How did you find me?" question on her booking form. Her costs for the test are about 45 minutes a week of her time, which she values at $40 an hour for this purpose, so $30 a week, plus a mailing tool she already pays for. Six weeks therefore cost her about $180.
| Week | Email theme | Opened by (illustrative) | Enquiries | Paid |
|---|---|---|---|---|
| 1 | A reminder that autumn light is short, with three open dates | 82 | 1 | 0 |
| 2 | A gallery from a recent family session, with the family's permission | 77 | 2 | 1 |
| 3 | What to wear, as a one-page guide | 91 | 1 | 1 |
| 4 | A holiday-card photo session with two dates | 88 | 3 | 2 |
| 5 | Answers to the three questions clients ask most | 70 | 1 | 1 |
| 6 | Last dates before the holidays | 85 | 2 | 1 |
The tally closes at 10 enquiries and 6 paid ones: 6 sessions at $450, which is $2,700 of booked work against $180 of cost. Even if she discounts the result heavily, assuming two of those six would have booked anyway (the tally shows two clients who also mentioned a friend's recommendation), four sessions is $1,800 against $180. The decision rule below turns on exactly this arithmetic.
The weekly AI routine: 45 minutes, same every week
Keep a standing brief that you paste at the top of each drafting session, then add the week's facts. The brief holds your voice, your prices and the rules. The facts are things only you know: the open dates, a recent job, a seasonal reason to book. This is the one prompt worth building carefully, because every week reuses it.
You are drafting marketing for a family and portrait photographer.
Voice: warm, plain, a little dry. British English. No exclamation marks,
no "capture memories", no "stunning". Never invent prices, dates or
testimonials.
Facts you may use: session fee $450 includes 30 gallery images;
headshot sessions $280; sessions run Saturdays and Sundays;
booking requires a $100 deposit at this link: [booking link].
This week's channel: email to past clients (about 200 people).
This week's theme: what to wear. Open dates: [list].
Write: 5 subject lines under 45 characters, one email of 150 to 180
words with a single booking link, and a one-line preview text.
Do not mention discounts.A typical reply (illustrative) offers subject lines such as "What to wear on the day" and "Wear this, skip that", then an email that opens "Hello [first name]" and moves through three sentences about solid colours. Three fixes are usual. First, the draft often tucks in a phrase like "creating timeless memories"; delete it. Second, it sometimes quietly invents a detail such as "I'll send a style guide", which you may not have; check every promise against what you actually do. Third, it tends to end with two calls to action; keep the booking link and cut the second.
The 45 minutes split roughly into 10 minutes for the facts and the prompt, 5 for generating, 20 for editing the draft until it reads like you, 5 for placing the photograph and scheduling, and 5 for logging last week's enquiries. If the edit regularly takes more than 20 minutes, the brief is too thin: add a sample of your own writing, a paragraph you wrote and liked, and ask the AI to match it.
If you choose Instagram instead
The same routine applies with different outputs. Ask for three captions a week and one story script, each ending with the same instruction: "Message me 'dates' for open sessions" or a link in the profile. A single, repeated way to enquire matters, since it lets the tally attribute the result.
Two platform facts help your drafting rules. Instagram now allows at most five hashtags per post, so tell the AI to give you at most five and to prefer specific ones over generic ones. And public posts from professional accounts can be shown in search-engine results unless you switch that off, which means a caption that names the service and the kind of session in plain words does some local-search work for free. A caption that says only "golden hour magic" does none.
Before and after, illustrative. An unedited AI caption: "There's nothing quite like the glow of autumn light on a family who loves each other. Book your session today!" The edited version: "Saturday's family session ran 40 minutes in a park with a dog who refused to sit still. Three autumn Saturday slots are left for family sessions: message me 'dates'." The second one has a real detail, names the service and gives one action. That is the editing standard to hold yourself to.
A trap here is counting likes as results. Likes are a measure of attention from people who are often other photographers. The only Instagram numbers that matter in this test are messages asking about a session, and the tally sheet records those. If you have a few hundred followers and the first four weeks yield no message at all, that is a real finding, not a failure of the AI.
If you choose local search
Here the weekly job is to keep your business profile active and your service descriptions specific. Ask the AI to rewrite each service description (family sessions, headshots, newborn) in 80 to 120 words that name what is included, the length of the shoot and where it happens, using words a client would type into a search box. Add one short profile post a week, and answer every review within three days.
The AI is most useful on review replies, where you can give it the review and three facts and ask for a warm reply of 40 words. Never let it add details of the session that the reviewer did not mention. Never paste a client's full name or a gallery link into a prompt.
Search is slow, so give it the full six weeks. Log the "Source" answer as the client says it, because someone who searched may well say "I found you on Google" or "your name came up", and both belong in the same bucket. If you are curious about the wider picture of what local businesses do with AI marketing, five worked campaigns for local businesses shows several shapes of campaign side by side.
Week six: the stop, scale or extend rule
Decide the rule before week one and write it above the tally, so you cannot bend it afterwards. Here is a workable version with three outcomes.
- Scale: paid enquiries from the channel are at least 3, and the value of those bookings is at least four times the six-week cost (your time plus tools). Keep the routine, add a second weekly output (a story, a second email) and give it another six weeks.
- Extend: there are 1 or 2 paid enquiries, or several unpaid ones. Run weeks seven to nine with one change only: a clearer offer, a different call to action or a new send day. Re-read the tally afterwards.
- Stop: no paid enquiries from the channel, and few or no unpaid ones after week four. Close the test, write down what you learned, and pick the next channel from the table.
The sum for the family photographer: 4 paid sessions credited to email, $1,800 of work, a cost of $180, a ratio of 10 to 1. She scales. A second illustration: a headshot photographer who tests Instagram gets 12 unpaid enquiries but no deposits, because people ask for prices and vanish. That is not a stop; it says the channel gets attention but the offer or the reply is losing people, so she extends for three weeks and changes how she answers price questions.
A third illustration shows why the cost line matters. A newborn photographer spends 3 hours a week on a local-search routine (valued at $120 a week) and lands one $600 booking in six weeks. That is $600 against $720 of time. It is not obviously a loss, since one happy client may refer others and the profile keeps working, but it is not a clear win either, so the rule gives "extend", not "scale".
Using AI to read the tally
At week six, strip names and initials from the sheet and paste the rows into a chat assistant. The point is to catch patterns you have stopped seeing after six weeks of looking.
Below is a six-week enquiry log for a photographer (names removed).
Columns: week, source as the client said it, shoot type, quoted
amount, paid yes/no.
1. Count enquiries and paid enquiries by source.
2. Say which shoot types converted and which did not.
3. Point out any week where enquiries jumped and what I did that week.
4. Do not guess causes you cannot see in the data; say "unclear".
[paste rows]Treat the answer as a second pair of eyes, not a verdict. Check the counts yourself against the sheet, since a chat assistant can miscount a table by a row or two. The "unclear" instruction is useful: it keeps the AI from explaining a spike with a story it invented.
Where the test misleads you
Several things make a test look better or worse than it is.
- Long booking cycles. A wedding photographer's clients decide months ahead. For that genre, judge the six weeks on enquiries that include a date and a budget, plus consultation calls booked, and check paid bookings at three months.
- Seasonal spikes. A holiday-card offer in week four will lift anything you do in that week. Note the season on the sheet so you do not credit the channel for the calendar.
- Referrals mistaken for the channel. If a client says "my friend sent me" after seeing your post, log both: source "friend", and a note "saw post". At the end you count the note as an assist, not a win.
- Changing the offer mid-test. A new price or a discount in week four ruins the comparison. If you must change it, restart the six weeks.
Keeping client details out of the prompts
Marketing drafts need your facts, not your clients' details. Use "[first name]" in sample messages, keep gallery links and surnames out of anything you paste, and check the privacy settings of the assistant you use, in particular the switch that controls whether chats may be used for model training. If you want a checklist for that, what to check in an AI tool's privacy policy and terms covers it. A photograph of a child in a newsletter or post needs the parent's clear permission, which you should record in the booking form rather than rely on memory.
If you would rather build the brief and tally with someone, that is what my AI implementation consultation covers; get in touch. Otherwise, take the sheet, the prompt and the rule above, pick your channel today and send or post the first piece this week. The first result worth reading arrives at week six, so start the clock now. For client replies once the enquiries come in, handling client emails with AI is the next step, and for turning quiet midweek slots into bookings, filling mini-session slots with AI marketing pairs well with whichever channel wins.
Questions photographers ask before the test
Is six weeks long enough for a wedding photographer?
Usually not for bookings, because couples decide months ahead. Keep the six weeks but count earlier signals: enquiries with a date and budget, and consultation calls booked. Judge the channel on those, then review paid bookings at the three-month mark.
Can I run two channels at once to save time?
You can, but you will not know which one worked, and the weekly effort splits in half. If you must, give each channel its own tally column and its own enquiry link, and treat the result as weaker evidence than a single-channel test.
Which AI tool should I use for the weekly drafts?
Any general chat assistant will do: ChatGPT, Claude or Gemini on a free or $20 plan. The tool matters far less than the brief you give it, which is why the prompt block below carries your prices, style and booking rules.
What if nobody enquires in the first two weeks?
That is normal for email and local search, and common for Instagram. Keep posting through week four before you change anything. A test stopped at week two tells you only that two weeks was too short.
Further reads
- AI Marketing for Restaurants: A Four-Week Plan Owners Can Follow — Another owner-run month plan with a one-hour weekly routine.
- AI Marketing for Salons: Fill Diary Gaps With Posts, Emails and Offers — How a diary-driven business fills slots with posts and offers.
- AI Culling vs Outsourced Editing: Which Saves Photographers More? — Where the editing hours go if you want time for marketing.
- How to Price Photography Packages: A Cost-per-Shoot Model With AI — Know your cost per shoot before judging what a channel earns.
- How to Build a Shared Prompt Library for Your Team — Keep the weekly brief somewhere you can reuse it.
- Restaurant Email Marketing: A Weekly Send Written by AI — A weekly email routine you can adapt for past clients.
- AI Marketing for Travel Agents: A Month of Posts and Emails — A four-week content calendar, a destination fact-check table, a supplier claims check and a four-email sequence for travel agents.
- Headshot Photography Business: Reach Corporate Clients With AI Outreach — Build a notes-based target list, draft three honest outreach emails with AI, and offer an on-site team day, with consent and spam rules covered.
- How to Use AI in a Photography Business: Start With Admin, Not Editing — Rank admin, marketing and editing by payback and risk, then start with three admin jobs: enquiry replies, shoot briefs and delivery emails.
- Photographer Gallery Delivery: Reminders and Print Upsells With AI — A dated six-email gallery sequence with prompts, filled-in templates and print offers whose margin you can check by hand.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Instagram help pages on hashtags and public posts; Mailchimp pricing page; ChatGPT, Claude and Gemini plan pages.