Shopify Campaign Autopilot: Should You Turn It On?

The Quick Verdict

Find the row that matches your store, then read the rest for the reasoning. Everything here is what Shopify's own pages said on August 15, 2026 — this feature is eight weeks old and still moving.

What Campaign Autopilot Actually Does

Shopify has three different things people call “AI” in the admin, and confusing them is the fastest way to misjudge this one. Sidekick and Shopify Magic help you make things — copy, segments, queries. Agentic commerce is about AI agents buying from you. Campaign Autopilot is the only one that spends your marketing budget, which is why the interesting question about it is not what it can write but what it can do without asking.

It lives under Growth > Autopilot in the admin. It looks at what your store sells and who buys it, proposes marketing tactics across the channels you have connected, and queues each one for you. The queue is the product. Everything else in this article is about what happens on either side of it — what has to be true before tactics appear, and what you should have decided before you approve the first one.

Does Autopilot spend money on its own?

This is the sentence worth reading twice before you decide anything else, because most of the anxiety around handing marketing to an AI dissolves once you know where the gate sits. Shopify puts it in the requirements page rather than the marketing copy:

The same page adds the ceiling: Autopilot can't spend beyond the guardrails you configure for paid advertising, and each paid tactic still needs your approval before it runs. On that page, two independent limits — the gate stops a tactic you never wanted, and the ceiling stops the tactics you did want from adding up to a number you never agreed to.

That page is not the whole rule, and the gap matters before you connect anything. Shopify's setup page documents a Permissions setting in the same part of the admin: “Permissions control whether Autopilot can run tactics on its own or asks for your approval before creating email campaigns and ad campaigns.” The two pages do not agree — the requirements page states the approval gate as a rule, the setup page states it as a switch you can set during setup or change later from Growth > Autopilot, and Shopify has not reconciled them. The resolution is the same one this article gives Starter merchants: open Growth > Autopilot, read your own Permissions before you connect a channel, and don't assume every tactic will wait.

Nothing about that is permanent, either. Shopify's launch post states that you can reject a recommendation, change your budget, or pause at any time, and that you decide how much control you want — from approving every campaign before it runs to giving Autopilot more room as campaigns evolve. The launch post frames autonomy as a dial you set, not a state you accept — one more reason to read your own Permissions before anything connects.

What the recommendations are built from

The recommendations are not generic playbooks. Shopify's overview page says the inputs are your store's products, customers and sales data, which is also why the output gets more specific as the store accumulates history — and why a brand-new store with three orders will see thinner suggestions than one with two years of seasonality behind it.

The question merchants ask next is whether that data goes anywhere else. Shopify says on its product page that the system uses information about your store and category alongside millions of Shopify stores; across the Autopilot overview, its requirements page, the product page and the launch post, read on August 15, 2026, no page uses the word “train” about a general model in either direction — see the summary of what we looked for and didn't find below. Shopify has published a position on this for Shopify Magic, which is a different feature: we cover what Shopify says about training AI on your store data in the guide that owns the built-in AI tools.

Who Can Turn It On Today

Merchants routinely conflate these two and then spend a week improving the wrong thing. One gate is yours: a checklist you can clear this afternoon — though Shopify adds that your store's eligibility is checked periodically and that Autopilot stops running tactics if the store becomes ineligible, which makes it a standard to hold rather than a box to tick once. The other is not yours at all, and no amount of store polishing moves it. Knowing which is which saves the week.

What your store must have first

Read this as a description of a store that can complete an order, because that is what every item on it adds up to. Autopilot buys attention; the requirements exist so the attention lands somewhere that can turn into a sale.

Requirements and considerations for using Campaign Autopilot, Shopify Help Center, read August 15, 2026 (linked in the quote above).

The one that trips people up is the last row. The Marketing staff permission is about people, not about the store: tactics can arrive perfectly well while the person who was supposed to approve them cannot open the section they live in. Sort that before you need it, not during a promotion.

The part Shopify decides

The plan rule is short and absolute: the requirements page excludes exactly one plan — the Agentic plan — and no other, so every other subscription plan is eligible on paper. Eligibility is a different statement from availability, though, because Shopify's overview page adds that the feature is in early access and available only to certain merchants.

How you ask matters less than people expect. On the Campaign Autopilot product page, read on August 15, 2026, the call to action is a Request early access button that leads into your Shopify admin at Growth > Autopilot. No separate form with an email field appears on that page — and Shopify does not use the word “waitlist” for this anywhere we read. What the launch post documents instead is a notification path: if you don't have access yet, still sign up and Shopify will notify you when it's available. The button covers both cases, and the decision stays Shopify's.

On Starter: two rules that don't meet

Put the two published sentences side by side. The requirements page says Autopilot is available on all Shopify subscription plans other than the Agentic plan. The Shopify Messaging page says all merchants with Basic, Grow, Advanced and Shopify Plus plans can use Shopify Messaging. Both are true, and they are about different things — one about a tool, the other about a channel that tool can orchestrate.

Where they meet is exactly where the documentation stops. None of the pages we read on August 15, 2026 — the Autopilot overview, its requirements and setup pages, the email-automations page, the Messaging product page or the Campaign Autopilot product page — defines which channels a Starter store sees inside Autopilot, and the word “Starter” does not appear on the requirements page at all. That is a gap in the documentation, not a rule we can infer in either direction.

What governs, then, is what your own admin offers. Open Growth > Autopilot and read the channel list before you plan a first month around email. If the email channel is there, you have your answer; if it isn't, you have it too — and either way it is worth noting the date you looked, because this is a feature that changed once already this summer.

What Campaign Autopilot Costs

The pricing question has a clean answer, which is unusual for a feature this new. Shopify states it in the launch post rather than burying it in a fee schedule:

The help documentation says the same thing in flatter language: Autopilot is free to use and doesn't have any fees, and you still pay for the marketing activities you approve and run through it. What that removes is the usual worry about automation software — there is no per-seat cost, no percentage of managed spend, and nothing that gets more expensive as the tool does more work.

What the email lane costs

Because the launch post says you pay through Shopify for email “just as you do today”, the email lane keeps its existing pricing rather than gaining a new one. That means 10,000 free emails to send at the beginning of every month, then $1 per 1,000 additional emails on Basic, Grow, Advanced and Plus, up to 300,000 emails a month. The unit is sends, not contacts — a distinction that changes the arithmetic more than anything else on this page.

Worked through: a month in which Autopilot's email tactics send 25,000 emails is 15,000 sends past the free allowance, so it costs $15. That is the shape of the whole email lane — cheap enough that the first tactic you approve there is close to a free experiment, on plans that carry the allowance.

One exception is worth knowing before you plan around the number: abandoned-checkout automations are always free and don't count towards the monthly limit. The mechanics of that automation — what it sends, when, and what it excludes — belong to our guide on Shopify's native abandoned cart recovery, and Autopilot orchestrates it rather than replacing it.

Who bills you for the ads

This matters more than it sounds, because a monthly ceiling set inside Shopify does not consolidate the paperwork. Meta's documentation inside the Autopilot section is explicit: ad spend is billed by Meta, not on your Shopify bill, so you keep an active payment method in Meta Ads Manager. Microsoft works the same way — you complete billing setup and any required verification in your own Microsoft Advertising account.

Shop Campaigns is the exception that makes the model interesting. You are billed when an order is placed as a result of a Shop Campaign, which means the cost follows the sale instead of leading it. The payment page also records a change most merchants missed: as of January 2026, Shop Campaigns fees are charged to your Shopify bill and are no longer deducted from your Shopify Payments payouts. The full economics of that channel sit in our guide to how Shop Campaigns bill you.

What none of those pages give you is a floor. Across the Autopilot setup, requirements and Shop Campaigns pages we read on August 15, 2026, no page names a minimum or maximum budget in money — the documentation describes a monthly budget and daily budgets without ever putting a number on either. Your ceiling is your decision, and there is no published floor to clear before you make it.

Request Access Now, or Wait?

Cost is one half of the answer. The other half is timing — whether a feature eight weeks old, still in early access, is something to adopt now or watch from a distance. The record so far is short enough to read in one go.

Two things stand out in that sequence. The channel list is not fixed — it grew by one in eight weeks, and Shopify names further directions without dates. And the access wording has not moved at all: the same “early access, certain merchants” sentence that shipped in June is what the page said in August.

Shopify's own expectation-setting is unusually plain about what the first weeks look like, and it is a better argument for asking early than any marketing line would be:

So the framework is simple. Waiting protects you from nothing, because access and approval are separate: a store with access and no approvals has spent nothing, and it has a queue of proposals it can read against what it already does. What waiting does cost you is the learning period Shopify names above, which starts when the system starts, not when you decide you are ready.

The risk that is real sits elsewhere — in approving before you have set a ceiling. That is a decision you make once, it takes minutes, and it is the subject of the checklist further down.

What Shopify hasn't published

No page we read names a date for general availability. The wording everywhere is early access, with no window, no cohort schedule and no “rolling out to all merchants by” sentence. Planning a Q4 around Autopilot being generally available is planning around something nobody has promised.

The second silence is louder. No ROAS, CPA or average-result benchmark appears on any page we read — the Autopilot help pages, the product page or the launch post; the performance figures on the product page sit inside an illustrative store mockup. Shopify replaced the number with an expectation, quoted above. Treat any third-party figure you find about Autopilot's performance as unverified against the vendor's own publications, because there is nothing official to check it against.

If the answer is now, the next question is which channel — and that starts with what each one demands of you before it can run at all.

What Each Channel Asks of You

Shopify documents four connected channels as of August 15, 2026 — Shopify Messaging, Shop Campaigns, Meta Ads and Microsoft Advertising. Read that list closely if you were expecting search: the search channel on it is Microsoft Advertising, which Autopilot runs as Performance Max campaigns alongside your other channels — the Meta campaign type sits in the table below. The list also moves: Microsoft Advertising arrived on August 4, 2026, and the launch post names ChatGPT Ads and Snapchat as arriving later, which makes them announced directions rather than channels you can approve today.

Channel pages in the Campaign Autopilot section of the Shopify Help Center, read August 15, 2026.

Both paid channels work the same way here: Shopify's launch post states that Campaign Autopilot can help create a new ad account on Meta and Microsoft Advertising if you don't already have one, or connect to ones you do, and the Meta Ads help page says Autopilot sets up your ad account. What you bring yourself is the Meta Business account it sits under, plus the payment method and the permissions you complete in Meta Ads.

What you can change before you approve

If brand safety is your objection to this whole category, this is the section that answers it — and the answer is not the same on both sides of the queue.

On the email side, the review is genuinely a review. Shopify instructs you to review the email preview, including the subject line, content, target audience and send schedule. You can edit the body text and adjust colors and branding, though Autopilot flags that changes might affect performance. The subject line is the odd one out: you can ask Autopilot to regenerate it, but you can't write a custom subject line directly, which is a real constraint for anyone with a house style for subject lines.

On the paid side, what you review is described more thinly. The Meta and Microsoft pages say to review the campaign details — the ad platform, campaign type, target audience and proposed budget — and the general tactic-management page says to review the tactic summary, including the channel, target audience and expected outcome. None of the pages we read documents previewing or editing the creative itself — the image or ad copy — for a paid tactic; what the tactic-management page does enumerate is three levels of editability across tactics — you can't change the customer segment, you can regenerate elements such as email subject lines and product selections, and you can customize visual elements such as email body content, colors and branding. The product page adds that Autopilot uses your catalog to generate ads and that no additional creative is needed, so the documented gap is about what the approval screen shows, not about who makes the image.

Which Channel Should You Approve First?

The table at the top of this article answers by plan and by what you already run. This answers by the combination — the account you already own, the one change you would need to make to a proposed tactic, and what you want out of the first month. Those three cross in ways a single profile row can't capture.

What to Set Before Your First Approve

Everything in this section costs nothing and takes an afternoon. It is also the part that decides whether the first month is a controlled experiment or a surprise on an invoice — and it starts with Permissions, because that setting decides whether the queue waits for you at all.

Start with the ceiling. the monthly budget is the maximum amount Autopilot can spend across your connected paid advertising channels each month — one number for everything paid, not one per channel. If you already spend on Meta outside Autopilot, set this figure with that in mind rather than as a fresh line item.

What the guardrails actually cap

The setup page describes guardrails as limits on what Autopilot can do on your behalf, “such as” target return on ad spend, target regions and target languages for ads. That “such as” is doing real work — it names examples, not a closed list, so the controls you find in your admin may extend past these.

The target-ROAS field is the one people fill in badly, because it invites a number from outside. Shopify publishes no benchmark for this field on the pages we read — the only defensible starting figure is what your own campaigns already return, and you can tighten it once the queue has produced something to compare.

The third setting is Permissions, documented on the same setup page and quoted in full in the approval-loop section above: it controls whether Autopilot runs tactics on its own or asks for your approval first. Read and set it before the first channel connects — every approval promise in this article is only as good as that switch.

What changes later, and what doesn't

Three levers behave in ways worth knowing before you pull them, and Shopify documents each of them. The one that catches people is timing: a guardrail you tighten today does nothing to a tactic that is already running, which is exactly why the ceiling belongs in place before the first approval rather than after the first surprise.

Managing Campaign Autopilot tactics and campaigns, and Setting up Campaign Autopilot, Shopify Help Center, read August 15, 2026.

Saying no once is not a preference you are teaching it — the middle row is the one merchants most often read backwards, and it matters if you were planning to shape the queue by dismissing things.

Two silences sit here as well. No page we read states how often new recommendations arrive, and none describes what happens to a pending tactic you neither approve nor dismiss — the setup and tactic-management pages describe dismissal and nothing about expiry (see the surface list above). Plan your review rhythm around your own calendar, not around a cadence nobody has published.

The checklist below turns all of that into the order these decisions are best made in. Steps one and two are eligibility work; three and four cap your spend, the Permissions check inside step six decides whether tactics wait for you — and all of them are worth reading before step five connects anything at all.

When Autopilot Isn't Enough

Official documentation never tells you where a product stops being useful, so this section is our judgement rather than Shopify's. It rests on two documented constraints and one absence of them: the audience rules, the scope of what Autopilot automates, and the fact that nothing in the Autopilot section addresses work outside marketing at all.

When you need to choose the audience

The email-automations page states that you can't change the target audience because Autopilot selects the segment to maximize results, and the Shop Campaigns page states you can't change the audience either. For a lot of stores that is a relief rather than a limitation — segmentation is work, and having it done is the point of the feature.

It is a genuine ceiling for the other kind of store. If your programme depends on segments you define — RFM tiers, lapsed-buyer windows, category affinity, VIP cohorts — then the lever you rely on is the one field the tool keeps. That is where a dedicated marketing platform earns its cost, and how segmentation pays off in Klaviyo is the comparison to read before deciding the two can't coexist. They usually can: Autopilot takes the lanes where the segment isn't your lever, and your own tooling keeps the ones where it is.

When the automation isn't marketing

It is easy to read “autopilot” as a promise about the business rather than about the marketing. The scope is narrower than the word suggests: proposals and campaigns on connected marketing channels. It does not move an order into your accounting system, ping a supplier when stock drops, tag a customer after a refund, or hold a workflow together across apps.

Those jobs belong to a different layer, and picking the right one is a decision of its own — our comparison of Shopify Flow, Zapier and n8n covers where each layer starts and stops. And where the requirement is neither marketing tactics nor off-the-shelf workflow — custom feed logic, a segmentation model of your own, an integration between your store and whatever spends the budget — that is custom development work. Most stores never reach that line, and the honest advice for them is to run the native channels well before paying anyone to extend them.

The Bottom Line

Campaign Autopilot is easier to evaluate than most AI features because the risky part sits behind settings you make first. With approval required in your Permissions and a monthly ceiling in place, its worst case is a proposal you decline; its best case is a channel operator that works while you do something else.

Frequently Asked Questions