Facebook ads strategy: the weekly loop for Shopify

What is a Facebook ads strategy?

Most answers hand you a funnel diagram.

A Facebook ads strategy is a written weekly loop that decides four things: what you test, what the copy says, what evidence ends a test, and who approves it. Meta’s help centre puts the learning phase exit near 50 optimisation events per ad set weekly, so the loop, not the audience, is the lever.

Account structure is a one-time decision. You set it, it settles, it stops being interesting.

The loop is the part you repeat, and it is the part almost nobody writes down.

Here is the test. Open Meta Ads Manager and answer one question out loud: what did you learn last week? Not what you spent. What you learned.

If the answer is a shrug, you do not have a strategy. You have a live account and a checking habit.

An established Shopify brand already knows the hard things. You know your margin. You know your best-selling variant. You know the three objections your support inbox repeats every week.

What is missing is a loop that turns that knowledge into new ads on a fixed rhythm, without rebuilding the whole thing from a blank page each Monday.

That is what this page is. Four jobs, one order, one record. And the fourth job is the one everybody skips.

Why does the weekly loop beat the account structure?

The account settles in days. The creative decides the month.

Meta’s delivery system now does most of the audience work, so the creative is what finds the buyer. Meta’s help centre also puts the learning phase exit near 50 optimisation events per ad set each week. The repeatable job is producing and judging creative, not rebuilding audiences.

Ten years ago the lever was the interest stack. You could out-target a competitor. That lever is mostly gone.

What you can still out-do is the rate at which you learn which message works on which person.

Rate of learning is a systems problem, not a talent problem. It depends on three things you control: how many real concepts you put in front of the machine, how honestly you judge them, and whether last week’s answer is still written down when you start this week.

Most brands lose on the third one. The test ran. Someone remembers roughly how it went. The note lives in a Slack thread. Six weeks later the same idea gets tested again by accident.

That is not a budget problem. That is a memory problem, and it is fixable.

What are the four jobs in the weekly Meta ads loop?

Four jobs. Each one produces something you can point at.

The four jobs are: decide what to test, write the copy that carries it, ship it with a decision rule attached, then read the result and record it. Each job has one output and one approver. A loop with only three jobs is a loop that never remembers what it learned.

Each job has a page under this one.

For the copy craft, go to the before and after Facebook ad copy rewrites. For the test method, go to how to run a creative test you can trust. For the creator pipeline, go to UGC ad examples for Shopify brands.

The rest of this page is the order they go in, and why that order is not negotiable.

What should you test in your Facebook ads this week?

Test concepts, not pixels.

A concept is a claim about why someone buys, written as one sentence you could argue with. “It survives the dishwasher” is a concept. A new button colour is not. Aim for three to five a month, pulled from your support tickets, your reviews, and the Meta Ad Library.

Concepts come from four places you already have, not from a brainstorm:

• Support tickets. The question people ask before they buy is the objection your ad has to answer.
• Three-star reviews. Three-star reviewers explain the trade-off. Five-star reviewers just say thanks.
• Returns reasons. A returns reason names the expectation your ad set wrong.
• The Meta Ad Library. Every active ad on any Facebook page is public and free to read. What has been running for months is what is working for them.

Write each concept on one line. Keep a backlog. When Monday comes you take the top line off the backlog instead of opening a discussion.

That backlog is the whole difference between a brand that tests weekly and a brand that tests when someone feels like it.

One rule that saves real money: a concept is not spent until it has run in at least two formats. A claim can die as a static image and win as a creator video. Do not bury the idea when you have only buried one execution.

What does Facebook ad copy have to do?

Three jobs, in order. None of them is “sound good”.

Facebook ad copy has to confirm what the image already promised, name the person it is for so the wrong reader leaves, and remove the one risk stopping the right reader. Meta recommends about 125 characters of primary text, so all three jobs happen before the “See more” cut.

The mismatch is worth dwelling on.

A scroller sees the image or the first two seconds of video first. Always. Your copy is read second, if at all.

So the first line of primary text is not an introduction. It is a caption that confirms the picture and adds the one thing the picture cannot say.

Then the qualifier. Good ad copy repels. If your product costs three times the category average, say what it is made of in line two and let the price shoppers leave before they cost you a click.

Then the risk. Not the guarantee page. The actual risk in this buyer’s head: will it fit, will it arrive before Friday, will it look like the photo. One line. Specific.

There are eight worked examples on the Facebook ad copy page, written as before and after pairs, so you can see exactly which word did the work.

These four jobs are the same four jobs every week, which makes them a system rather than labour you keep buying. You are not under-staffed. You are un-built.

Not sure which of these four jobs is the one costing you? The assessment reads your week and names the one job worth building first. Take the assessment. About three minutes. Free. No pitch.

Where do creator and UGC ads fit in the loop?

They are a format, not a strategy.

Creator and UGC ads carry proof: a real person using the product in a real room. The entry point is not a creator programme. It is one concept filmed two ways, once by a customer and once by a paid creator, so you can see whether the format or the concept did the work.

The useful way to think about it: a studio image proves what a product looks like. A creator video proves what owning it is like.

Those answer different objections, so judging them against each other on one metric tells you nothing.

They fail for one predictable reason. A brand books five creators, hands each of them the brand’s own polished script, and gets five expensive versions of an ad it could have shot itself.

The UGC ad examples page lists the shot patterns that keep earning their place, the brief that gets them made properly, and the disclosure rules you cannot skip.

How do you know a Facebook ad creative test is finished?

Because you wrote the rule before you launched.

A stopping rule has three parts: the metric you read, the number that ends the test, and what happens next in both directions. Meta’s help centre documents the learning phase behind that volume, near 50 optimisation events per ad set weekly. “Let us see how it goes” is not a rule.

The failure mode here is not statistics. It is human.

You open the dashboard, you find the number that says the thing you hoped, and you call it. Everyone does this.

The only defence is writing the rule before you can see the answer. The rule is how the version of you that had not seen the numbers gets a vote.

Meta’s business help centre documents the learning phase and the roughly 50 weekly optimisation events behind it. Read it once, then set budgets so a test can actually reach that volume.

Splitting a small budget across eight ad sets is the most common way a brand makes its own account unreadable. The creative testing page has the metric ladder and the test log to write it in.

What happens to the paid traffic that does not buy?

Most of it leaves. Plan for that before you raise the budget.

Baymard Institute’s running average across documented studies puts cart abandonment near 70%, so most of the traffic you pay for leaves with items still in the cart. Recovering it costs nothing extra in ad spend, which makes it the cheapest revenue in the account.

That figure is worth checking yourself: Baymard Institute publishes the running average and the list of studies behind it.

This is where the paid loop touches the email side.

If a real number of paid sessions add to cart and leave, the next hour of work is not another creative test. It is the abandoned cart email sequence, which catches buyers you have already paid to reach.

Fix that leak, then come back here. One leak at a time.

What does the Meta ads week actually look like?

Same order, same days, every week.

Monday: pick the concept, 15 minutes. Tuesday: write the copy and the brief, 45 minutes. Wednesday: ship with the stopping rule attached, 30 minutes. Friday: read last week and record it, 30 minutes. Four short sessions, because a loop only compounds if it survives a busy week.

Two things make this survive contact with a real week.

First, the order is fixed. You do not decide on Monday what kind of week it will be. You take the top line off the backlog.

Second, everything lands in one record. Not a dashboard. A written record of what ran, what it cost, what you decided, and what is waiting on your approval.

Here is the shape of that record. We run this loop on our own marketing, because our own marketing was the job that never got done, and this is the format it produces:

Staged data, shown to explain the shape of a run record. It is not a customer result.

The held row is the important one. It carries the reason it was held, and nothing goes live without your yes.

FAQ: What else do people ask about Facebook ads strategy?

The five questions below decide budget: how to build a Facebook ads strategy from scratch, how many ads to run, whether to use Advantage+, how often to refresh creative, and whether the ceiling is the offer. Each answer assumes an established Shopify brand running its own ads.

How do you build a Facebook ads strategy from scratch?

Write the four jobs down before you touch the account: decide the test, write the copy, ship with a stopping rule, read and record. Then pick one fixed day for each. Meta’s learning phase, near 50 optimisation events per ad set weekly, sets the budget floor for a readable test.

How many Facebook ads should I run at once?

Fewer than you think, usually three to five per ad set. The constraint is volume: Meta documents the learning phase exit near 50 optimisation events per ad set each week. More creatives on the same budget means thinner data on every one of them, and thin data cannot be read.

Should I use Advantage+ or manual campaigns?

Use whichever produces a result you can read, then keep it. The four jobs on this page work in both, because they decide which concepts you feed the system and how you judge them. Automated campaigns move the targeting decision to Meta. They do not write your stopping rule for you.

How often should I refresh Facebook ad creative?

On a fixed cadence, not on a feeling. Weekly production with a monthly read suits most established Shopify brands. Variety of concept matters more than frequency: ten versions of one claim fatigue together, because they are one ad wearing ten outfits.

Do I need new creative or a better offer?

Check the offer first. If every concept you test lands in the same narrow cost band, the ceiling is usually the offer, not the copy. Creative testing finds the best expression of an offer. It cannot rescue one the market has already priced.

Keep reading

• Facebook Ad Copy Examples: Before and After Rewrites
• Ad Creative Testing: How to Run a Test You Can Trust
• UGC Ad Examples That Work for Shopify Brands