Author: Claire Winslow Brown

Your ads are getting clicks. The store is not getting orders. Before anyone asks for another round of creative, open the ad and the page it sends people to, side by side.

Does the product match? Is the advertised offer easy to find? Can someone on a phone buy the thing without dismissing three pop-ups or discovering an unexpected shipping charge?

To tell whether your ads or landing page need work, measure each step separately: who sees the ad, who pays attention, who clicks through, who actually reaches the page, and who completes the next action. Strong ad engagement plus weak on-site conversion is a reason to investigate the destination. It is not, by itself, proof that the page is at fault.

A successful campaign needs the ad, audience, offer, website, and measurement to work together. The useful question is where the evidence tells you to intervene first.

What tells us an ad is doing its part?

Start with the job you hired the campaign to do. Video views can be a sensible awareness objective. A sales campaign needs evidence of qualified shopping activity and, eventually, profitable purchases. Cheap attention is not a substitute.

These are the signals we would bring into a performance review. Compare them with the same account’s history and similar audiences, placements, formats, and objectives. A cold prospect and a returning customer are not a fair head-to-head comparison.

Signal	What to look for	What it does not establish
Audience and delivery	Relevant geography, placement mix, prospecting versus retargeting, and actual search terms for search campaigns	A targeting setting does not prove every impression reached a likely buyer
Hook rate	Whether the opening earns a brief view; explicitly name the view threshold and denominator	Stopping a scroll does not show purchase intent
Video retention	Whether viewers stay long enough to understand the product, offer, and next step	A watchable video can still attract the wrong shoppers
Outbound click-through rate	Clicks leaving the platform divided by impressions, where that metric is available	Clicks do not guarantee the destination loaded or the visitor wanted to buy
Cost per relevant click or landing-page view	Cost of getting a potential customer to the destination, alongside downstream behavior	A cheap visit is not necessarily a valuable visit
CPA, purchase value, and lead quality	Commercial outcomes using consistent attribution and a realistic margin target	Platform attribution alone does not prove incremental revenue

Be especially careful with “CTR.” All-click CTR can include interactions that never take someone to your website. Use the click type that fits the question and keep it consistent across comparisons.

Hook rate also needs a definition. For a Meta video report, you might calculate three-second video plays divided by impressions. Label that exact calculation; do not present it as a universal platform standard. A TikTok report may use two-second views instead. TikTok’s video metric definitions distinguish ordinary views from focused views, which can include interactions. Those numbers are not interchangeable.

A strong opening followed by a sharp drop before the product appears suggests a different creative problem from a weak opening. And a sensational promise can lift CTR while making the eventual sale harder. The creative still owns the expectations it creates.

Follow the click before judging the landing page

First, check that people arrive. Compare outbound clicks with measured landing-page views and site sessions over the same period. A new gap warrants investigation: slow loading, redirects, broken URLs, accidental clicks, consent choices, or missing tracking can all contribute.

Do not treat the difference as an exact count of people who abandoned the page. Platforms and analytics use different counting rules. One person can click more than once, and some visits cannot be measured.

Then follow a consistent on-site funnel. For ecommerce, that might be landing session, product view, add to cart, checkout, purchase. For lead generation, use page visit, form start, completed submission, and qualified lead. Define whether each count represents events, sessions, or people; repeated events can make a funnel misleading.

• People arrive but rarely explore or add to cart: investigate message match, product clarity, price, trust, mobile usability, and visitor intent.
• People add to cart but do not finish: inspect shipping charges, delivery dates, account requirements, payment failures, and checkout usability.
• Forms start but do not submit: test validation errors, required fields, and the mobile keyboard experience.
• Submissions look healthy but sales says the leads are poor: revisit targeting, the promise, qualification, and follow-up. A shorter form is not automatically the fix.

Google’s GA4 engagement definition is broader than buying: an engaged session can qualify through time, a key event, or multiple page views. Use bounce rate and engagement as clues, not a pass/fail grade for the page.

When does the evidence point more strongly to the website?

A visible defect is a good place to start. If the advertised size is sold out, the discount fails, or the submit button is covered on mobile, fix that. You do not need a statistical debate to repair a broken purchase path.

For less obvious problems, look for patterns. Did conversion fall across several otherwise stable campaigns after a website release? Does one destination underperform other relevant pages for comparable traffic? Is the loss concentrated on a device where you can reproduce a usability problem?

Those findings strengthen a website hypothesis. They still need context. Branded search visitors may already know what they want; cold social visitors may be meeting the product for the first time. Comparing their raw conversion rates can make a perfectly usable page look broken.

Google advises keeping the ad and destination aligned, including the offer and requested action. On search campaigns, Quality Score components distinguish expected CTR, ad relevance, and landing-page experience. Treat those as diagnostic context, not proof that a campaign is profitable.

The same clicks can produce very different acquisition costs

Here is a hypothetical example, not a BPM client result. Two periods each spend $1,000, produce 50,000 impressions and 1,000 outbound clicks, and record 800 landing sessions. Outbound CTR stays at 2%; cost per outbound click stays at $1.

In the first period, 32 of those sessions include a purchase: a 4% purchasing-session rate and $31.25 in ad spend per purchasing session. In the second, only eight do: 1% and $125. Assume one order per purchasing session for this simplified example.

The ad’s click metrics look unchanged. The business result does not. Before ordering new ads, inspect what changed after arrival, but also check whether the audience, placement mix, or intent of those clicks changed. Equal click counts do not mean equal traffic quality.

The reverse matters too. If better-qualified creative attracts fewer clicks but more purchases, a lower CTR can be a good trade. Evaluate the campaign against what your margins can support, not whichever metric makes the report look best.

Use this worksheet to choose the next test

Put one row per campaign and landing-page combination in a shared document. Record the date range, objective, audience, device, offer, and attribution settings, then fill in:

1. Impressions, the explicitly defined hook rate, retention, outbound clicks, and spend.
2. Measured arrivals, relevant product or form actions, completed outcomes, and qualified leads or net order value.
3. The first meaningful drop-off compared with an appropriate baseline.
4. The suspected cause, an alternative explanation, and evidence that would contradict your preferred explanation.
5. One proposed change, an owner, a budget, a measurement window, and a success metric.

Verify tracking against actual orders or received leads before interpreting an abrupt collapse. Allow for conversion delay. Twenty visits and no orders is not enough to declare a page broken; the amount of evidence needed depends on the underlying conversion rate and the size of the change you want to detect.

Where traffic allows, randomly split comparable visitors between the existing page and a focused revision while keeping the offer and ads stable. Judge completed outcomes and quality, not just button clicks. A simple before-and-after comparison is weaker because promotions, seasonality, and audience mix can move at the same time.

Agree on the fix, the owner, and the decision

The media team can diagnose a checkout problem without having permission to change the checkout. That makes ownership part of the campaign plan. Decide who can edit the page, who approves the offer, who checks tracking, and when the team will review the result.

If a material conversion problem cannot be fixed yet, discuss limiting spend rather than buying more traffic into it. Creative tests can continue where they answer a useful question, but they should not conceal an unresolved site issue.

If performance has faded on a previously successful ad, our creative-fatigue diagnostic helps narrow that question. For the broader campaign, BPM’s paid social team and Shopify website work address connected parts of the customer experience. Bring the ad, the destination, and the numbers to the same conversation. That is how you get to a useful next step.

Before you send an ad back to the creative team, open the product page. Check the price, the available sizes, and what shipping costs at checkout. A perfectly good ad can become expensive when the thing it is selling gets harder to buy.

Creative fatigue is still worth investigating. You just want enough evidence to give the designer a useful brief.

Meta ad creative fatigue means repeated exposure is reducing an ad’s ability to get a useful response. Treat it as a hypothesis to test, not a diagnosis you can make from frequency alone. Before replacing the ad, check where performance changed: delivery, clicks, the shopping experience, or measurement.

Check what else changed before blaming the ad

Pick two comparable reporting windows. Use the same attribution setting, conversion event, and metric definitions. Include the same weekdays, and allow recent conversions time to appear. A weekend promotion and an ordinary Tuesday are different selling conditions.

Look at the individual ad, then the wider account. If several unrelated ads decline together, investigate shared changes before commissioning a whole new creative batch. Check budget edits, placement mix, product availability, promotions, and website releases.

Frequency is an average across the people reached in the selected period. It is useful context, but it hides differences between people and changes with the reporting window. A blanket “replace every ad at frequency three” rule skips too much of the story.

A creative-fatigue diagnostic you can use in the next meeting

These are investigation prompts, not automatic pause rules. More than one issue can happen at once.

What changed?	What to investigate	A useful next check
Repeat exposure rises while outbound click-through rate falls	Creative wear, audience mix, or a less compelling offer	Compare delivery and response for the same ad over comparable windows; test a meaningfully different concept
Clicks hold up, but fewer visitors buy	Product page, checkout, stock, shipping, price, or traffic quality	Walk the purchase path on mobile and compare site conversion by landing page
Impressions cost more while click response stays similar	Auction conditions or delivery changes	Review CPM and placement mix before assuming the creative stopped working
Reported purchases drop abruptly across ads	Tracking, consent changes, reporting delay, or a real store problem	Compare store orders and measurement diagnostics for the same period
One ad loses delivery while account results remain healthy	A shift in allocation	Check whether the account is still meeting its commercial target

Write down the suspected cause, the evidence against it, and the next test. That last part matters. “Creative fatigue” is not a useful meeting outcome unless somebody knows what to do next.

A falling conversion rate can make the ad look guilty

Here is a simplified example, not a BPM client result. An ad spends $600 and produces 300 clicks. In the first period, those clicks lead to 12 purchases: $50 per purchase. In the next period, the same spend and click count produce six purchases: $100 per purchase.

The cost doubled even though the cost per click stayed at $2. Something after the click, or the quality of the people clicking, deserves attention. This does not prove the landing page is responsible. It does show why swapping the headline is an incomplete response.

Check whether the best-selling size sold out, a discount ended, shipping became more expensive, or the page stopped working properly on a common device. Compare actual store behavior with platform reporting; they will not necessarily match because they measure different things.

For a closer look at the handoff from ad to website, use our guide to diagnosing clicks that do not turn into conversions.

Give the next creative test a real question

If the evidence points toward creative, test a different reason to buy. For a storage product, that might mean replacing a beauty shot with a demonstration of how much fits in a small cabinet. Keep the offer and destination consistent where practical so the result is easier to interpret.

A new background color can be worth testing, but it does not automatically create a new buying argument. Record the concept, intended audience, objection addressed, and outcome you will judge. Decide the budget and evaluation window before launch, based on the account’s conversion volume and acceptable cost.

Ad response changes over time in ways that a single snapshot can miss. A research paper on detecting creative fatigue explores performance trajectories rather than relying on one metric. The paper evaluates its method on synthetic data. It does not establish a universal Meta threshold or guarantee that any particular test will work.

Judge the result against what the business can afford. Our guide to profitable ecommerce ROAS explains why a platform number alone cannot settle that question.

If you need a team to connect the creative brief with account performance and margin, explore BPM’s paid social services. Start with the specific decision: what should stay running, what needs a new concept, and what needs fixing on the store?

Your bestseller deserves a look. So does the product nobody has properly explained yet. Neither deserves the whole ad budget just because somebody likes it.

Choose ecommerce products to advertise by checking what each sale contributes, whether you can fulfill demand, and whether you have a convincing reason for a new customer to buy. Then run a bounded test. Revenue rank and margin percentage alone leave out too much.

Image: original ICONIC product photography shown in BPM’s creative portfolio. The examples and numbers in this article are hypothetical, not ICONIC performance data.

Start with dollars left per order

Calculate the money left before advertising after the expected discount, product cost, fulfillment, payment fees, shipping subsidy, and returns allowance. Use one consistent treatment of refunds so you do not subtract the same loss twice. Exclude sales tax collected for remittance from revenue available to spend.

Then decide how much must remain for overhead and profit. What is left is a planning ceiling for acquisition cost, not a promise that you can buy customers at that price.

Consider two single-product orders. Product A brings in $80 after discounts and leaves $32 after variable order costs. Product B brings in $50 and leaves $25. If the business requires $12 to remain from A and $10 from B after advertising, the planned acquisition ceilings are $20 and $15 respectively.

Product B has the higher contribution percentage before ads: 50% versus 40%. Product A still has more dollars available for acquisition. Which one deserves spend depends on whether actual acquisition costs fit those limits and whether the operational checks below pass.

These figures are illustrative. For multi-item baskets, account for the order’s full economics instead of assigning the entire purchase value to the product shown in the ad. Our ecommerce ROAS guide explains the relationship between margin and return targets.

Shopify’s profit reports depend on recorded product costs. Check that those inputs are complete, and understand which costs the report includes before treating its gross profit figure as money available for ads.

Check stock at the variant level

A product can be in stock while the version people want is unavailable. Look at sizes, colors, bundles, and shipping destinations. Send somebody through the exact buying path the ad will use.

Compare stock coverage with replenishment time and the demand you might create. Shopify’s inventory reporting can provide a historical sales-rate baseline. Adjust your planning for launches or promotions; yesterday’s pace is not a guarantee of tomorrow’s.

If you are clearing old stock, give that campaign its own objective and limit. A clearance decision can reasonably accept different economics from an ongoing customer-acquisition program. Mixing the two makes the next budget meeting harder.

Can the ad give someone a reason to care?

Write one sentence explaining why a person who has never heard of the brand should consider this product. If the sentence could describe every competitor, the brief needs more work.

A storage container might be easy to demonstrate: show what fits, how it stacks, or how the lid works. A supplement may need more explanation and careful support for its claims. A gift may depend on an occasion and a believable delivery date. Different products ask different things of the creative and the page.

Look for usable evidence: a clear demonstration, detailed product information, permissioned customer feedback, or a comparison you can substantiate. Attractive packaging helps the image; it does not settle the buying decision.

A product-selection worksheet for your next planning meeting

Create one row per candidate product or bundle and fill in these fields:

1. Customer and occasion: Who is this for, and what prompts the purchase?
2. Expected order economics: Net revenue, variable costs, required contribution, and planned acquisition ceiling.
3. Supply: Sellable units by important variant, replenishment date, and fulfillment limitations.
4. Buying evidence: Sales and conversion history, traffic source, return reasons, and what remains unknown.
5. Creative idea: The buying reason, the evidence for it, and the asset needed to show it.
6. Test: Budget cap, destination, audience, decision window, and the result that earns another test.

Mark each candidate “ready to test,” “needs work,” or “hold,” with the reason. Do not average away a serious problem. A strong margin cannot compensate for an unavailable product, and beautiful photography cannot fix an unsupported claim.

Give new products a fair test without inventing certainty

A bestseller may benefit from email, loyal customers, or people searching for it by name. A new product sent to cold traffic starts somewhere else. Raw conversion rates across those groups do not prove which product is better for acquisition.

Keep the test conditions as comparable as practical, record the differences, and allow enough time for the buying cycle. Use a pre-agreed budget limit rather than declaring a winner after one sale. Stop for a broken purchase path even if the test period is unfinished.

Future repeat purchases can justify a different acquisition target, but use observed customer cohorts and a payback period the business can fund. An optimistic lifetime-value estimate should not quietly become permission to lose money on every first order.

Bring the shortlist to your paid social team and, where relevant, your paid search team. The same product might need a demonstration in a feed and a precise answer to an existing search. The first decision is which product earns a test. Increasing its budget is a separate decision.

You should not have to become a media buyer to hire one. But your agency cannot see a delayed shipment, a change in margin, or a promotion that somebody mentioned in a hallway.

A paid social agency needs current business constraints, usable creative, appropriate account access, and a clear route to decisions. The agency’s job is to turn those inputs into a plan, explain what is missing, and make the next step easy. The client’s job is to supply the facts and approvals the agency cannot invent.

This is a working checklist for that handoff, at kickoff and whenever something changes. For the broader timeline, see what your first 90 days with a paid social agency should deliver.

Image: Cacao & Cardamom seasonal creative from BPM’s portfolio, an example of the product assets and occasion-specific messaging that go into a brief.

Tell us what a good order looks like

Revenue is a starting point. Share which products leave enough money after discounts, shipping support, fulfillment, and expected returns. If you cannot share detailed costs, agree on approved acquisition-cost or contribution targets with the person who owns the numbers.

Explain whether the priority is first-time customers, repeat orders, a launch, or clearing a particular product. Those are different assignments. A returning customer buying a bestseller may be a good order without proving that a new-customer campaign is working.

Write down the budget ceiling, who can approve a change, and any cash or stock constraints. “Grow as much as possible” is much easier to act on once “possible” has a definition. Our guide to bringing finance into the agency conversation can help with that discussion.

Share the changes that affect the promise in the ad

A promotion brief should include the actual offer, eligible products, exclusions, start and end times with a time zone, and the destination page. Add the person who will confirm that the discount works.

For a seasonal gift campaign, the order deadline and shipping promise matter as much as the headline. If the delivery cutoff changes, the ad and page may both need an update. Put that change in the agreed communication channel; do not assume the media team will discover it at checkout.

The same goes for low stock, new packaging, price changes, product substitutions, and website releases. An agency should ask for a regular update and flag gaps. A client should have one reliable place to send it.

Send source material, not just a logo folder

Useful creative inputs include original product photos, demonstration footage, approved claims with supporting material, common buying objections, and the questions customer support keeps answering.

Tell the team which assets have permission for paid advertising and what restrictions apply. A creator’s organic post is not automatically a usable ad asset. Record the approved usage, duration, and any editing limits rather than leaving the designer to guess.

Feedback works better when it identifies the issue. “This shows the old lid” gives the team a fix. “Can we make it pop?” gives them another round of interpretation. The agency should explain the concept and ask focused questions so you are reviewing a decision, not reacting to a mystery.

Use this shared handoff checklist

• Business target: Client supplies the commercial goal and approved limits. Agency restates them as campaign and reporting decisions. Complete when both sides can explain what earns more spend.
• Product and offer: Client confirms stock, pricing, terms, and shipping dates. Agency checks the ad against the destination. Complete when the advertised path works.
• Creative inputs: Client provides current assets and usage permissions. Agency identifies missing shots or evidence. Complete when the brief can be produced without guesses.
• Access and tracking: Client grants agreed access or connects the relevant owner. Agency tests the access and measurement it needs. Complete when the task works, not when an invitation is sent.
• Approvals: Client names the final approver and backup. Agency supplies a clear review request and deadline. Complete when feedback is consolidated and launch authority is explicit.
• Website fixes: Both sides identify who can implement a change and what is in scope. Complete when a reported problem has an owner, priority, and next action.

Add an owner, due date, and link to the relevant file beside each item. Keep customer details and other sensitive information out of a broadly shared checklist.

Give the right access, then test it

Keep business ownership of your accounts. Use the platform’s supported permissions instead of sending a personal password. Grant what the agreed work needs and review access when responsibilities change.

For example, Shopify supports collaborator accounts with controlled permissions. The agency should specify what it needs to inspect or change. Read-only reporting access and permission to edit a live storefront serve different purposes.

Someone still needs to test the actual workflow. Can the team see the correct account? Can it verify a purchase event? Can the person responsible for the landing page make the agreed fix? Finding that out on launch morning wastes everyone’s time.

What if something is not ready?

The agency should name the dependency, explain its effect, and offer a workable next step. If new footage is delayed, perhaps an approved product demonstration can run first. If checkout is broken, reducing or pausing affected spend may be the responsible move.

Missing inputs should never become a blanket explanation for poor performance. The agency still owns its strategy, execution, reporting, and recommendations within the engagement. A useful update says what is blocked and what work can continue.

When you discuss paid social support with BPM, bring the questions you want off your plate. We can define the scope and the handoffs together, so hiring help actually feels like getting help.