The audit reflex: what to check before you rebuild anything

Every week, someone tells us the same story in a different shape. An ad account that worked for months suddenly stops converting. A product gets rejected from a sales channel with no clear reason given. An account gets flagged for verification with no explanation attached. The instinct, every time, is to build something new: a new ad, a new campaign, a new application.

That instinct is usually wrong. In almost every case we looked at this week across paid ads, email, product pages, and platform compliance, the actual fix wasn't a new tactic. It was finding the one setting, field, or number that had quietly changed, and correcting that instead.

Here is where to look before you rebuild anything.

Check what's turned on by default

Shopify's Facebook and Instagram sales channel ships with a marketing box that, left checked, automatically offers a percentage off to anyone who hands over an email address through a Meta ad. It stacks on top of whatever offer you already run on your own site. Nobody at the brand turns this on. Meta turns it on for you, tucked under Settings, then Marketing, inside that app. One store owner only found it after noticing customers redeeming a code she never created.

The same instinct applies to the Shop app. Collection thumbnails can quietly pull a random, blurry product photo instead of the one a brand actually uploaded. Neither of these will tank a business on its own. But they are both examples of a default setting working against a brand without anyone noticing, and they cost five minutes to check.

Know your real cost per acquisition before you touch an ad

There is a version of ad management that runs entirely on feel: an ad looks tired, so it gets paused. A number looks bad, so it gets a new headline. That approach is slower and less reliable than just doing the math.

Pull the last 30 days of account data and calculate the real average cost per acquisition. From there, the rule is simple. Once a specific ad has spent two to three times that number and it still is not breaking even, it is done, not paused. It is not going to suddenly recover. One account kept a losing ad running well past $3,700 in spend, watching the return keep sliding instead of stabilizing, when the 30 day average made it clear the ad should have been cut weeks earlier.

When performance flatlines overnight, audit the foundation first

One brand owner described seven straight months of growth on Meta, then a month that simply stopped, with the same budget and the same offer. It is tempting in that moment to blame the algorithm and launch a wave of new creative. Sometimes the algorithm really has changed: post iOS 14 tracking limits and Meta's more recent Andromeda updates have genuinely made the platform work with less data than it used to, which raises the bar on how sharp creative needs to be.

But before assuming that is the whole story, go back to the foundation. Confirm the pixel is still firing correctly. Check whether page load speed quietly slowed down. Compare reported conversions against what the store itself shows. More than one slow month has traced back to something small and fixable, not a permanent platform shift.

A stalled AOV is a catalog question, not a mechanism question

A t-shirt brand had every average order value mechanism already in place: a bundle builder positioned above the add to cart button, a free shipping threshold, a free gift unlock, frequently bought together suggestions, and a post purchase upsell. Average order value still sat around $47 no matter what got rearranged or retested.

The more useful question wasn't which mechanism to add next. It was which products were actually selling. A sales by SKU audit often shows that only three or four designs out of a much larger catalog are moving real volume, and those bestsellers may not naturally appeal to the same audience at all. If the two best selling products don't share a customer, no amount of bundling copy will make people buy both. Catalog depth and audience overlap are worth auditing before adding another app.

One blank field can quietly break a sales channel

A brand moving its product data over to metafields, for cleaner and more consistent listings across hundreds of one of a kind products, started seeing new products get rejected from the Facebook and Instagram sales channel with no clear reason in Shopify's own error messages.

The cause turned out to be simple. The standard product description field was left completely blank once everything moved to metafields, and Meta's policy reads a blank description as incomplete information. The fix did not require abandoning metafields. A single generic sentence in the description field gave the channel something to read, while the detailed specs stayed organized in metafields exactly as planned.

Platform verification denials are a paperwork problem, not a business problem

A brand that had been live on TikTok Shop for weeks suddenly received a re-verification request, then a denial, then another denial after resubmitting with extra documents. It is one of the more common frustrations among brands running TikTok Shop, and it rarely means anything is actually wrong with the business.

Before resubmitting again, audit every document character by character. The legal business name, spelling, punctuation, and ID numbers need to match exactly across every single file, with no exceptions. Upload high resolution originals directly from the source, not screenshots, not cropped images, not anything carrying a watermark. A single mismatched middle initial between a bank statement and a business registration is enough to trigger an automatic rejection. When a clean resubmission still gets denied, some brands have had better luck starting an entirely new application rather than continuing to appeal the same one.

Know your email benchmarks so you can tell what's actually broken

It is hard to diagnose a problem against a number you have never checked. For email specifically: a signup popup converting at least 3% of viewers into subscribers is healthy. Total email revenue should land somewhere around 25 to 30% of overall store revenue for an established ecommerce brand. Campaign open rates should sit at 35 to 40% or higher, and click rates around 1.5 to 2%.

One brand discovered over 7,000 customers who had purchased exactly once and never come back, against only 400 who had shopped three or more times. That group requires no new acquisition cost to reach again. A flow triggered 60 days after a first purchase, surfacing bestsellers or new arrivals, can convert a real share of them before another dollar goes toward new customer ads.

The point of all of it

None of this is complicated. A hidden discount toggle, a blank description field, a mismatched document, an unchecked benchmark: these are small, ordinary things, which is exactly why they get missed. Something that used to work, or should be working, quietly isn't, and the fix is rarely a bigger, newer, more expensive tactic.

Before launching the next ad, app, or campaign, spend five or ten minutes checking what changed first. Most of the time, that is where the actual answer is sitting.