Real brands. Specific numbers.

A consent setting nobody noticed cost $750K and sent CAC up by $2M.

Chapter 1: a Shopify consent misconfiguration cut first-party data to the ad platforms. Chapter 2: the 90% traffic collapse it caused took three days to find.

Baseline: 3 days to detect, found by manual analysis

A consent banner nobody accepted cost us three days and $750K.

The Beard Struggle

International remarketing was dead. Nobody knew why for weeks.

Improper phone number formatting broke every Klaviyo SMS flow outside the Americas, quietly costing $19K a month.

Baseline: weeks undetected, root cause found in 1 day

Every international SMS silently failed. Nobody saw it for weeks.

A $40K/mo revenue leak, now caught the same day.

A theme update quietly broke a checkout step. Traffic looked normal, so nothing flagged it until the monthly close.

Baseline: 4-week average detection lag on silent leaks

We were losing $40K a month and it never showed up in a dashboard.

Solawave

Creative refresh time halved by ranking ads on revenue, not clicks.

Working out which ad actually drove sales took two to three days of pulling reports and cross-referencing Shopify orders.

Baseline: 6-week refresh cycle, creative ranked on CTR

We were scaling the ads with the best clicks, not the best revenue.

Hiya Health

Creative strategy rebuilt around 90-day LTV, not day-one conversion.

Every ad was judged on day-one conversion because that was the only number fast enough to act on. Cheap converts churned by month two.

Baseline: creative judged on day-one conversion only

Our best day-one ad was our worst 90-day ad.

Half Price Drapes

From hearing about problems from customers to catching them in under an hour.

Conversion problems surfaced through support tickets, hours or days late, and every tool reported a slightly different number.

Baseline: problems reported by customers first, hours to days

We used to learn about breakages from our own customers.

ThirdLove

Ending the blended-metrics argument with one number everyone trusts.

Meta claimed one attribution number, Klaviyo another, Shopify a third. Weekly meetings opened with a debate instead of a decision.

Baseline: four tools, four different revenue numbers

Every Monday was an argument about whose number was right.

Three tools became one after outgrowing what a single dashboard could answer.

Triple Whale was not broken. It just stopped answering enough questions once the brand crossed $120M GMV.

Baseline: three tools, none answering the why

The dashboard told us what moved. It never told us why.

A regional margin drop caught two weeks before the board review.

Margin looked healthy nationally while one region quietly lost money. Every report rolled up to a single number.

Baseline: regional margin reviewed quarterly

I would have found out the same day I had to explain it to the board.

Four storefronts run from one Slack channel.

Four brands meant four dashboards and four sets of alerts. The CEO logged into each one just to know if anything needed attention.

Baseline: four dashboards, four alert streams

I used to spend Monday mornings logging into four accounts.

The conversions were happening. The ad platforms never heard about them.

Browser-side tracking dropped a quarter of purchase events after iOS changes. Server-side recovery put them back into the optimisation loop.

Baseline: browser-only tracking, post-iOS match rates drifting

A quarter of our conversions were never reaching the ad platforms.

Attribution was solved. Decisions still were not.

The team already had multi-touch attribution. What they lacked was anything telling them which campaign to cut and which creative to scale.

Baseline: attribution platform in place, decisions still manual

Attribution told us the split. It never told us what to do on Monday.

Retention flows only work if the data reaching them is complete.

Recovered first-party signal was fed back into the email and SMS platform, making post-iOS lifecycle programmes 30% more effective.

Baseline: retention flows firing on incomplete post-iOS profiles

The flows were fine. The data going into them was not.

A billing failure spike looked like churn until someone checked.

Subscription cancellations climbed in a monthly report. The cause was a payment retry configuration, not customer intent.

Baseline: subscription health reviewed in a monthly cohort report

A payment failure spike looked like churn for three weeks.

Replatforming off Magento without losing the numbers.

A platform migration is where tracking, feeds, and flows silently break. Continuous monitoring made the cutover observable instead of hopeful.

Baseline: replatforming with no continuous view of what broke

Replatforming is where tracking quietly dies.

Warehouse-native intelligence at $250M+.

At enterprise scale the truth lives in the warehouse. Consequential reads it natively and puts marketing decisions on the same numbers as finance.

Baseline: warehouse reporting refreshed weekly, marketing separate

The warehouse had the truth. Marketing had a dashboard.