From a monthly figure to a daily one: the revenue that used to take a month now arrived in a day. The dashboard numbers had been fine; the issue was elsewhere.
The brand sat on WordPress with a broken on-site funnel and a page-test speed score of 6, in the red.
The first decision was migration. We took the numbers to management to argue for Shopify and they agreed. A UI/UX specialist came in, and only two additions: a cart bar showing how far the customer was from free shipping, and a news carousel in the header.
On the selling side, two pieces.
First, customer segmentation. Last year's customers were split across the site's three categories, and each segment received a dedicated SMS with a genuine discount code that appeared nowhere on the site or in any ad, plus its own UTM and a landing page for the category they had actually bought from. The customer receives a message about the thing they already buy, at a price nobody else can see.
Second, creative. Every existing creative plus the winners, new creative in varied formats including deliberately ugly ones, influencer material, and an offer tied to a specific day: a "what do you think happens on day X" creative running on prospecting with all engagers and visitors excluded, so existing buyers keep buying normally instead of waiting for the day, and only new people enter.
Meta structure ran CBO across all variations (VC, LLA, CH), plus ABO, a whitelisted partner, and ASC.
Then the mistake, because nothing runs smoothly.
While scaling I got one wrong. A campaign I was extremely confident in: CBO, lookalikes, custom all variations, 1-day and 7-day prospecting. I set a budget large enough that Meta sent a confirmation email. No sleep. At 7am, ROAS 0.5. I punished it at 9, still nothing, and closed it. Two errors: overconfidence in the campaign, and a custom audience ad set that was still populating. By day's end ROAS recovered slightly, but the money was spent.
Rules that came out of it: exclude anyone who purchased in the last 15 to 30 days so you don't inherit returns and damage brand reputation. In any CBO, set a minimum spend per ad set so it can't decide on a false signal. While scaling, keep a tab and a timer open per winning campaign, logging last ROAS and purchases at intervals. And watch two things in Shopify: sessions and conversion rate. Sessions rising while CR falls means stop scaling.
If you want to sell at large numbers, most of the work is outside the dashboard, and scaling runs on written rules rather than confidence.
The lesson
Most of the work is outside the dashboard
