Your media buyer's rules sheet (24 Sep), rebuilt on our own break-even figures. Use it before you turn off, keep or scale any ad. ROAS = pounds back for every £1 of ad spend. Cost per sale = ad spend ÷ sales (Meta calls this CPA). If a cost-per-sale number looks odd, go by ROAS instead, because ROAS adjusts for order size.

The media buyer's sheet used £27 break-even and 1.94 ROAS for every country. That was too loose for the UK and too strict for the US and Canada. US and Canada orders ship from China with no VAT, so each sale keeps more profit. Australia stays at 1.51 / £58.11 and New Zealand at 1.70 (paused). These figures are rechecked every month through Q4.

New-people ads can cost the most, because they bring new customers who buy again later. Hot audiences should always be the cheapest. Today they cost about the same as new-people ads, so check how those audiences are set up.

Do this: Product Aware is the only stage losing money in both countries. Cut it first if budget has to come from somewhere.

“Stays above” means over several days once the spend minimum is passed, not one bad day. New-people ads may run up to 10% past break-even, because the customers they bring buy again. Warm and hot ads get no leeway. If an ad fails only in the US, pause the US copy and keep the UK one running. US ad prices have more than doubled in a year.

1. People scroll past (under 34% watch 3 seconds, or under 6.4% watch 15 seconds) → the video is the problem. Replace it or turn it off.
2. People watch but do not click (click rate under 1.31% UK / 1.15% US) → change the headline, thumbnail or button first.
3. People click but do not add to basket (under 12.1%) → the page is the problem, not the ad. Try another landing page. Leave the ad alone.
4. People add to basket but do not buy (under 26.9%) → price, postage, stock or checkout. Not an ad problem.
5. Cost to show the ad keeps rising and people see it again and again → the ad is worn out, not bad. Make a fresh version.
Give it more time (extra £30–50 UK, £50–80 US, £45–70 Canada) if people watch and click well and only the sales are missing. Never give more time to an ad people scroll past.

On a good day: only trust the morning read once there are at least 5 sales, or spend of 3× the target cost per sale, by about midday. Judge on Shopify sales ÷ total ad spend, not Meta alone. Raise the budget 50%, then 50% again at the next check if it holds. Check every 2 hours on the day's total so far. Go back to the normal budget at the end of the day, whatever happens.
On a proven winner: “proven” = 7+ days above target ROAS with 20+ sales. Raise 20–30% every 2–3 days. If ROAS drops toward break-even after a raise, go back to the previous budget. If ad costs rise but clicks do not, it has hit its ceiling: make a fresh version instead of adding money.

October — cheapest time left to test. Let new-people ads run up to break-even if people are watching and clicking well. This is where November's winners come from.
November — busiest month. Ads will cost 30–50% more to show. Judge on ROAS, not cost per sale, because orders get bigger too.
December — last year got much less efficient (UK ROAS 3.25 → 2.54). Go back to strict targets and protect profit.
Break-evens are recalculated at the start of each month. This card is updated when they change.

The biggest leak is not the ads. Of every 100 people who add to basket, fewer now buy: 27 over the year, 23 in the last 30 days, 20 in the UK last week. Check the postage shown at checkout, items out of stock, and whether the free socks offer shows properly. At November volumes, fixing this is worth more than any single ad decision.