1 · The headline
Wed to Fri this week against Wed to Fri last week
11 to 13 November compared with 4 to 6 November: the same three weekdays, one week apart.
Sales
£39,082
▼ 11%£43,858 last week
Ad spend
£15,784
▲ 2%£15,465 last week
Meta and Google together.
MER
2.47
▼ 13%2.84 last week
Total sales divided by total ad spend. £2.47 of sales for every £1 of ads.
New-customer orders
251
▼ 13%288 last week
Cost per new customer
£62.89
▲ 17%£53.70 last week
Total ad spend divided by new-customer orders.
Contribution
£580
▼ 78%£2,671 last week
What is left of sales after VAT, product costs, shipping, handling, payment fees and ad spend.
What happened
Gold launched on Tuesday 10 November and sold out the same day: £30,817 of sales and 407 orders, the biggest day of the fortnight. The three days after it were quiet. That is a post-launch dip, and it was deeper than Pott usually sees, because many of your regular customers and email subscribers had just bought.
Almost the whole fall in sales came through email and returning customers. The ads themselves were steadier than the headline suggests: your main Meta campaign held its cost per new customer. Two other things added to the picture: the Christmas tests on Meta cost more per new customer this week, and Google showed your products less often after bid targets were raised on 11 November.
Order value, discounts, subscriptions, stock, tracking and searches for the Pott name all looked normal.
2 · Day by day
Ten days of sales and ad spend, with Gold in the middle
Ad spend was steady at about £5,000 a day throughout. Sales are what moved: a big spike on launch day, then three softer days.
Daily sales and ad spend, 4 to 13 November
Daily MER (sales for every £1 of ad spend)
See the daily numbers
| Day | Sales | Orders | Ad spend | MER | New-customer orders |
|---|---|---|---|---|---|
| Wed 4 Nov | £11,963 | 182 | £5,346 | 2.24 | 82 |
| Thu 5 Nov | £17,037 | 250 | £5,080 | 3.35 | 109 |
| Fri 6 Nov | £14,859 | 240 | £5,038 | 2.95 | 97 |
| Sat 7 Nov | £18,521 | 254 | £5,282 | 3.50 | 109 |
| Sun 8 Nov | £16,938 | 247 | £5,310 | 3.19 | 95 |
| Mon 9 Nov | £12,759 | 219 | £4,776 | 2.67 | 84 |
| Tue 10 Nov (Gold) | £30,817 | 407 | £5,283 | 5.83 | 119 |
| Wed 11 Nov | £13,304 | 197 | £5,049 | 2.63 | 89 |
| Thu 12 Nov | £13,011 | 213 | £5,239 | 2.48 | 75 |
| Fri 13 Nov | £12,767 | 194 | £5,497 | 2.32 | 87 |
Much of this week's missing demand landed on 10 November instead. Taking the six days from 8 to 13 November as one block, sales averaged £16.6k a day at an MER of 3.20, against £14.6k a day and 2.84 for 4 to 6 November. Looked at that way, the period around Gold was ahead of the week before, not behind it.
3 · Where the gap came from
Email accounts for nearly all of the fall in sales
Each order is counted against the last thing the customer clicked before buying. Sales that ended on a Meta or Google ad click were level with last week. Sales that ended on an email click fell by £4,648, out of a total gap of £4,815.
Change in sales, by last click
11 to 13 Nov against 4 to 6 Nov.
Change in new-customer orders, by last click
37 fewer in total (288 to 251).
Inside the email number
Email's own reporting, 4 to 6 Nov compared with 11 to 13 Nov.
£9,584 → £4,728
▼ 51%£13,050 → £9,919
▼ 24%90 → 70
▼ 22%1,608 → 1,124
▼ 30%191 → 169
▼ 12%For context: on Gold launch day alone, 214 returning customers spent £16.9k. "Christmas Bestsellers" also followed four Gold emails in three days, and it had the lowest open rate of the fortnight (58.8%, against 63 to 66% for the others).
4 · Meta
Evergreen held steady. The tests cost more this week.
Meta spend rose 5% (£13,076 to £13,744) and brought in about the same number of new customers (174 to 167), so Meta's cost per new customer went from £75 to £82. Splitting that by campaign shows where the change is.
Meta cost per new customer, by campaign
Ad spend divided by the new customers credited to each campaign. Measured in Triple Whale using "Clicks and Deterministic Views", which counts people who clicked an ad plus people it can confirm saw one before buying.
"All testing combined" is the Christmas creative tests, the landing page test and last week's Fir test. The new round of Christmas tests only went live on 12 November, so it has had two days and is too early to judge.
Evergreen + Gifting spend
£9,363→£9,354
Level. New customers were level too (129 to 128).
Evergreen + Gifting cost per new customer
£72.64→£73.25
Your core Meta activity is doing what it did last week.
All testing spend
£2,605→£3,210
Up 23%. This is where the extra Meta spend went.
All testing cost per new customer
£74.85→£103.56
More spend, slightly fewer new customers (35 to 31).
The same story on every way of measuring
ROAS is the sales credited to Meta divided by Meta spend. Different models give ads more or less credit, so the level differs, but all of them moved the same way: a little lower, with new-customer costs up by 7 to 14%.
| How credit is given | ROAS 4–6 Nov | ROAS 11–13 Nov | Change | Cost per new customer |
|---|---|---|---|---|
| Clicks and Deterministic ViewsClicks plus confirmed ad views | 1.67 | 1.54 | −8% | £75 → £82 |
| Triple AttributionEvery ad platform clicked gets full credit | 2.14 | 1.86 | −13% | £78 → £89 |
| Linear, paid channelsCredit shared across the ad clicks | 1.76 | 1.59 | −10% | £87 → £98 |
| First clickAll credit to the first click | 1.64 | 1.56 | −5% | £95 → £101 |
One more signal behind the higher testing costs: Meta reached 13% fewer people this week (about 214,000 down to 186,000) and showed ads to the same people a little more often. Nothing here looks like tired creative. People clicked at a slightly higher rate than last week.
5 · Google
Google showed your products less often from 11 November
Bid targets were raised on 11 November. A higher target asks Google for more return on each pound, so it bids in fewer searches. Efficiency held, but volume dropped: fewer impressions, less spend and fewer new customers from Shopping and Performance Max.
Daily times your products were shown on non-brand searches
Searches that do not include the Pott name, which is where Google finds new customers.
Google ad spend
£2,389→£2,043
Down 14%.
Non-brand impressions
110.8k→82.8k
Down 25%.
New customers, Shopping + Performance Max
51→34
Down by a third.
Cost per new customer, non-brand Google
£43→£54
Still well below Meta's £89 measured the same way.
Even at £54 it costs less than Meta to bring in a new customer, so this is volume worth getting back. Searches for the Pott name were healthy all week: impressions up 19% and clicks up 18% on last week. The Google and Meta costs here both use Triple Attribution so they compare like for like.
6 · Launch history
How Gold compares with 13 earlier launches
We looked at every launch since September 2025 and compared the days after each one with the seven days before it. Thirteen were clean comparisons (no sale or Black Friday in the way). Because this view uses the week before launch as its baseline, Gold's fall in sales reads as 16% here, against the 11% in the headline.
Sales after a launch, against the week before it
Most launches keep sales above the pre-launch week. Three earlier ones dipped the way Gold has: Raku (Oct 2025), Teal (Jan 2026) and Emily Doran (Aug 2026).
The first three days after launch: Gold against a typical launch
Change against the seven days before launch. The fall in new customers and the rise in their cost are within what Pott has seen before. The fall in email and returning-customer sales is the largest of any clean launch.
See the three similar launches
| Launch | Sales, days 1–3 | New-customer orders, days 1–3 | Cost per new customer, days 1–3 | Sales, days 4–7 | Sales, days 8–14 |
|---|---|---|---|---|---|
| Raku (Oct 2025) | −17% | −19% | +14% | −15% | +1% |
| Teal (Jan 2026) | −11% | −10% | +20% | 0% | −16% |
| Emily Doran (Aug 2026) | −5% | −15% | +18% | −10% | +17% |
| Gold (Nov 2026) | −16% | −13% | +19% | due 17 Nov | due 24 Nov |
| Typical launch (middle of 13) | +30% | +3% | +12% | +25% | +22% |
In the three similar cases, new-customer orders stayed 16 to 19% down through days 4 to 7, and sales were back to the pre-launch level by days 8 to 14 in two of the three (Teal was the exception, in a slow February). When the cost per new customer rises after a launch, it has typically settled back by days 4 to 7. For Gold that points to recovery between roughly 17 and 24 November. It is a small sample, so we treat it as a guide and check it against the real numbers.
Gold early access. Sold out the same day.
Days 1 to 3. The dip covered on this page.
Days 4 to 7. New customers may still run below normal.
Days 8 to 14. Sales usually back to the pre-launch level.
Why Gold dipped more than most: most launches have a public launch or a follow-up email the next day that carries the momentum. Gold sold out within the day, so there was nothing left to follow up with. The three earlier launches that dipped were one-off drops as well.
7 · The plan
What we recommend from here
Most of this week is a lull that should pass on its own. The plan is to protect what is working, fix the two things that would not fix themselves, and refill email demand.
It is bringing in new customers at the same cost as last week. No change needed.
MetaBring testing back to about last week's level by trimming the weaker test sets that have already had two weeks, and let the new Christmas tests finish their run before judging them.
MetaOne adjustment back towards the earlier targets to recover non-brand volume, then leave them alone for two to three weeks so Google can settle.
GoogleThe cost rise is in line with earlier launches and has typically eased within a week. Cutting now would lower new-customer volume going into the busiest part of the year.
Meta + GoogleThe fastest way to refill email and returning-customer demand after the Gold burst.
EmailA next-day email after each drop, even when the product has sold out, is a low-cost way to keep momentum going.
EmailFirst recheck
Wednesday 18 NovemberDays 4 to 7 after Gold are complete. We look at new-customer orders, cost per new customer and how email has responded.
Second recheck
Tuesday 24 NovemberDays 8 to 14 are complete. If new-customer orders are still more than about 10% below the pre-launch 96 a day, and cost per new customer is still above about £58, we treat it as more than a post-launch lull and revisit the plan.
About the numbers. Sales, orders and new-customer figures are from Shopify and match Triple Whale to within one order on every day (the split by last click in section 3 uses Triple Whale, which is why its total gap is £4,815 against £4,776 on Shopify). Ad spend is Meta plus Google. Email figures are from Klaviyo. Contribution uses the cost settings in Triple Whale. All days are UK days, pulled on Saturday 14 November once Friday was complete.
Campaign-level Meta figures use Triple Whale's Clicks and Deterministic Views model, which shares credit between campaigns, so new-customer counts can be fractions and are rounded here. The launch history compares daily averages for each window with the seven days before launch; "typical" is the median of the 13 launches without a sale or Black Friday alongside.
Prepared by Kova Commerce for the Pott Candles team. Please don't share this link outside the team.