KC
Kova
Commerce
Pott Candles · September 1-8 vs Ailis' forecast
Internal, Casey only · not for the client
Built 9 September 2026 · data through 8 September
The answer

Nothing slowed down. The forecast assumed a level ordinary days have never run at.

September 1-8 landed at 88.8% of Ailis' sheet, -£6,013 short over eight days. Her subscription line is ahead. The entire gap is her online line, which is set at £4,000 to £5,000 a day on days with no launch and no sale. Online revenue has run about £3,500 a day on ordinary days since the start of August, and September 1-8 ran £3,599. New customers, ad efficiency, ad fatigue, the consolidation and the website were each tested below and none of them moved. Two things Casey raised are real and are now in the page: Cook and the Citronella starter packs went out of stock, worth about £543 a day against early August, and the seasonal lift is real but lands after mid-month. Without the stock-outs September's ordinary days would be running ahead of August's, which is what every prior year shows.

Where the gap isThe ordinary-day baselineSeasonal shiftStock-outsNew customersReturningSpendMeta efficiencyFatigueAutumn adsConsolidationWebsiteGooglePagesWhat we can do
£47,508
Revenue, Sept 1-8 (Shopify, net of refunds)
£53,521
Her sheet, days 1-8
-£6,013
Gap · 88.8% of plan · -£752 a day
£1,717
Subscriptions vs her line (ahead)
-£7,204
Online (new + returning) vs her line

01 · Where the gap is

Cumulative revenue, her sheet vs actual, days 1-8
Her sheetActual
The gap by her own lines, days 1-8

Her subscription line includes subscription first orders, so they are counted there here too (£6,176 of them in 8 days). Same total, same basis as her sheet.

02 · Her online number only ever existed on launch and sale days

Online revenue per day (new + returning, excluding subscriptions), 1 August to 8 September
Ordinary dayLaunch or sale dayAugust ordinary-day average £3,518Her September online line
£4,114
August average online per day, all 31 days
£3,518
August average on the 27 ordinary days
£3,599
September 1-8 average, no launch, no sale
£4,000-5,000
Her sheet's ordinary-day assumption

Four August days made the August average: the Emily Doran collab launch on 14 August (£11,068 online) and the Bank Holiday 20% off weekend on 28-30 August (£7,866, £6,210, £7,416). Strip those four days out and August ran £3,518 a day. September 1-8 ran £3,599 with four ordinary emails and no launch. That is the same business at the same speed. Her sheet spreads August's launch-inflated average across launch-free days, which is why every ordinary day in September reads as a miss.

What this means for the month. The fix is in the sheet, not the store: rebase ordinary days to about £3,500 and put the launch and email bumps on top, which is how our own tracker shapes days. Her September total then depends almost entirely on the 12 September Rust & Amber launch and the ad ladder behind it, not on ordinary days catching up.

02b · The seasonal shift is real, and it arrives after mid-month

September 1-8 vs August 1-8, revenue per day, by year
New-customer revenueTotal revenueReturning revenue
UK Google search interest, "candles" (index), by period
August1-15 September16-30 September

Casey is right that September carries a seasonal tailwind, and the last three years show exactly what shape it takes. Two things matter for reading the first eight days:

What it changes. Nothing about the ordinary-day verdict on the sheet, and everything about where the month is won: the seasonal demand and the launch both land in the same fortnight, which is the argument for holding budget for the 12 September step rather than spending it on ordinary days now.

03 · Each suspect, tested

Every candidate cause Casey raised, with the test used and the verdict. Green means the data cleared it. Amber means it is real but small.

Is it Citronella and Cook going out of stock?

⚠ Yes, part of it
  • Line revenue from the two: £630 a day in Aug 1-8, £481 across August, £87 a day in Sept 1-8. That is £543 a day that used to be there against early August.
  • Cook is completely out. All 10 Cook products are at zero (refills, candles, starter packs, the quarterly subscription); only the MatchPott is in stock. Its last sale was 7 September, one unit.
  • Citronella: the Standard and Grand starter packs and the Standard and Grand candles are at zero. Refills in all three sizes (364 units) and the Petite starter pack (44) are in stock. The Standard starter pack, the one new customers bought (34 in August), is the one that is gone.
  • The Cook and Citronella Meta campaigns were switched off on 30 August, and Citronella new-customer orders went from 23 in Aug 1-8 to 2 in Sept 1-8.

This is why the ordinary-day total is flat rather than up with the season. Fig, Roses and Tomato Vine absorbed most of it, but not all. Cook is a Q4 product and her sheet has a "Cook is BACK" email on the 19th (lever 6). Citronella is an outdoor summer line and its window is closing anyway.

Is it new customers?

✓ Not this
  • 34.5 new customers a day in September, identical to the August average (34.5) and +19% on Aug 1-8.
  • Blended CAC on Shopify first orders: £42 against £48 for August. Cheaper, not dearer.
  • New-customer revenue is £2,340 a day vs £2,483 in August (-6%). The whole difference is basket size: AOV £72 to £68 as starter-pack share fell 69% to 63%.

Acquisition volume is fine. The one soft edge is the basket, which is a landing-page and offer question, and it is lever 4 below.

Is it returning customers?

✓ Not this
  • Returning (non-subscription) revenue is £2,031 a day, -18% against the August average.
  • But against August's last ordinary week (20-27 Aug, no launch, no sale) it is +15%.
  • August's returning spike was the 20% Bank Holiday sale and the collab codes (IOW10, EMMA100, KATIESHIPPING). September has none of them. Discounts given fell from £511 to £322 a day.
  • Subscriptions on her basis are £19,238 vs her £17,521: ahead.

This is the August pull-forward we measured on 3 September, not churn. The base is intact; it simply is not being pulled forward by a sale this week.

Are we underspending?

✓ Not this
  • Ad spend 1-8: £11,670 against her budget of £11,668. On budget to the pound.
  • Meta is running £1,144/day and Google £315/day. The evergreen budget was set to £1,100 on 4 Sept; our ladder asked for £1,220 from the 4th and £1,372 from the 8th, so we are about £800 under our own plan in total. Not the £6,000.
  • Should we spend more? At today's 1.60x new-customer ROAS the response curve puts the next pound at about 1.32x, below the 6-month payback line (1.37x) and above the 12-month one (1.05x). More spend at today's efficiency pays back in 8-12 months, not in September.

Spend is not the reason for the gap and extra spend at current efficiency is not the profitable answer either. The ladder step on the 12th is justified only because the launch resets efficiency (Tomato Vine moved CAC 24% in June). Keep the 19 September gate: step on only if trailing CAC on the 12-18 rung is under £47.

Has Meta efficiency dropped?

✓ Not this
  • Account ROAS 1.99x in August, 1.96x in Sept 1-8. Cost per purchase £35.5 to £35.6.
  • Ten straight weeks inside a 1.7x to 2.4x band, at £900 to £1,500 a day.
  • Purchases per click are up (4.10% to 4.76%). The account is converting the clicks it buys at least as well as in August.

Whatever changed in the account structure, the money is coming back at the same rate it did in August.

Are the ads fatiguing?

✓ Not fatigue, it is auction cost
  • Test: the same 33 old evergreen ads that ran in both windows, 17-31 Aug vs 1-8 Sept. ROAS 2.04x to 1.84x, CPA £34 to £40.
  • All of that slip is CPM +17% and click-through -9%, so each click costs 29% more. Purchases per click rose 11%. Fatigued ads convert worse; these convert better and cost more to show.
  • Weekly frequency sits at 1.1 to 1.6. Nobody is seeing these ads too often.
  • CPM hit £20.70 on 8 September, the September high. Last year CPM climbed from £14.31 to £17.28 through September the same way.

Ad 707, a quarter of evergreen spend since mid-August, is drifting gently (2.27x → 2.99x → 1.82x → 1.90x → 1.79x → 1.70x by week). Worth a refresh in the autumn set, not an emergency.

Is it the autumn ads in the evergreen campaign?

⚠ A small real drag
  • 16 autumn ads inside evergreen took £1,165 (14% of its spend) at 1.36x, £53 CPA, beside old ads at 1.84x, £40.
  • The ring-fenced Testing Autumn campaign ran the other 15 at 2.38x, £26 CPA on £410, led by Ad 852 (2.54x).
  • The drag is worth about £71 a day of revenue at the same spend. Real, and fixable today (lever 1), but about a tenth of the daily gap.
  • Their landing-page-view rate is 87% vs 97% for the old ads: the homepage relink is costing some of the clicks.

Ad 837, the Fig hero, is the exception at 1.80x and earns its evergreen place. The rest belong in the test campaign until they beat the old ads.

Is it the consolidation into one campaign?

✓ Not this
  • The account went from 9 spending campaigns on 19 Aug to 6 on 20 Aug, 3 on 31 Aug, then Roses paused on 4 Sept.
  • The campaigns folded in were running 0.91x to 2.03x (Seasalt 0.91, Incremental 1.29, Orangery 1.53, Testing Jul 1.70, Cook 1.91, Citronella 2.03), all below evergreen's 2.05x.
  • Evergreen's own line fell from 2.05x to 1.77x because it absorbed their money. The account did not move: 1.99x to 1.96x.
  • The one debatable pause is Roses: 2.57x and £30 CPA, the best cold campaign in the account (lever 2).

Consolidation concentrated the same efficiency into fewer lines. Looking at the evergreen campaign on its own makes it look like a decline; looking at the account shows it is not.

Is the website converting worse?

✓ Not this, it is converting better
  • Triple Whale pixel conversion rate: 5.46% on August's last ordinary week, 7.04% in Sept 1-8.
  • Meta purchases per click up, landing-page views per click 97% on the old ads.
  • Sessions 1704 to 1658 a day (-3%): slightly less traffic because Meta is spending 16% less than August and there is no launch email in the window.

Pixel sessions undercount true traffic, so read the direction, not the level. Direction: fewer visits, more of them buying.

Are any pages converting worse?

⚠ One page, for one audience
  • Meta traffic, purchases per landing-page view, 17-31 Aug vs 1-8 Sept: Homepage 5.0% → 6.5% · Tomato Vine 5.2% → 5.1% · Orangery 6.6% → 4.9% · Roses 7.8% → 9.2% · Subscribe & save 14.0% → 15.6%. Nothing paid-social lands on is converting worse; the homepage is converting better for ad visitors. (Covers the 47% of Meta spend whose ads carry a plain link; the rest use dynamic link fields.)
  • Google brand traffic landing on the homepage is the one decline: conversions per click 18% → 22% → 19% → 15% → 17% → 14% by week from w/c 3 Aug to w/c 31 Aug, on flat clicks (443, 578, 625, 680, 644, 666). Starter-pack product pages from Shopping: 6% → 11% → 8% → 7% → 9% → 5%.
  • Google reports conversions against the click date and back-fills late ones, so the newest week always reads low. The w/c 17 Aug figure is fully matured and is already below early August.
  • Who lands on the homepage from a brand search? Mostly existing customers. Early August's homepage carried the Emily Doran collab, then the Bank Holiday sale. September's carries the autumn story with no offer. The same returning-customer pull-forward, seen from the page side.

Triple Whale pixel conversion for the whole site is up. Pixel and Meta agree the site is not the problem; the homepage is simply converting fewer returning brand searchers without a sale on it. The CRO re-audit already found the homepage shows no social proof and 82% of mobile sessions never scroll past the collection tiles: both fixes are queued there.

Is it Google?

⚠ One soft spot
  • Brand search is flat: 116 clicks a day in August, 116 in September. Brand demand has not dropped.
  • Shopping High fell from 2.13x to 1.22x on £207 a day, £52 per conversion against Meta's £36.
  • All Google: 6.10x to 3.57x, mostly that one campaign.

Google is 22% of spend. Shopping High is the only line in either platform that got materially worse in September, and it is lever 3.

04 · What we can do before and around the launch, profitably

The launch on the 12th is the big lever and it is already planned. These are the things that improve results at the same spend on ordinary days, ranked by what they are worth. Together the four same-spend levers add about £337 a day of revenue, roughly £137 a day of contribution, or £7,420 of revenue over the rest of September. That closes a bit under half of the ordinary-day gap to her sheet. Cook's return adds roughly £208 a day on top once it is back. The rest of her online number only ever existed on launch and sale days.

Sized levers, revenue per day at the same spend
1

Move the autumn ads out of the evergreen CBO

16 autumn ads inside evergreen took £1,165 (14% of its spend) at 1.36x and £53 CPA, against 1.84x and £40 for the old ads beside them. The ring-fenced Testing Autumn campaign runs 2.38x at £26. Keep Ad 837 (the Fig hero, 1.80x) in evergreen, move the rest to the test campaign and let them earn their way back.

Kova, today
+£71/dayrevenue, same spend
2

Switch Roses back on

Roses was the best cold campaign in the account (2.57x, £30 CPA, £143/day in August and £57/day in September) and was paused on 4 Sept in the consolidation. Evergreen's marginal pound is running about 1.77x. Restore it at £60/day.

Kova, today
+£46/dayrevenue, same spend
3

Trim Google Shopping High, move half to Meta

Shopping High fell from 2.13x in August to 1.22x on £207/day (£52 per conversion vs Meta's £36). Cut it to £100/day, push the £100 into the evergreen budget (that alone gets Meta to the £1,220 rung our ladder asked for), and check the feed carries Fig and Tonka with autumn titles before the 12th.

Kova, this week
+£74/dayrevenue, same spend
4

Get the new-customer basket back to August

New-customer AOV slipped from £72 to £68 as starter-pack share fell from 69% to 63%. Two causes we can touch: the autumn ads land on the homepage (their landing-page-view rate is 87% vs 97% for the old ads) and the canonical Starter Pack page carries no subscription offer. Point the autumn ads at the Fig starter pack, and put the subscription option back on the canonical Starter Pack page.

Kova relink today; Ailis for the PDP
+£146/dayrevenue, same spend
5

Fix the two checkout leaks already found

Zigpoll verbatims (3 Sept): the subscription checkout shows £65-69 recurring when the ad says £25-29, and the "New to Pott £5" code fails for some buyers. Both are unfixed. Neither is sized here because we cannot see how many abandon, but they are free to fix and they sit exactly where the autumn launch traffic will land.

Ailis / dev, before the 12th
Not sizedfree to fix
6

Get Cook back in stock before the 19 September email

Every Cook SKU except the MatchPott is at zero: refills, candles, starter packs and the quarterly subscription. Cook sold £208 a day of line revenue across August and £29 a day in September. Her sheet has a "Cook is BACK" email on the 19th, so the product presumably lands before then. Worth confirming the date now: if it slips, that email has nothing to sell.

Ailis, confirm this week
+£208/dayAugust run rate, not in the total
7

A refill reminder that is not a discount

Returning revenue is not down on ordinary days (£2031/day vs £1769 in the last ordinary week of August), but August's returning spike was a 20% sale plus collab codes, and September has neither. Fig refills are the top returning product this month. A Klaviyo flow that fires 6-8 weeks after a refill purchase, at full price, replaces the sale lift without training the list to wait for 20% off.

Ailis (Klaviyo)
Not sizedfree to fix

What not to do. Do not chase the sheet with budget on ordinary days. At today's efficiency the next pound returns about 1.32x on first order, which loses money in the month and only pays back on a 12-month view. Do not add a discount to close the returning gap; August's sale is exactly why September's returning line looks soft. And do not read the evergreen campaign's own ROAS as the health of the account.

The one number to watch after the 12th. Trailing CAC on the 12-18 rung. Under £47, step to the 19 September rung. Over, hold and close the month at £2,848 a day. Her sheet's £248,177 needs 1.85x new-customer ROAS and £36 CAC for the rest of the month; we are at 1.60x and £42 today, with CPMs rising.

Appendix · campaigns that spent since 1 August

CampaignFirst dayLast daySpendROASCPA
Evergreen (old Tomato vine)1 Aug8 Sep£26,2281.97x£36
Roses May 20261 Aug4 Sep£4,7312.57x£30
Testing | Jul 20261 Aug30 Aug£4,6591.70x£41
Citronella May 20261 Aug30 Aug£4,6292.03x£36
Orangery (Aug 2026)3 Aug19 Aug£2,8191.53x£47
Purchase | Incremental Attribution (Jul 2026)1 Aug20 Aug£2,4821.29x£56
Cook Candles1 Aug30 Aug£2,4481.91x£37
Seasalt Refresh Purchase (Jul 2026)1 Aug19 Aug£1,6370.91x£78
Sales | Hot | Retention | Subscription Opt-In | 7D Click 1D View1 Aug8 Sep£1,4754.10x£15
Testing Autumn 20262 Sep8 Sep£4102.38x£26

Sources: Shopify orders on the KC dashboard definitions (revenue net of refunds, new customer = first-ever order, subscription = recurring tag), Meta Marketing API insights at account, campaign and ad level, Google Ads API v25, Triple Whale pixel, Recharge, and her sheet "May targets - September (2)" pulled 8 September. Working folder: clients/pott-candles/sep-1-8-review.