A forensic reverse-engineer of Q4 2025 (what we ran, when, on what offers, with what result) and a data-validated acquisition-timing thesis, built into a decision-ready plan for Q4 2026 across paid, email/SMS, content, offers and budget.
Prepared 22 Jun 2026 | Currency GBP | Owner: Ailis Topley | DRAFT for Casey review (not sent to client)
Cohort verdict: the "go aggressive early" bet is supported by the data
Customers Pott acquired in September and October 2025 are roughly 1.5x more valuable than those acquired during the BFCM rush. The Sep+Oct cohort (n=1,860) shows a 53.0% repeat rate, GBP 173 average LTV-to-date and 28.4% subscription attach, versus 32.7% repeat, GBP 113 LTV and 16.6% attach for the Nov+Dec cohort (n=4,292). The gap is not just a "they have had longer to spend" artifact: at equal maturity (cumulative revenue per customer at month-3), Sep GBP 132 and Oct GBP 140 still beat Nov GBP 110 and Dec GBP 89.
So: turn acquisition spend on hard from early September, treat the September-October window as the cohort-building phase, and let BFCM monetise the demand that phase plus email created, rather than treating BFCM as the moment to go and find new customers cheaply (they are the least valuable cohort of the year).
GBP 1.11M
Store revenue, 1 Sep 2025 - 5 Jan 2026
Shopify blended (truth)
GBP 281k
Paid media spend Q4 2025
Meta 237k + Google 44k
1.53x
Sep/Oct LTV vs BFCM cohort
repeat 1.62x, attach 1.71x
GBP 62.9k
Black Friday (28 Nov), top day
Early Access (27 Nov) #2 at 52.8k
What we are doing differently in 2026 (the headline)
Start the cohort build in early September, not November. Last year the spend curve only stepped up meaningfully in late October and peaked in November; the most valuable customers of the whole quarter were already being acquired in Sep-Oct at far better efficiency and far higher repeat behaviour. In 2026 we front-load prospecting into September, run the proven pre-launch Early-Access signup mechanic again (it produced the second-biggest revenue day of the year off a 2,114-person list), and lean Google into non-brand Shopping and PMax for incremental reach while recognising that brand Search is harvesting, not creating, demand.
Three honesty notes up front.
Revenue windows differ by source and are labelled on every table: Meta and Google were pulled 1 Aug - 31 Dec 2025; Shopify daily and Klaviyo were pulled 1 Sep 2025 - 5 Jan 2026. Do not sum across windows.
Platform ROAS figures in Parts 2 and 3 are platform-attributed (Meta pixel last-touch; Google last-click) and both platforms claim the same orders. Shopify is the blended revenue truth. The Attribution reality check section re-scores both against Triple Whale's deterministic linearAll model: Meta's real contribution is about 1.2x and Google's about 4.6x, not the reported 2.2x and 7.8x.
The Part 5 content review is the one section grounded mostly in performance patterns rather than the visual assets themselves (the physical Q4 shoot was not retrievable in this pass). It is flagged clearly in that section.
Game plan // At a glance
Two halves of Q4: build the cohort, then harvest and gift
The whole plan in one view. Q4 splits cleanly into a September-October build and a November-December harvest and gift, and every lever (spend, content, email) shifts at that line. Each move below names the section that proves or tests it, so the plan sits up top and the evidence sits underneath.
The plan on three lanes
Sep-Oct build phaseBFCM peakNov-Dec harvest & gift
SepOctNovDecJan
Spend
Front-load prospecting
BFCM peak
Taper
Content
Starter Pack, candle-clear
Gifting (clearly candles)
Email / SMS
List build + waitlist
BFCM
Gifting + SMS
A thematic view of the levers. The campaign-level calendar with exact dates is in the Q4 2026 calendar ↓. Green is the high-value build window; blue is the harvest and gifting work.
Sep-Oct vs Nov-Dec, lever by lever
Lever
Sep-Oct · Build the cohort
Nov-Dec · Harvest + gift
Spend
Turn prospecting on hard, well above last year's GBP 26k September. The earliest, cheapest months produced the most valuable customers and the strongest paid efficiency. Cohort thesis ↓Attribution reality check ↓
Spend into the BFCM peak, then taper Google as conversion rates fall in mid-December. Last year the biggest-spend month was the least efficient one. Meta ↓Attribution reality check ↓
Content
Product-forward Starter Pack and refill-ritual creative with total candle clarity. The Starter Pack is the first purchase for 40.6% of new customers. Offers ↓Content ↓
Build the dedicated gifting library we did not have last year, and make sure it unmistakably reads as candles, not ambient lifestyle. Content ↓
Email / SMS
Grow the list and stand up the Early-Access waitlist, the mechanic that produced the #2 revenue day of the whole year. Email & SMS ↓Meta ↓
Fire BFCM and gifting sends, segment hard to protect open rates at peak cadence, and scale the underused SMS channel. Email & SMS ↓
Spend plan: from dashboard targets to campaign allocation
The financial dashboard sets a revenue target for each month but does not yet lock a spend figure. Working back from those targets at last year's blended efficiency gives the spend envelope; the plan then front-loads it, because September and October build the most valuable cohort.
Month
Revenue target (dashboard)
2025 actual paid
Implied at 2025 efficiency
Recommended (front-loaded)
Sep
GBP 194k
GBP 31k
GBP 47k
GBP 64k
Oct
GBP 430k
GBP 57k
GBP 105k
GBP 118k
Nov
GBP 741k
GBP 102k
GBP 181k
GBP 182k
Dec
GBP 580k
GBP 71k
GBP 141k
GBP 110k
Q4 total
GBP 1.95M
GBP 260k
GBP 474k
GBP 474k
Revenue targets: KC financial dashboard (pottcandles plan), live, no spend goals set yet. 2025 actual paid = Meta + Google by month (Meta Graph API, Google Ads API). Implied = revenue target / 4.10x, last year's Sep-Dec blended MER (store revenue GBP 1.07M on paid GBP 260k). Same Q4 envelope, reshaped: the recommendation moves roughly GBP 30k out of December into Sep-Oct. Posture for Casey to size, not a committed figure. Why front-load ↓Efficiency by month ↓
The recommendation roughly doubles September and October against 2025 (Meta's September was only GBP 26k last year) while growing the already-large November and December more modestly. That is the cohort evidence and the linearAll finding (early months were the most efficient) expressed as budget.
Where the Meta budget went, and how to re-weight it
Campaign type (Meta 2025)
Spend
Share
ROAS
2026 adjustment
Evergreen prospecting
GBP 188k
79%
2.0x
Keep as the volume engine, but start in early September, not late October
Event sales & offers (BFCM, Boxing Day, GWP)
GBP 24k
10%
3.6x
Protect and scale at the peak; the most efficient bucket of the year
Testing
GBP 11k
4.5%
2.2x
Maintain, weighted toward the new gifting creative
Seasonal drops (scent launches)
GBP 8k
3.5%
2.7x
Increase; run as waitlist-launched drops (Pudding, Raku, Starlight over-indexed)
Retention / remarketing
GBP 6k
2.5%
3.0x
Scale modestly; subscription retention ran 3.1x
Early-access leads
GBP 0.3k
0.1%
14x*
Fund properly and earlier; it seeded the #2 revenue day of the year
Gifting (dedicated)
~GBP 0
-
-
New line for Nov-Dec, paired with the gifting creative fix below
Meta campaign-level spend, Aug-Dec 2025 (reconciles to GBP 237k). *Early-access is lead generation feeding the email list, not a direct-sales ROAS. Google (GBP 44k) sits outside this split and should weight to non-brand Shopping and PMax. Campaign detail ↓
The content gap we are fixing. Last year's Q4 leaned on image-heavy, ambient creative: there was little dedicated gifting content, and what ran did not clearly communicate that the product is a candle. The 2026 plan treats this as the single biggest creative fix. A product-clear Starter Pack story carries the build phase, then a purpose-built gifting library for November-December shows the candle, the scent and the gift occasion without ambiguity. Detail and the performance evidence in Content ↓
Part 1 // Foundation
The cohort thesis, validated
The entire "go aggressive in September" bet rests on whether early-autumn customers are genuinely more valuable than BFCM discount shoppers. We tested it on the real customer base before recommending anything.
Repeat rate by acquisition month / the cliff is the Oct-to-Nov line
Jul
45.1%
Aug
54.6%
Sep
50.2%
Oct
54.6%
Nov
36.6%
Dec
27.6%
Same source as the table below. Sep-Oct (green) sit at the top of the year; the BFCM cohort (amber) falls off a cliff. Subscription attach and orders-per-customer follow the identical shape.
First-order-month cohorts, 2025
First-order month
Cohort size
Repeat rate
Avg LTV to date
Avg AOV
Orders/cust
Sub attach
Jul 2025
304
45.1%
GBP 149
GBP 61.75
2.59
24.7%
Aug 2025
630
54.6%
GBP 170
GBP 63.36
2.88
26.5%
Sep 2025
670
50.2%
GBP 167
GBP 64.55
2.72
27.6%
Oct 2025
1,190
54.6%
GBP 177
GBP 71.80
2.63
28.8%
Nov 2025
2,402
36.6%
GBP 125
GBP 71.84
1.84
19.1%
Dec 2025
1,890
27.6%
GBP 98
GBP 66.89
1.50
13.4%
Source: pott-customer-dataset.json (29,978 customers, first-ever order keyed to Shopify), analysed 22 Jun 2026. LTV is lifetime-to-date and unequal by cohort age; the maturity-normalised table below controls for that.
Sep + Oct vs Nov + Dec
Sep+Oct
Nov+Dec
Lift
Repeat rate
53.0%
32.7%
1.62x
Avg LTV
GBP 173
GBP 113
1.53x
Orders/cust
2.66
1.69
1.57x
Sub attach
28.4%
16.6%
1.71x
n = 1,860 vs 4,292. Same source.
Maturity-normalised (kills the "older cohort" objection)
Cumulative revenue per customer at a fixed month index, so a 5-month-old cohort is compared to others at the same age.
Cohort
Cum @ m3
Cum @ m4
Aug
GBP 123
GBP 136
Sep
GBP 132
GBP 142
Oct
GBP 140
GBP 152
Nov
GBP 110
GBP 112
Dec
GBP 89
GBP 89
Source: forecast-data/POTTCANDLES/segmented-cohorts.json (generated 21 Mar 2026). Cohort sizes differ slightly from the table above (different generation date).
Verdict. The thesis holds on every metric and survives the maturity check. A September-October customer is worth about 50% more in revenue, repeats 1.6x as often, and subscribes 1.7x as often as a BFCM-acquired customer. BFCM still matters enormously for total revenue, but as a monetisation and returning-customer event, not the cheap-acquisition event it is often assumed to be. The action that follows: open the acquisition taps in early September and treat Sep-Oct prospecting as building the asset that BFCM then harvests.
Part 2 // Paid social
Meta, reverse-engineered
What ran, when, on what offer, with what result, plus the pre-launch signup to Black Friday launch sequence we believe was executed well.
GBP 237k
Meta spend, Aug-Dec 2025
2.23x
Blended ROAS (Meta-attributed)
GBP 529k value, 6,966 purchases
GBP 88k
November spend (peak month)
5x the August baseline
3.31x
BFCM week ROAS (best in-season)
on the biggest spend week
Monthly spend ramp
Aug
GBP 17.8k
Sep
GBP 25.7k
Oct
GBP 50.8k
Nov
GBP 88.1k
Dec
GBP 54.7k
Source: Meta Graph API v21.0, act_552116272670678, time_range 2025-08-01..2025-12-31, pulled 22 Jun 2026. The step-up only really arrived in October; September was still a 26k month, below where Part 1 says it should be.
Top campaigns: volume sits in evergreen prospecting, ROAS sits in warm + event offers
By spend
Campaign / angle
Spend
ROAS
Evergreen broad prospecting
119.5k
1.86x
Advantage+ Shopping (cold)
68.8k
2.23x
BFCM 2025 (manual bids)
14.1k
3.83x
Testing (creative/aud)
10.6k
2.17x
Dec GWP - Free Refill
5.6k
2.42x
By ROAS (spend > GBP 500)
Campaign / offer
ROAS
Spend
Boxing Day - up to 30% off
4.58x
3.9k
BFCM 2025 (manual)
3.83x
14.1k
Retention / Subscription 7D
3.14x
4.6k
Retention (existing)
2.93x
1.2k
Christmas Line (seasonal drop)
2.83x
4.3k
Source: Meta campaign-level insights, same window. Tiny-spend ROAS outliers excluded (BFCM Early-Access leads 14.0x on GBP 261; a returning-customer adset 8.87x). Pattern: broad prospecting carries the volume at ~1.9x; warm/returning and event-sale offers carry the efficiency.
The pre-launch signup to Black Friday launch sequence (confirmed, and it worked)
26 Nov (BF minus 2): Early-Access leads campaign. Objective OUTCOME_LEADS, single broad adset excluding customers (last 365 days), optimising offsite conversions. A cheap waitlist play (GBP 261, 44 leads) that fed the email list rather than driving direct revenue.
28 Nov (Black Friday): dual BFCM sales launch. Two campaigns fired together (Manual Bids + Highest-Volume Backup), both value-optimised, each with twin adsets (broad prospecting + returning customers). A clean primary-plus-backup bid structure on the launch day.
Run-up: Autumn (24 Sep), Rhubarb scent (4 Oct), Christmas Line (16 Oct) and Evergreen (30 Oct) stacked reach before the event (reach jumped to 640k-677k in the 17-24 Oct weeks).
Wind-down: Retention (9 Dec), Remarketing (11 Dec), December GWP Free Refill (11 Dec), then Boxing Day up-to-30%-off (26 Dec).
The arc is unambiguous and well built: build reach through October, capture an early-access list days before BF, launch a dual-campaign sale on the day, then cascade into retention, remarketing, gift-with-purchase and a Boxing Day clearance. This is the single most reusable play in the account and the Klaviyo data (Part 4) shows the signup mechanic paid off directly.
Cost read: auction pressure, not saturation. CPM sat at GBP 12-16 in Aug-Oct, spiked to GBP 18-20 across 31 Oct - 4 Dec (the BFCM auction window), then fell back to GBP 10-13 by late December. Frequency stayed controlled (1.9-2.7) for almost the entire window, rising only to 2.6-3.7 in late December as budget concentrated on a shrinking audience. The cost increase into BFCM was everyone bidding harder, not Pott over-serving its own audience, which means there is room to enter the auction earlier and harder in 2026 without saturating.
Part 3 // Paid search
Google, reverse-engineered
Mix by week, the brand vs non-brand demand curve, and the question that matters: is paid Search just harvesting demand that Meta and email created?
GBP 44.4k
Google spend, Aug-Dec 2025
77.5%
Spend in Shopping
Search 13.2%, PMax 9.3%
2.81x
Non-brand Shopping ROAS
the real scaling line
~93%
Brand Search impression share
near-saturated harvest
Yes, Search is largely harvesting. Brand Search ran at a 36.8x last-click ROAS with ~92-94% impression share every week through BFCM. That combination means brand Search is capturing intent created upstream by Meta, email and organic, not generating new demand. It accounted for about 69% of Google conversions on 22.7% of spend. The genuinely incremental Google lines are non-brand Shopping (2.81x) and PMax (4.21x). Treat the headline 7.83x blended ROAS as flattered by brand harvest and never as a measure of incremental return.
Brand Search: a small slice of spend, most of the credit
Share of Google spend
22.7%
Share of Google conversions
69%
Brand Search takes under a quarter of the budget but is credited with roughly 69% of conversions, at ~93% impression share. A small-spend, large-credit gap is the signature of harvesting demand that Meta, email and organic created, not generating it.
Weekly efficiency around the BF/CM spike
Week start
Spend
ROAS
Conv rate
CPC
17 Nov
GBP 3.5k
10.1x
14.4%
GBP 1.06
24 Nov (BF week)
GBP 4.8k
12.1x
19.8%
GBP 1.17
1 Dec (CM week)
GBP 4.2k
7.5x
13.4%
GBP 1.16
8 Dec
GBP 5.0k
6.6x
13.0%
GBP 1.34
15 Dec
GBP 5.0k
5.2x
12.8%
GBP 1.70
Source: Google Ads API v21, CID 3762144525, segments.date 2025-08-01..2025-12-31, pulled 22 Jun 2026. Conversions are Google last-click and overstate incrementality given the brand pattern.
BF week was the efficiency peak (12.1x, 19.8% conversion rate). The two highest-spend weeks (8 and 15 Dec, ~GBP 5k each) were the least efficient, with CPC climbing to GBP 1.70: classic scaling past peak intent. The lesson for 2026 is to spend into the demand peak and pull Google budget back as conversion rates fall, rather than chasing volume into December.
Only 4 of roughly 24 campaigns carried any spend; the rest were dormant. Non-brand Shopping (the single campaign at GBP 30.2k, 68% of total) is the engine and the right place to invest incremental Google budget in 2026.
Cross-channel // Attribution reality check
What each channel actually contributed (Triple Whale linearAll)
Parts 2 and 3 use each platform's own ROAS, where Meta and Google both claim full credit for the same orders and the totals double-count. This section re-scores Q4 against one deterministic, click-tracked model that sees the whole path. It does not replace the platform numbers, it stress-tests them.
How to read this. Triple Whale's linearAll model follows each order's real click path (its deterministic pixel) and splits that order's revenue equally across every touchpoint, paid and email and organic and direct. So a paid channel only keeps the share of an order it genuinely touched, and channels the ad platforms cannot see (email, organic, direct) finally get counted. Spend is unchanged, so linearAll ROAS = attributed revenue / the same spend. The gap between the two is the over-claim factor. This pull covers 18,357 orders, 1 Aug to 31 Dec 2025, and reconciles to GBP 1.19M, within a few percent of the Shopify blended truth in Part 4 (different window).
GBP 876k → 489k
Meta + Google: claimed vs deterministic
platforms claim 74% of revenue, really drove ~41%
2.23x → 1.21x
Meta ROAS, reported vs linearAll
1.85x over-claim
7.83x → 4.56x
Google ROAS, reported vs linearAll
1.72x over-claim, still genuinely strong
GBP 287k
Email contribution platforms never see
tied with Meta as the #1 channel
Paid channels: what the platform claims vs what linearAll keeps
Channel
Spend
Reported rev
Reported ROAS
linearAll rev
linearAll ROAS
Over-claim
Meta
GBP 237k
GBP 529k
2.23x
GBP 287k
1.21x
1.85x
Google
GBP 44k
GBP 347k
7.83x
GBP 202k
4.56x
1.72x
Reported = Meta omni_purchase value (pixel last-touch) and Google last-click conversion value, same Aug-Dec window as Parts 2-3. linearAll from Triple Whale get-orders-with-journeys-v2, pulled 23 Jun 2026. Over-claim = reported / linearAll.
The November bet was the weakest, on deterministic numbers too. Re-scored month by month, Meta's linearAll ROAS was highest in the under-invested early months and lowest at peak spend: Aug 1.89x on GBP 17.8k, Sep 1.29x, Oct 1.28x, and November just 1.04x on GBP 88.1k, essentially breakeven once credit is shared. Pouring the largest budget of the quarter into BFCM bought the worst marginal efficiency of the quarter. This is the same conclusion Part 1 reached from cohort LTV, now confirmed from the opposite direction: the early-autumn spend worked harder than the BFCM spend.
Meta by month
Spend
linearAll rev
linearAll ROAS
Aug 2025
GBP 17.8k
GBP 33.6k
1.89x
Sep 2025
GBP 25.7k
GBP 33.1k
1.29x
Oct 2025
GBP 50.8k
GBP 65.2k
1.28x
Nov 2025
GBP 88.1k
GBP 91.2k
1.04x
Dec 2025
GBP 54.7k
GBP 63.7k
1.16x
Meta monthly spend from Part 2; linearAll revenue is Meta's shared share of all orders placed that month. Spend rose 5x from Aug to Nov while linearAll ROAS fell from 1.89x to 1.04x.
The real Q4 channel mix, once every touch is counted
Under linearAll the quarter is not a paid-media story with email in support. Three channels sit at almost equal weight with Google close behind, and over half of revenue lands in channels the ad platforms structurally cannot attribute.
Invisible to ad platformsPaid (self-reported high)Mixed / other
Email (Klaviyo)
24.2%
Meta
24.1%
Organic / Social
23.6%
Google
17.0%
Direct
3.4%
Other (loyalty, influencer, SMS)
7.7%
Triple Whale linearAll, Aug-Dec 2025. Green bars (email, organic, direct = 51% of revenue) are channels Meta and Google cannot see in their own dashboards. Exact GBP and platform-visibility notes in the table below.
Channel
linearAll rev
% of revenue
Visible to ad platforms?
Email (Klaviyo)
GBP 287k
24.2%
No
Meta
GBP 287k
24.1%
Yes (self-reported 1.85x high)
Organic / Social
GBP 280k
23.6%
No
Google
GBP 202k
17.0%
Yes (self-reported 1.72x high)
Direct
GBP 40k
3.4%
No
Other (loyalty, influencer, SMS, misc)
GBP 92k
7.7%
Partial
Triple Whale linearAll, Aug-Dec 2025, GBP 1.19M total. "Other" bundles Smile loyalty (GBP 36k), influencer (GBP 34k), SMS (GBP 4k) and a long tail. Paid social and search together hold 41.2% of attributed revenue; email, organic and direct together hold 51.1%.
What this changes for the plan. Nothing in Parts 1 to 7 reverses, and two recommendations get stronger. First, front-loading into September is now supported by deterministic ROAS as well as cohort LTV: the cheap early months were also the efficient ones. Second, email and organic are not support acts, they are half the revenue and invisible to the platforms optimising the spend, which is the strongest possible argument for protecting list growth (the Early-Access mechanic in Part 2) and for not reading platform ROAS as the scoreboard. The honest planning number for incremental paid is closer to Meta 1.2x and Google 4.6x than the reported 2.2x and 7.8x.
Part 4 // Flagship: email, SMS & the daily revenue story
Where the revenue actually moved
Daily store revenue overlaid with the sends and the BF/CM/Boxing-Day anchors, the highest-revenue sends and what they share, and an honest read on list fatigue.
GBP 452k
Klaviyo campaign revenue
email 97% / SMS 3%
GBP 253k
Klaviyo flow revenue
campaigns:flows 64:36
46.8%
Email open rate (2.13M delivered)
1.41% click
GBP 31k
Top single send: Early Access (27 Nov)
2,114 recip, 77% open, 50% click
Monthly store revenue (Shopify, the blended truth)
Sep
GBP 105k
Oct
GBP 238k
Nov
GBP 406k
Dec
GBP 318k
1-5 Jan
GBP 38k
Source: Shopify Admin GraphQL, orders 2025-09-01..2026-01-05, pulled 22 Jun 2026. GBP 1,105,378 total, 16,754 orders. Excludes cancelled and fully refunded; partial refunds not deducted, so slightly above admin Total Sales.
The top revenue days line up with sends and anchors
Date
Store revenue
Orders
What drove it
28 Nov
GBP 62.9k
895
Black Friday live (full-list AM send)
27 Nov
GBP 52.8k
668
Early Access send to the 2,114 signups
1 Dec
GBP 34.0k
528
Cyber Monday
29 Nov
GBP 32.4k
429
BF weekend tail
30 Nov
GBP 28.4k
443
BF weekend tail
15 Nov
GBP 21.3k
269
5th Birthday campaign
22 Nov
GBP 20.2k
248
Pudding Pott collection launch (segmented)
26 Dec
GBP 12.7k
195
Boxing Day (below BFCM, not a top-10 day)
Source: Shopify daily spine + Klaviyo campaign sends, same windows. Nine of the top ten days carry a matching Klaviyo send. The 27 Nov Early-Access email directly created the second-biggest revenue day of the year, validating the signup mechanic from Part 2.
Black Friday, Christmas Offer, Starlight launch, Boxing Day, 5th Birthday, New Season. 39-55% opens at 40k+ recipients. These win on volume and anchor the revenue spikes.
List fatigue: mild and recoverable, concentrated in the BF and early-December stretch. Cadence spiked to 12 email sends per week across BF week and early December; open rate dipped to its two lowest weekly readings of the quarter (43.9% then 40.2%) and December click rate (1.2%) fell below October's (1.8%). Opens recovered to 53-57% once cadence eased in late December. No structural decline, but the implication for 2026 is to protect open rates by segmenting harder during the peak rather than blasting the full list 12 times a week. SMS is underused (7 sends, GBP 13.5k, strong 14.3% click) and is the clearest channel headroom.
Part 5 // Content & creative
What to shoot for 2026
Mapping creative to performance, and a content plan tied to each funnel stage.
Scope honesty. The physical Q4 2025 shoot assets were not retrievable in this pass, so the stated concern ("image-heavy, did not clearly read as candles") could not be visually verified frame by frame. The recommendations below are grounded in what the performance data proves worked (offer and angle level from Meta and Klaviyo), plus the ad-level data sitting in the raw Meta pull for a follow-up creative teardown. Treat the visual critique as a hypothesis to confirm against the actual assets.
What the performance pattern says
Product-launch and waitlist creative outperformed broad brand creative. The highest revenue-per-recipient sends were all specific product moments (Pudding, Raku, Starlight, Early Access), not general brand storytelling. The implication is that clear product-forward creative, with an explicit "this is the thing, here is the offer, here is when," converts; ambient brand imagery does not.
Evergreen broad prospecting on Meta ran at 1.86x while event and warm creative ran 2.8-4.6x. Some of that is audience, but it is consistent with the concern that cold prospecting creative was not doing enough product-clarity work.
The Starter Pack is the acquisition hero (Part 7): 40.6% of Q4 2025 new customers started there. Creative should make the Starter Pack and the refill-ritual unmistakable.
Recommended 2026 content plan
Funnel stage
What the creative must do
Format mix
Pre-launch signup (Sep-early Nov)
Tease the BFCM early-access list; build the waitlist asset that produced the #2 revenue day
Short video + static, scarcity-led
September cohort build
Sell the Starter Pack and the refill ritual with total product clarity
Product-forward video, demo-led
BF launch
Unmistakable offer and urgency; dual prospecting + returning cuts
Static + short video, offer-forward
Gifting (Dec)
Gift use-occasion, bundles, ready-to-give framing
Lifestyle + product, gifting context
Last chance / Boxing Day
Clear deadline and clearance framing
Static, urgency-led
Across all of it the missing-angle hypothesis to test is motion and use-occasion: show the candle being used, the scent storytelling, and the gifting moment, because the static, image-heavy approach is what the client flagged and the data is at least consistent with that concern.
Part 6 // New vs returning & budget posture
How hard to push new acquisition, and when
Measurement caveat. A clean order-level new-vs-returning split as-of-order-date was not cheaply available from Shopify (the customer order-index is only exposed as a current lifetime snapshot, which overstates "returning" because Q4 first-timers who have since reordered now read as returning). The customer-level cohort evidence in Part 1 is the more reliable lens and it is unambiguous.
The scaling tension, stated plainly
Growth requires more new customers, and new customers convert less efficiently than the returning base. But Part 1 proves the resolution: the new customers acquired in September and October become the high-LTV, high-repeat, high-subscription-attach cohort that carries the following year. So the tension is resolved by timing, not by trading growth against efficiency. Acquire hard when the resulting cohort is most valuable (Sep-Oct), then shift the mix toward monetising the returning base and the freshly built list through BFCM and gifting.
Recommended posture for Q4 2026
Start prospecting spend in early September, ahead of last year's late-October step-up. The September 2025 Meta month (GBP 26k) was under-invested relative to the cohort value it produced.
Acquisition vs retention split: weight toward acquisition (Meta non-brand prospecting + Google non-brand Shopping/PMax) through Sep-Oct; rebalance toward retention and returning-customer offers (email/SMS, retargeting, subscription) from mid-November.
Google discipline: fund non-brand Shopping and PMax for incrementality, keep brand Search capped (it harvests), and pull Google budget back as conversion rates fall in mid-December.
Protect efficiency at the peak: Meta showed BFCM cost was auction pressure not saturation, so entering earlier is viable; just hold frequency in check with segmentation rather than full-list blasting.
Part 7 // Offers & products
Evergreen outsold seasonal, and it was not close
Top products, 1 Sep 2025 - 5 Jan 2026
EvergreenSeasonal
Starter Pack
GBP 243k
Christmas Starter Pack
GBP 52k
Seasonal Subscription Box
GBP 43k
Standard Seasonal Sub Box
GBP 32k
Free Standard Xmas Refill
GBP 30k
Standard Refill Subscription
GBP 29k
Grand Starter Pack
GBP 21k
Shopify line items, same window. The Starter Pack alone is 4.7x the top seasonal item and about 22% of window revenue. Units and product types in the table below.
Product
Revenue
Units
Type
Starter Pack
GBP 242,993
3,698
Evergreen
Christmas Starter Pack
GBP 52,229
747
Seasonal
Seasonal Subscription Box
GBP 43,118
625
Seasonal/sub
Standard Seasonal Sub Box
GBP 32,202
467
Seasonal/sub
Free Standard Christmas Refill
GBP 29,646
1,078
Seasonal
Standard Refill Subscription (Rolling)
GBP 29,162
1,420
Evergreen
Grand Starter Pack
GBP 21,459
238
Evergreen
Source: Shopify line items, 2025-09-01..2026-01-05, pulled 22 Jun 2026. Across all 236 products a name-based split is roughly GBP 693k evergreen vs GBP 454k seasonal (~60/40, directional, scent refills like Fir/Myrrh are debatable).
Claim validated. Evergreen outsold seasonal, and it is driven by one SKU: the Starter Pack alone (GBP 243k) is 4.7x the top seasonal item and about 22% of window revenue. It is also the #1 acquisition product, the first purchase for 40.6% of Q4 2025 new customers. The Starter Pack is the front door to the brand and the value ladder into subscription.
2026 lineup and offers to re-run
Lead with the Starter Pack as the evergreen hero across the whole quarter, with the refill-subscription value ladder front and centre (respecting Pott's value-before-price principle, not discount-led).
Re-run the proven offers: the Early-Access signup mechanic (it built the #2 revenue day), the dual BFCM campaign structure, the Boxing Day up-to-30% clearance (best Meta ROAS at 4.58x), and the December GWP Free Refill.
Run seasonal as gifting, not as the acquisition engine: Christmas Starter Pack, Noel refills and the Seasonal Subscription Box are strong gifting and returning-customer plays, but evergreen carries new-customer acquisition. Keep seasonal drops as limited, waitlist-launched moments (Pudding, Raku and Starlight all over-indexed on segmented launches).
Protect the subscription ladder: the Sep-Oct cohort attaches subscriptions at 28% vs 17% for BFCM buyers, so the earlier acquisition window is also the better subscription-acquisition window.
The plan on a timeline
Q4 2026 calendar / Gantt
Anchors for 2026: Black Friday 27 Nov, Cyber Monday 30 Nov, Boxing Day 26 Dec. Every move keyed to the phase it belongs to.
SepOctNovDecJan
Cohort build (Meta + Google prospecting)
Sep ramp through mid-Nov
Non-brand Shopping / PMax push
Sep - early Dec
Early-Access signup mechanic
late Oct - BF
BFCM dual-campaign launch
27-30 Nov
Gifting + GWP Free Refill
Dec gifting
Retention / returning offers
mid-Nov - Jan
Boxing Day clearance
26 Dec
Content shoots delivered
Aug-Sep
Phase plan derived from the Q4 2025 sequence (Part 2) shifted to 2026 anchors, with acquisition front-loaded into September per the Part 1 verdict.
Where the money goes
Budget posture by phase and channel
Anchored to the Q4 2025 actual of about GBP 281k paid media (Meta GBP 237k + Google GBP 44k). The 2026 shift is to move spend earlier, not necessarily to spend dramatically more, and to weight Google toward incremental lines.
Phase
Meta
Google
Primary objective
New vs returning lean
Sep cohort build
scale up vs 2025 (was GBP 26k)
non-brand Shopping/PMax
Prospecting, Starter Pack
New-heavy
Oct reach build
hold high (was GBP 51k)
grow Shopping
Prospecting + seasonal teasers
New-heavy
BFCM (late Nov)
peak (was GBP 88k Nov)
spend into the peak
Dual launch + early-access
Mixed (both)
Dec gifting
moderate (was GBP 55k)
pull back as CvR falls
Gifting, GWP, retargeting
Returning-lean
Boxing Day
short burst (4.58x in 2025)
tail only
Clearance
Returning-lean
Acquisition vs retention split recommendation: roughly 70/30 acquisition-weighted through Sep-Oct, rebalancing to roughly 45/55 returning-weighted from mid-November as the focus shifts to monetising the list and base that the early phase built.
Spend baselines: Meta Graph API and Google Ads API, Aug-Dec 2025, pulled 22 Jun 2026. Percentages are posture recommendations for Casey to size against the final 2026 budget, not committed figures.
Make it happen
Action checklist
Lock content shoots for Aug-Sep delivery. Brief product-forward and use-occasion creative for Starter Pack, refill ritual, and gifting; add motion. OWNER: Ailis + Kova creative
Confirm the Q4 2025 shoot critique against the actual assets, and run an ad-level creative teardown from the raw Meta pull. OWNER: Kova
Stand up the Early-Access signup mechanic by late October. Leads campaign excluding existing customers, feeding a dedicated Klaviyo segment. OWNER: Kova paid + email
Front-load September prospecting on Meta (broad + Advantage+) and Google non-brand Shopping/PMax. Set September budget above the 2025 GBP 26k Meta level. OWNER: Kova paid
Rebuild the dual BFCM campaign structure (manual + backup, prospecting + returning adsets) for 27-30 Nov. OWNER: Kova paid
Plan the December GWP Free Refill and Boxing Day clearance (best Meta ROAS line in 2025). OWNER: Kova paid
Cap brand Search, fund non-brand Shopping/PMax, and set a rule to pull Google budget back as conversion rate falls in mid-December. OWNER: Kova paid
Protect email open rates at the peak by segmenting rather than blasting the full list 12x/week; scale SMS (underused at 3% of campaign revenue). OWNER: Kova email
Lead all acquisition with the Starter Pack and keep the refill-subscription value ladder visible; run seasonal as waitlist-launched gifting drops. OWNER: Kova + Ailis