Q4 2026 · internal
What to build next, and what to stop building
9 September 2026 · from 625 tagged ads, 2,405 reviews and two years of orders
Every ad in the account has been tagged with who it is written for and how far along they are. Two things are genuinely missing, one pile is built and switched off, and three lanes are carrying almost everything. This page says what to make, for whom, and by when. It is a message plan: campaigns stay broad and are never split by persona or stage.
Ads tagged
625
104 live today
Written for a gift-giver
0
in the biggest month of the year
Three lanes carry
79%
73% of what is live
New assets proposed
11
plus 5 recuts of footage we own
01 · The plan
Five things, in the order they matter
Ordered by what it costs to make against what it is worth. The first needs a shoot. The next two need nobody but us. The last two already exist.
1
Gifting
Shoot this
6 assets
Finished by 31 October
Coverage today0 of 625 ads
Gift-Giverproblem-aware
Why
- November is the biggest month of the year at 15.2% of orders, and 51.3% of November revenue comes from people buying for the first time, the year's peak.
- Three quarters of gift orders are Starter Packs, so the product already exists. What is missing is pictures of it being given.
- Gifting is almost entirely a non-subscriber act: of 86 buyers who said the order was a gift, only 9 had an active subscription. This is acquisition, not retention.
What to make
- The Starter Pack as a complete gift in one box, opened on camera. Pottery out first, candle second, MatchPott last.
- The gift that is still there a year later. Same Pott, different scent, same shelf. This is the only lane where "kept for years" is a gift argument rather than a durability claim.
- A December cut for the man buying. Male share roughly triples in December and he will not open seven product pages the way she does, so this one names two or three safe choices and stops.
unboxing + showcasestyled-realcustomer or none
Worth knowing. This is the one recommendation with no in-account performance behind it, because the lane has never run. It rests on order data and the poll. It is the best-evidenced bet available, but it is a bet.
2
Objection-handling
Make without a shoot
4 assets
Live by mid-October
Coverage today0 of 625 ads
Scent-Led Home Curatorproduct-aware
Why
- Price is her number one hesitation in the poll (109 of 183 non-trivial answers) and 26% of the 168 unhappy reviews. It is the most consistently evidenced objection in the whole dataset.
- Not being able to smell it first is second, and she is already trying to solve it herself: median 7 product pages opened per order, 37 at the 90th percentile.
- Nothing in the account answers either. No shoot is needed for any of this.
What to make
- The price answer, told as arithmetic rather than reassurance: the pottery is bought once, the candle is what you buy again. Never framed as waste.
- A scent-family guide. Which of these is you, in four groups, so a stranger can pick without smelling. This also fills the comparison device, which has one ad and it is 84 days old.
- A "what it actually smells like" cut that uses the reviewers' own words. The scent is what 45% of all reviews talk about, the biggest theme in the corpus.
- Size and colour: the third named hesitation, and the cheapest to answer.
direct-address + comparisonraw-nativefounder or customer
3
The press card
Make without a shoot
1 asset
This week
Coverage today0 of 625 ads
Scent-Led Home Curatorproduct-aware
Why
- Pott has held an Ideal Home placement since February 2026 that has never appeared in an ad.
- It is a graphic, not a shoot. Cheapest item on this page by a distance.
What to make
- One card, the quote and the masthead, on the strongest existing autumn still.
proof-displaydesignednone
Worth knowing. Confirm the placement is still fair to cite before it runs, the same check we are applying to the Tonka review quote.
4
The refill demo, switched back on
Already built, switch it on
3 assets
With the Q4 batch
Coverage today44 built, 6 live
Scent-Led Home CuratorRefill Loyalistsolution-aware
Why
- This nearly went on the page as a gap and it is not one. The mechanic is well covered, it has just been switched off.
- It does double duty: 11% of unhappy reviews are about getting the old wax out, so the demo is also a product fix.
- Recut, do not reshoot. The footage exists.
What to make
- One clean 20 to 30 second demo, no voiceover, warm water to new refill in one take.
- Two recuts of the strongest paused teaching ads with the current autumn scents in frame.
demo + teachingraw-nativenone
5
The founder, who has gone quiet
Already built, switch it on
2 assets
November
Coverage today88 built, 7 live
Scent-Led Home Curatorsolution-aware
Why
- An entire messenger lane has gone dormant: 88 ads built, 7 running.
- The 2023 survey has people volunteering "a delightful, likeable owner" and "authentic, small business" unprompted, without being asked about the founder at all.
- The object is the second biggest review theme at 30%, behind only scent. She is buying something made by someone.
What to make
- Two pieces to camera from the workshop, using the making footage from the September drop as b-roll.
- One should carry the price answer in her own voice, which is the highest-value thing the founder can say.
direct-addressdocumentaryfounder
02 · What to stop making
Three lanes are doing almost all the work
Not because they are bad. Because there is nothing left to learn from another one, and every asset added to them takes delivery away from a lane that has never been tested.
founder-craft192 built · 20 live
social-proof163 built · 13 live
sensory-desire141 built · 43 live
Together 496 of 625 ads, and 76 of the 104 currently running. Seven of the nine assets delivered on 9 September land in these same three lanes.
03 · Not on the plan, deliberately
What we are not making, and why
- Reaction and skit. Off-brand. Both registers fight the calm, crafted thing the product is.
- Narrative brand film. Deprioritised. Hero films rarely earn direct-response delivery at this spend.
- Urgency and offer. Parked by design. Pott runs a launch calendar, not a sale calendar.
- Waste and landfill guilt. Tested repeatedly and it does not convert here. Sustainability belongs in the copy as something she gains, never as something she owes.
04 · Sizing
Why 11 new pieces and not thirty
- 104 ads are already live on the evergreen budget. Meta concentrates rather than explores at that count, so most of a large batch would never deliver.
- The September autumn drop is the worked example: 16 of those ads inside evergreen are running 1.36x against 1.84x for the older ads they sit beside, roughly a £560 drag.
- 11 genuinely new assets, 5 recuts. Only the gifting six need a camera; the rest is either desk work or footage already shot.
- So new work goes into the £50/day testing campaign first, and only what earns its place graduates. Retire spent-out incumbents rather than adding budget.
- New-customer share of revenue peaks at 51.3% in November and bottoms at 23.5% in April. Q4 is the acquisition engine, so the gifting work is the one item on this page that cannot slip.