The Fir test: what we learned
7 new versions of the Ad 1000 idea, each given its own fixed budget (£50 a day). Ran 17 to 27 October 2026, £3,391 spent. Every version carried Ad 1000's exact copy, so the image or video was the only thing that changed.
In short
- Yes, the real-tree scent is the appeal, and we found 3 ads to run alongside Ad 1000 (1029, 1027, 1033), with 2 more worth keeping on (1030, 1031).
- The scent promise works without the needle-drop line. "Smells like a fir tree, gorgeous!" (1027) cost £24 per order. A casual customer-style photo (1029) did just as well, with the most orders of any static.
- The imagery does matter, but polish doesn't. The winning quote on a new Grand photo (1028) lost. The mood-led product video with no stated promise (1032) lost too. Turn both off.
What we were testing
Ad 1000 (the Holly static with the "Christmas tree without the needle drop" review) was taking about a fifth of the account's spend. Ailis built this batch to follow it.
| Ad | Ailis's file name | What it was exploring | Answer |
|---|---|---|---|
 1027Short review, Standard starter pack | Short review, Standard starter pack | Is the realistic Fir scent enough on its own, without the needle-drop comparison? (Similar imagery, but not identical, so Meta treats it as a new ad.) | Yes |
 1028Winning quote, New Grand image | Winning quote, New Grand image | The winning message on a different photograph. Does it still perform? | No |
 1029True-to-life review, Informal photo | True-to-life review, Informal photo | The same promise with a more informal, customer-style feel | Yes, best static |
 1030Not-fake review, Lifestyle photo | Not-fake review, Lifestyle photo | Answering the worry that pine candles smell artificial | Solid |
 1031Founder story | Founder story | Why we made Fir and the brief behind getting the scent right | Too early |
 1032Scent-led product video | Scent-led product video | Bringing the fragrance and Christmas feeling to life in a shorter product video | No |
 1033Pine disinfectant, Starter pack demo | Pine disinfectant, Starter pack demo | Opening on disappointment with other pine candles, then showing ours and how it works | Yes, best video |
Results
Ranked by orders and cost per order first, then return.
Ad 1000 · Holly · "Christmas tree without the needle drop"
Statics (£1,923)
Videos (£1,468)
What we learned
1. The hypothesis holds: the real-tree scent is the appeal
Five of the seven versions cost less per order than Ad 1000 did over the same days. Each one says, in a customer's words or by naming the problem, that it smells like a real tree. The one version that set a Christmas mood without stating the promise (1032, the shorter product video) was the worst ad in the test.
The batch met its aim. We have three new ads that can run alongside Ad 1000 (1029, 1027, 1033) and two more worth keeping on (1030, 1031).
2. The scent promise works without the needle-drop line
1027's "Smells like a fir tree, gorgeous!" has no needle-drop comparison and still cost just £24 per order. The needle drop is a nice extra, not the reason Ad 1000 works. That frees us to use any strong "real tree" review.
3. How much the imagery matters: polish doesn't, what's in the photo does
Polish doesn't. 1029 (informal, customer-style photo) brought in the most orders of any static at £25 each, level with 1027's studio shot. So we can make these quickly from real home photos.
What's in the photo does. 1028 kept Ad 1000's exact winning quote and changed only the photograph, and it lost at £71 per order. Plenty of people clicked and added to cart, but only 21% went on to buy, against 53% to 68% for the two best statics. Our best guess is that the Grand 3-wick shows a bigger, pricier candle than the Starter Packs people land on. That's a hypothesis, and it's one photo, so it isn't proof that every new image would fail.
4. Opening on disappointment with other pine candles is the strongest video
1033 opens with "Warning: smells like a real Christmas tree" and the common letdown of pine candles that smell like disinfectant, then shows ours and how it works. It took the most budget, had the best click rate of the videos (1.9%) and brought in 25 orders. The static version of the same idea, "not fake at all" (1030), was solid too.
5. The founder story is worth another look
1031 got £38 per order but only 8 orders, so it's too early to call. Keep it running at a low level rather than building more founder content yet.
The Christmas creative test shows the same thing
In the separate Christmas test (live since 21 Oct), the "tree without the needle drop" caption won with a creator on camera saying "it smells exactly like a Christmas tree" (Ad 1056, £28 per order, 17 orders). Again, it's a person saying the scent claim.
What to do next
Keep and scale
- 1029, 1027 and 1033 get their own budget, using the same posts so their likes and comments carry over.
- 1030 and 1031 keep running at a low level.
- Expect cost per order to rise as spend goes up. Check again at about double the spend.
Turn off
- 1028 (quote on Grand photo)
- 1032 (scent video, no customer voice)
- Ad 1000 itself stays on. It's still the account's biggest ad.
Make next
- 3 to 4 statics: different customer "real tree" reviews on casual, real-home photos.
- 2 videos: problem-led openers ("most pine candles smell like disinfectant"), each one a different take.
- Stop making mood-only product videos that don't state the real-tree promise. Hold founder content until 1031 has more data.
Caveats
- As Ailis said, some versions change more than one thing at once (1029 has a new review and a new photo style, for example), so read each result as a direction rather than a clean single-variable answer.
- Each version got 3 to 25 orders, short of the ~30 we'd want for a firm verdict. The winners and losers above are clear. The gaps between 1027, 1029 and 1030 are too small to rank them against each other.
- The budget wasn't split evenly between the videos even with a fixed budget per version (1033 got £896, 1032 got £268). That's partly a result in itself: Meta held back on the weaker ads.
- Returns are Meta's own figures (7-day click, 1-day view). All test versions were shown only to people who haven't bought before.
- The 25 October Cook email went out during the test and may have nudged some orders.