Why September has felt flat, and what we're doing about it
Prepared by Kova · 22 September 2026 · Meta and Shopify data to 21 September
Last September dipped in exactly the same way. Ad returns fell from about 2.8x in late August 2025 to 1.7x to 1.9x in the first half of September, then picked back up from the last week of the month.
Two things made it a bit sharper this year: ads cost about 20% more to show than in August, and our Tomato ads have run out of steam while still taking the biggest share of the budget.
Merging the campaigns in late August wasn't the cause. The two weeks straight after it were our best of the quarter.
+40%
new customers vs the same week last year (218 vs 156)
+20%
cost to show our ads, September vs August
1.31x
Tomato ad return last week, down from 2.75x in July
September dips every year
Weekly Meta return on ad spend. Hover or tap a point for the exact figure.
20262025
Late August is strong every year, then the first two weeks of September soften. Last year the recovery started around 22 September and October ran at 2.5x. Rust & Amber launched about 10 days earlier than last year's autumn drop, so the lift may land a little sooner, but it would be unusual to see it before the last week of the month.
What's different this year: the Tomato ads are tired
Tomato carried the account through summer, and it still makes up about a third of what new customers buy first, so demand hasn't gone anywhere. What has worn out is the ads. We haven't had fresh Tomato creative since mid-August, and our best-performing Tomato ad (a customer testimonial that ran about £9,200 in five weeks) slid from 3.0x to about 1.2x before we switched it off on 19 September. Meta kept giving Tomato 40% to 60% of the budget even as it faded. Meanwhile the Autumn range was returning nearly 2x on half the money.
Week of
Tomato share of spend
Tomato return
Autumn range return
20 Jul
34%
2.75x
not live
10 Aug
35%
1.88x
not live
24 Aug
59%
1.98x
not live
7 Sep
48%
1.48x
1.98x
14 Sep
40%
1.31x
1.94x
People are still clicking at the same rate. Fewer of them buy once they land, which is what tired creative looks like.
What we checked and ruled out
Merging the campaigns. In late August we moved to one main campaign because every product sells on the same story: refillable, handmade, any scent. The two weeks after the switch returned 2.40x and 2.06x, our best of the quarter. The dip started two weeks later, alongside the cost increase and the Tomato fade.
Launching lots of new ads. Adding ads can knock a campaign back into Meta's "learning" mode. Ours stayed out of learning throughout September, so that hasn't happened.
What we're doing
Fresh Tomato creative, as a priority. Same message, new faces, hooks and formats. This is the biggest lever we have right now.
Switching off the Tomato ads that have faded, so the budget shifts to the Autumn and Orangery ads that are working.
Testing new ads before they go into the main campaign. New ads get a week in a small test campaign. We move the winners across once a week, in one batch, and switch off the same number of tired ads when we do.
Judging the month from 28 September, when last year's recovery was well under way, rather than on these two weeks.
Meta figures use 7-day click / 1-day view attribution, as reported in Ads Manager. "Return" is Meta-reported revenue divided by spend. New customers are first orders in Shopify, excluding subscription renewals. Weeks run Monday to Sunday (2025 weeks are aligned to the nearest equivalent week).