Case study · Amazon.de · October 2026

Glass bottles & seasonal home decor

Same category, same season, same prices: 4.3× the units sold after optimization

An Amazon seller fully optimized 66 products with HORAiZON ONE and deliberately left 138 products unchanged. We measured 14 days before and 14 days after go-live, against a control group from the same category.

Of the 66 optimized products, 23 sit in categories with a large enough control group (glass bottles, tealight holders; 67 unchanged products there). Only these count for the core comparison. Period: 29 Aug–11 Sep vs. 15–28 Sep 2026.

Starting point

The seller offers a broad range on Amazon.de: glass bottles for filling (swing-top, liqueur and shaped bottles in many sizes and set sizes), seasonal home and Christmas decor such as tealight holders, and garden accessories. The catalogue holds just over 200 products, many of them in variant families. Until September 2026, every listing ran with its original content.

The challenge: measuring fairly

Method

1. 1Splitting the range: 66 products are optimized, 138 products stay unchanged as the control group.
2. 2Full optimization with HORAiZON ONE: Title, bullet points, description and backend keywords. For 18 of the 23 products in the comparison categories, the complete image gallery as well (main image plus up to eight additional images).
3. 3Go-live in three days: All 66 listings went live between 12 and 14 September 2026; the live content was checked.
4. 4Measurement: 14 days before (29 Aug–11 Sep) against 14 days after (15–28 Sep), launch days excluded. The data are Amazon's sales and traffic figures per product.

Result

Stage 1 – Same category, same season

Of the 66 optimized products, 23 sit in categories with a large enough control group: glass bottles (16 optimized, 56 unchanged) and tealight holders (7 and 11). Only these 23 against 67 products count for the core comparison. The other 43 optimized products, such as mood lights, artificial snow or garden furniture, have no usable control group in their category.

The optimized products gain in two places at once: they are viewed more often, and they sell better per view. In the control group only seasonal traffic grows, and the conversion rate even dips slightly. The gain is also broadly spread: 83% of the optimized products grew sales, against 44% of the unchanged ones. The result does not hinge on a few outliers.

Stage 2 – The same product family

The strictest test comes from a glass tealight-holder set: within the same variant family, five variants were optimized and four active variants stayed unchanged. Same product, same season, same customers.

In addition – Best Sellers Rank

In Home & Kitchen, the Best Sellers Rank of optimized product families improved by a median 13%, while the rank of unchanged families worsened by 8% (snapshot 15–17 Sep against 30 Sep–2 Oct 2026).

Does it pay off?

An optimization of the scope in this case study (texts and image gallery, 100 credits) costs approx. €17 per product on the HORAiZON ONE Business plan (€999 for 6,000 credits, about €0.17 per credit). What we measure is the control-adjusted extra revenue: revenue beyond what the control group in the same category gained anyway over the same period.

Extra revenue = revenue after − revenue before × growth factor of the control group, calculated per category. Revenue is not profit, so payback is calculated on contribution margin, assuming the extra revenue is spread evenly. The lower credit rate for variants is not factored in.

Content does not move a season. But within the same season, the optimized listing sells several times as much.

Putting it in context

• Short period: We measured the first 14 days after go-live. The data say nothing yet about how long the effect lasts.
• Small starting base: In the comparison categories, sales before the optimization were modest, and multipliers look large on small bases. That is why we also show how broad the effect is (83% against 44%).
• Robustness: Looking only at products with full data coverage (at least 10 of 14 days in both periods), the picture holds: units ×4.2 against ×1.6.
• Whole range: Across all 66 optimized products the figures point the same way (sales +21% against +6%, units +48% against −7%). Because the seasonal mix differs, they are not the basis of the core finding. Optimized garden furniture lost ground at the end of its season despite the optimization.
• What the figures leave out: Advertising costs, margins and returns. Prices and advertising stayed practically unchanged over the period; even so, 14 days cannot prove a single cause.