Why a cook-set family earned less on the same ad budget

What the data shows Fact

• The export repeats each parent's revenue on every child row.
• Summing that field gives USD 652,742, which counts each parent once per child.
• Summing child-level revenue gives USD 252,518. Every later figure uses this definition.

Reading Inference

Brand size is always read at child level, and parent figures are counted once per parent. The summed parent field overstates this line 2.6 times.

What the data shows Fact

• Cook sets and pots: USD 147,706 a month, 58% of the line.
• Camp tables: USD 78,085, 31%.
• Utensil kits: USD 26,727, 11%.

Reading Inference

The diagnosis starts where the revenue is. Cook sets and pots are the commercial center of this line, so they are examined first and the other families are not given equal weight.

What the data shows Fact

• A-102 is the largest child at USD 64,732 a month.
• A-118 and A-207 rank third and fourth.
• The three share one parent listing, which is the scope of this diagnosis.

Reading Inference

Diagnosing the children that carry the most revenue keeps every action close to a commercial result. The family is treated as one unit because its three variants share a listing and an advertising budget.

What the data shows Fact

• Conversion moved by 0.4 points or less on each variant across the three months.
• A-102 and A-207 have no product video. A-118 has one.
• The A+ content explains specifications but has no comparison module and no customer-use content.
• The listing holds a 4.6 rating from 2,140 ratings, unchanged over the period.

Reading Inference

The missing video and customer-use content are facts. That adding them would lift conversion is a hypothesis to test, not a result to promise. Because conversion did not fall, the listing does not explain the lost profit.

First constraint? No. It is an opportunity, tested after the constraint is fixed.

What the data shows Fact

• Family sessions: 37,160 in July and 36,804 in September, down 1%.
• A-102 lost 14% of its sessions. A-207 gained 10% and A-118 gained 9%.
• By September A-207 drew more sessions than A-102 and converted them at 9.4% against 13.2%.
• Search volume for the two main terms fell 3% over the period.

Reading Inference

Demand did not leave the family. The same number of shoppers arrived, but more of them landed on the variant that converts worst and earns least.

First constraint? No. Demand is stable.

What the data shows Fact

• A-102's top-of-search impression share on its two main terms fell from 31% to 12%.
• Its organic rank on those terms slipped from 3 and 4 to 7 and 9.
• A-207's advertised keywords more than doubled, from 190 to 412, while its organically ranking keywords went from 41 to 43.

Reading Inference

A wide gap between advertised and organic keywords does not by itself prove that advertising is ineffective. Here the search-term report settles the question: the new A-207 terms brought spend without orders, which the next step shows.

What the data shows Fact

• Family ad spend: USD 17,400 in July, 18,300 in August, 18,200 in September.
• A-102's share fell from 57% to 43%. A-207's rose from 21% to 34%.
• Broad match took 41% of A-207's spend in September, up from 18% in July.
• Six unrelated terms took USD 3,180 over two months and returned 29 orders.
• Sponsored Products carried 78% of family spend and Sponsored Brands 22%. No Sponsored Brands Video ran.

Reading Inference

The total did not change. The allocation did. A-207's broad match absorbed the budget, and A-102's proven terms lost the placement that fed them.

First constraint? Yes.

What the data shows Fact

• A deal ran on A-102 and A-118 during Prime Day week in July.
• No promotion ran in August.
• A 10% coupon ran on A-207 from 5 September to month end.

Reading Inference

The coupon stacked a discount on top of a variant that was already losing money after ads. It should move to a variant that can afford it.

First constraint? No. It made the constraint more expensive.

What the data shows Fact

• The category leader sells at USD 32.99 against USD 34.99.
• It holds 5,310 ratings against 2,140.
• It runs a product video, an A+ comparison module and Sponsored Brands Video. A-102 runs none of the three.

Reading Inference

First place is worth pursuing only while contribution stays acceptable. Rank is not a goal by itself. The content gap is the cheapest part of this list to close; the review gap closes only with time.

What the data shows Fact

• A-207's ACOS was 35% in September against a break-even of 29%.
• A-102 ran at 22% against a break-even of 57%, and A-118 at 20% against 94%.
• A-207's landed cost is 46% of its price, which leaves a 14% margin before ads.
• Landed cost did not change on any variant during the period.

Reading Inference

A-207 cannot carry its current advertising at any volume. Fixing the search terms brings it back toward break-even, but its landed cost still leaves little room, so it needs a decision of its own.

First constraint? No, but it sets the limit on what A-207 can spend.