What the account looked like when we took it
On paper this was a strong business. It was a top-ten brand in its category on Amazon UK, with 3.5% of category revenue, and it took more for every unit sold than any other brand in the top ten: 1.3 to 2.6 times the price of brands that outsold it several times over. Shoppers who reached the listing bought: paid traffic converted at 8.4%, and Sponsored Products returned £3.22 for every £1 spent.
The problem was that almost nobody reached it. Across 555 category search terms, carrying 288,083 searches a month, the brand held a top-ten organic position on seven. Six of those seven were its own brand name. A shopper who already knew the brand could find it. A shopper who did not, could not.
That is the most common shape we see in a premium account, and it is the one a generalist audit misses. The usual checklist flags the listing, the images and the cost of sale, and here all three were fine. What was wrong was a visibility problem sitting behind good numbers, which is why an account audit has to look at where the brand ranks for terms it does not own, not only at how the traffic it already gets performs.
Three things the previous setup had not acted on
The pack sizes were priced backwards. On the three product ranges that carried the account, the larger pack cost more per serving than the smaller one: 4%, 7% and 11% more. Those three ranges took £24,178 of the £27,757 that reached listing level in the 30 days we measured. A shopper comparing pack sizes was being told not to buy the bigger one.
The reviews were split. The brand sold eight product families with 441 reviews between them, and duplicate listings for two of its best sellers split those reviews further. A shopper saw a thinner review count on each page than the business had earned.
The budget sat in the wrong ad type. Sponsored Brands was taking 29% of the advertising budget and returning £1.40 per £1, converting at 5.1%. Sponsored Products returned £3.22 and converted at 8.4%. Moving that budget was worth about £2,300 a month of working media at no extra cost.
The first two are the client's to fix, not ours. We cannot reprice their packs or merge their listings. That matters to how the plan is built, and it is the part most agency case studies leave out: the plan names which gates the client owns, and the steps that depend on them wait until they are cleared.
Why the order matters
The plan buys ranking first and spends less later. For the first two months it pays for position on eight search terms, carrying 32,473 searches a month, at a planned total advertising cost of sale of 22.4%. That is deliberately expensive. From month 3 it holds page one, and from month 4 advertising steps back as organic rank carries more of the sales. The modelled cost of sale falls to 10.1% by the end of year one.
The reason for that order is the pricing. A brand earning that much per unit needs far fewer sales than the brands around it to reach the same revenue. So every rung above it is cheaper for this brand to climb than for the brand standing on it. Buying the position early and trading out of it for ten months is cheaper, per pound earned, than a slow ramp that reaches the same place later. We modelled both paths, and we chose the push. Why a growth figure means nothing without an efficiency figure beside it is the test the whole plan is written against.
Programmatic advertising waits until month 4 at the earliest. It will run on remarketing only, with a holdout region, because retargeting needs an audience that the search push has not built yet. That is the argument in our post on when Amazon DSP is worth it and when it is not.
What has moved, and what has not
Ranking has moved faster than planned. In the week to 13 September, 11 of the 13 target search terms improved. The biggest term in the plan carries 35,216 searches a month. It moved from 84th to 10th in unpaid results in a single week, and the plan did not target it until month 15. The brand now holds first position in both paid and unpaid results for its own name.
Sales have not kept pace. At 18 September the month was running about 12% behind the plan line. Three things were holding it back at that point, and one has since been fixed.
First, the account cannot spend the budget it was given. After 17 days of September, advertising spend was 62% of the planned pace. The ranking bids have reached the ceiling set in the account's guardrails, and that ceiling only rises by a deliberate decision, never by default. The cost of sale looks better than planned, 16.2% against 22.4%. On this account that is bad news, not good: it means we are buying less ranking than the plan needs. That is also why lowering the cost of sale is the wrong goal for some accounts.
Second, one pack size of the best-selling range lost its Add to Basket button for more than ten days. The cause we identified was a 25%-off promotion at another retailer on the same product. Amazon withdraws the button when it finds a product cheaper elsewhere, and no bid can fix that. The button came back on 18 September.
Third, stock on some lines has run close to empty. The menopause-range advertising is held back until cover recovers, because advertising a product about to sell out spends money to create a stock-out.
What the next ninety days are for
October is month 3, where the plan stops buying position and starts holding it, at 966 units a month. That is 52% above where the account traded when we started. It is also the first quarterly review of the plan with the client. That review will decide three things: whether the ranking gains hold now that spend is below plan, whether the bid ceiling rises, and when the pack pricing and the listing merges happen.
The ranking has already outrun the plan. The sales have not caught up yet, and it is too early to say whether they will. If you run a premium brand with good conversion and poor reach, a free account audit will show you where you rank for the terms you do not own. Our pricing is published if you want to model the cost first.