The management pack, on your rhythm

Cycle 20 took 751 orders in its first seven days, 21% ahead of the same days last year (622). The tracker projects 1,503 (range 1,278–1,727), in line with its smoothed average of 1,465. The year to date is 25,932, +19%. Cycle 19 closed at 1,469 after six orders were cancelled with nothing bought, 5% ahead of last year’s 1,404. No decline at the level of the whole business: the tracker reads Growing.

Against the prior six cycles no campaign type moved beyond its usual swing. Against the same three cycles last year, rebuilt search (+56%), the search not yet rebuilt (+26%), all paid (+42%) and cost per order (+41%) are dearer beyond it. So far in cycle 20, rebuilt search is heading dearer ($333 a credited sale against $248) as group A’s spend rose after the repricing, and product shopping too ($187 against $132). Rebuilt search is the line to fix; the repricing is the fix under test.

Red is beyond that line’s usual swing: 1.5 times the median swing in its own history, or 25% either way, whichever is smaller. Everything else is inside its swing: treat it as noise. Cycle 20 is projected from seven days; tested over 35 closed cycles, the first-week cost per sale landed on average within 1% of the close, with no lean either way.

Cycle 19: $251K, 10% below last year’s $277K (last year’s strongest cycle), on 5% more orders. Ad spend of $180K against $116K absorbed the growth. Cycle 20 projects $268K against $213K last year, from seven days, against a dear cycle last year.

A caveat we state rather than hide: labour is charged as a fixed share of every order, as if it were fully variable. Much of it is fixed in the short run, so this line penalises a strong cycle and flatters a weak one. Read the trend, not the single cycle.

Over the last three cycles, profit after each product’s own ad spend is level with the prior six and 9% below last year. Products Delta, Gamma and Epsilon give profit back against last year. Cycle 20 is heading level ($253K projected against $257K a cycle): Product Alpha carries it ($74K against $56K), Product Delta runs below its average ($62K against $81K), and Product Iota projects a loss after its own spend.

Cash contribution = orders × each product’s contribution per order before labour (net income after credit notes, less parts, stock, indirect costs, payment fees and shipping), minus the product’s own search and shopping spend. Profit = the same after labour. Spend no product owns (brand search, general shopping, other paid) sits in gross contribution above. Coloured only beyond each product’s usual swing; the total row has no band.

Rows add up: A + B = the live groups; live groups + trailing credit = all own search. The trailing credit is sales credited since the switch to clicks the old campaigns bought before it, valued at each product’s break-even, so the whole search book is read before the A / B split.

Since the rebuild, the whole search book has 935 credited sales against the 772 the old campaigns would have given (+22%), at $274 a sale against $224: $51K of profit against break-even, where the old campaigns would have made $80K. Group B is the only group losing money ($435 a sale, −$41K against break-even); group A makes money ($303 a sale), but less than before.

The driver: the money moved, but the sales cannot be read yet. Group A’s higher target and bid cap, and group B’s lower target, moved the spend as intended. Cycle 20’s cost per sale is not a verdict; the first read is mid-cycle 21. Product Iota’s slower week is not the repricing; on Product Eta’s group B it is not ruled out.

Brand search is read on impression share, closing clicks and its cost trend, not against a break-even. Credited sales are after the brand discount: the first brand click in a sale’s journey earns 30% of the credit a product-search click would, so the credited figure is deliberately low.

Orders are counted once per customer checkout, from the order system, and reconciled to the accounts. Credited sales use our own attribution rule on raw click journeys (first and last click weighted, brand clicks discounted, a 90-day click window); the ad platforms’ own conversion counts are not used. Cost per sale = spend ÷ credited sales over the window, read on the tracker’s shape: the last three cycles against the six before and against last year, coloured only beyond the measure’s own usual swing. Break-even = full cost per order including labour and shipping. Gross contribution = orders × each product’s contribution per order before advertising, minus all ad spend, two-product orders counted once.