How well the engine matches the research

Media raised 10% every quarter; the change in first-quarter units, in all seven categories over 5 worlds. Countermove is at 83% of the published average. Seven in ten purchase occasions come from habitual buyers who do not react to media in the quarter, and brand media mostly works through memory, which builds slowly. The game is deliberately cautious here so that media never looks like a quick fix; later studies of TV advertising in packaged goods also find short-run elasticities well below 0.12 (Shapiro, Hitsch & Tuchman 2021, Econometrica).

Same test, read on units a year later with the plan still running. Once memory has built, Countermove lands a little above the published long-run average. Going dark costs share for years, as Ehrenberg-Bass and Kantar find for brands that stop advertising.

One quarter of doubled media on the balanced mix, then back to normal; the extra units the next quarter as a share of the extra units in the burst quarter. Countermove carries more than the average: purchase habit carries any extra sales forward, not only advertising's, and brand media keeps feeding memory, which Assmus et al.'s single-equation models count only in part. A year after the burst, 56% of the first quarter's lift remains for brand media against 49% for activation, which is the slow and fast pattern the research describes.

The same media budget all year, split from 20:80 to 90:10; two-year net income for each split. Countermove lands near the IPA average: profit peaks at 60:40, and share a year on peaks at 60:40. Activation in Countermove fades within the quarter and only reaches switchers, while brand media builds memory for years. 60:40 is an average across many brands, not the answer for any one of them: Binet & Field's later work (Effectiveness in Context, 2018) finds the best split ranges up to about 80% brand by category, position and horizon, and in the game the Advisor's optimiser works out each brand's own split.

The standard plan with the media budget held from 4% to 17% of net sales, gross profit after media summed over two and over three years; the A&P-to-sales ratio of the budget that earns most, in all seven categories over three worlds. Over two years, Countermove's most profitable media level is leaner than what the theory and the category's own margin suggest (soft drinks 8%, energy drinks 7%, snacks 14%, beauty & personal care 8%, household care 4%, coffee 12%, oral care 11%), and the Advisor's optimiser, which also sets price and deals and judges on two years from the quarter in hand, often recommends a leaner budget still. Over three years the most profitable level rises to 16%: brand memory keeps paying after a two-year window closes, so most of the gap is the horizon, not the money being wasted. Going dark now loses about what the research finds (Ehrenberg-Bass: sales 16% down after a year without mass media, 25% after two; Countermove: 17% and 27%), because brand memory settles on a floor plus the share of voice rather than falling towards nothing. The optimiser's leaner answer mostly comes from the price rise it picks alongside: a dearer brand converts fewer price-led shoppers, so each euro of media wins less. So read Countermove's media advice as "over two years, lean wins more often than the industry norm suggests"; over a longer horizon the norm holds.

Campaigns drawn by the default process: the spread of what a euro of brand media builds (1 s.d. of 0.43 on a log scale) against the spread of media money across the brands and budget options of three reference markets (0.64). Countermove gives the creative a smaller part than Nielsen on purpose: with the full weight, the luck of one campaign's draw would decide too many games. Process (budget, pre-testing, brand codes) shifts the odds; it never guarantees.

Average linkage across campaigns drawn by the default process. Below average, part of the shortfall builds memory for the most mentally available rival, usually the leader. The share is Countermove's assumption: sponsorship research finds people credit prominent brands they never saw (Johar & Pham, 1999), but no study gives a figure for ads; the debrief shows the money.

Loss per quarter on air after the first quarter for an average rational, balanced and emotional ad; weak ads wear faster and strong ones slower. Kantar finds no clear link between an ad's age and its performance on average, so the rates are kept low.

A new campaign in Q1 on each approach, same media money, 9 worlds; value share differences.

The same campaigns as above, read on the brand's sensitivity to price and deals at the end of year two.

A one-off shelf-price move (+3% and −5%), read at Q4, averaged over seven categories. Countermove's shelf price bites 70% as hard as the published mean. Much of the data behind that mean is weekly store data, where short price cuts supply most of the price variation, and those move volume more than a lasting shelf-price change. Countermove's deal response is below (−2.45). A lasting move reaches about the category's stated elasticity once habits adjust (the table below), and more in the most promoted categories (coffee, snacks), where frequent deals have trained shoppers to watch price; habitual buyers react to a price above what they are used to, and past about 10% above the market they react with the full elasticity, so taking price every quarter costs far more volume than taking it once.

Deal elasticity is measured the way store studies do it: the rate of sale in deal weeks against regular weeks, over the depth of the discount. Shelf elasticity is the first-quarter response to a shelf-price move.

Deep deals with stock loading in Q1 against light deals, every category and world. Shoppers who stocked up buy less the next quarter, and retailers who were loaded order far less. The bump splits three ways, as van Heerde, Gupta & Wittink (2003) describe: shoppers switching from rivals, loyal buyers buying ahead (all of which comes back as the dip, 35% of the bump here), and a little new category volume.

For each category, the three brands at the end of the year: purchase frequency must rank the same way as penetration, and penetration must vary far more than frequency (in Countermove, 3.1× against 1.4× on average).

Every fourth combination of the quick calls, held for a year, in four markets: 768 quarterly moves of the player's value share. The largest single move was 2.34 points; nine plans in ten moved share by under 2.2 points over the year.

Checked for every company in 3,528 simulated quarters: all seven categories in all seven markets, on a quiet plan and an aggressive one, in the three-brand world and in the four- and five-brand worlds of team play. Gross sales less trade spend equals net sales: largest gap under one part in a million.

Same quarters as above. Shares are measured after the post-deal dip, so a deal's hangover shows in share too.