In 1885, Hermann Ebbinghaus published the results of an experiment he ran on the only subject who was always available: himself. He built about 2,300 nonsense syllables, strung them into lists of thirteen, and memorized each list until he could recite it twice without a mistake. Then he came back after a measured gap - twenty minutes, an hour, about nine hours, a day, two days, six days, a month - and learned the same list again, timing both runs. The difference he called savings: the share of the original work he no longer had to redo.
The table he printed is 141 years old, and I think it is still the cheapest marketing consultation ever published. Twenty minutes after learning, 58.2 percent of the work was saved. An hour later, 44.2 percent. A day, 33.7. Six days, 25.4. Thirty-one days, 21.1. Read the drops, not the totals. The forty minutes between the first two measurements cost him 14 points of savings; the twenty-five days between day six and day thirty-one cost 4.3. Ebbinghaus opened the discussion of his own numbers by acknowledging that forgetting "would be very rapid at the beginning of the process and very slow at the end." Everyone quotes the rapid half. The slow half is this whole article. A single pass through anything sheds most of itself fast, and whatever survives the shedding is strangely durable.
The durability has been checked, hard. In 2015, Jaap Murre and Joeri Dros at the University of Amsterdam rebuilt the experiment - one subject, seventy hours of syllable time - and concluded that the curve "has indeed been replicated." It is not the quirk of one obsessive German. And the flat end holds across spans Ebbinghaus never attempted. In 1984, Harry Bahrick reported on 733 people tested across fifty years on the Spanish they learned in school: retention fell for the first three to six years, then "remains unchanged for periods of up to 30 years." Most of his subjects had barely touched the language since; he named the surviving residue the permastore. What gets through the early crash costs almost nothing to keep.
The most expensive hour
Every promotion plan I have ever been shown is built as if the curve ran the other way. The whole media budget lands on a single day, and that day sits inside the steepest window on the table - the same crowd, sometimes the same hour - where each extra sighting arrives while the memory is still loaded and mostly buys what has not left yet. Then the spend stops, the crash runs unopposed, and whatever the review measures weeks later is the 21.1 percent the table said would remain. The single day did what single days do: it paid full price for the fastest-decaying attention there is.
The alternative is not more money. It is shaped money. In 2006, Nicholas Cepeda and four co-authors published a meta-analysis in Psychological Bulletin - 317 experiments across 184 articles - and the practical finding is a scheduling rule: the gap that best preserves a memory grows with how long you need it to last. Every study they examined with a retention interval beyond a month showed a benefit from spreading the learning across weeks rather than a single day, and the average benefit over the massed version ran 15 percent. For anything you want remembered at the end of a month - a purchase cycle, a subscription decision, a name - the literature's answer is dull on purpose: show up across the month.
Stage lights hit the crowd from one direction, for as long as the show runs. What each person carries out is the part that survives the walk home.
The calendar does the work
My day job is Bid Spotlight, a leaderboard where the ranks go to the highest bids. The board resets on the first of every month with the minimum bid back to a dollar, which means a placement is not a day of noise. It is a position held across a month of separate days - sightings spread across weeks, the exact shape the meta-analysis favors at exactly this horizon - and the top ranks go out again in the daily social edition, one-day gaps, the shortest spacing that still beats same-day massing. The curve is also why holding the slot beats storming it. Each new sighting lands on a memory that already survived the crash, where a refresh is cheap, and resets the whole curve from a higher start.
The buyer's version of the lesson costs nothing to run. Decide how long the memory needs to last, then buy that shape. One day of spend, then silence, is the massed condition with a media plan attached. A modest position held for weeks beats a spectacular one held for an afternoon, for the same reason Ebbinghaus lost more in his first hour than in his final twenty-five days.
Respect the curve: take a slot at bidspotlight.com and let the month do the remembering.