What is Flight dates?

A campaign running from 1 October to 15 November covers 31 days in October and 15 in November, so days in flight is 46. A A$92,000 total gives a daily rate of 92,000 ÷ 46 = A$2,000. Compress the same total into 23 days and the daily rate doubles to A$4,000, which buys further down the inventory curve and usually raises the price per result.

Flight dates are the start and end of a campaign's run.

Stated that way they sound administrative. With a fixed total they are not, because the dates and the total together determine the daily spend rate, and the daily spend rate determines how much inventory the campaign has to buy each day.

Days in flight = end date − start date + 1

A campaign from 1 October to 15 November runs for 46 days. A A$92,000 total across those days is A$2,000 a day. The same total across 23 days is A$4,000 a day — the same money, twice the rate, and a materially different campaign.

The cheapest and most responsive inventory in any market is bought first.

A campaign spending A$2,000 a day competes for the portion of available inventory that amount can buy. One spending A$4,000 has to reach further: broader audiences, less premium placements, auctions it would otherwise decline, times of day it would otherwise skip.

The result is a higher cost per result, and it is not a failure of the campaign. It is what buying more of a finite thing per unit time costs.

The same effect runs in frequency. A compressed flight against a fixed audience accumulates repetition twice as fast, so a campaign that would have reached a comfortable frequency over six weeks reaches an uncomfortable one over three.

Both are predictable before the campaign runs, from the daily rate alone, and both are routinely discovered afterwards.

Start before the thing starts. A campaign on conversion-based bidding needs time to leave its learning period, during which delivery is uneven and cost is unrepresentative. A flight that begins on the first day of a promotion spends the promotion's best days learning.

End before the thing ends. Two reasons. An automated strategy with budget left and few days remaining will loosen what it pays to spend it, producing the campaign's most expensive results on its last day. And advertising an offer in its final hours reaches people who cannot act in time.

Make it long enough to settle. Flights of a few days rarely leave learning on conversion-based bidding. Where the window genuinely cannot be extended, optimising toward a shallower and more frequent event gives the model something to work with in the time available.

Check the implied daily rate against what the account normally spends. A rate several times the usual is a decision to pay more per result, which is sometimes correct — a launch that has to land on a date — and should be made knowingly.

Weekends and weekdays convert differently in most businesses, and a short flight that happens to contain a different mix from the period it is compared against will look better or worse for no reason.

Public holidays move demand and move competition, and they fall differently across markets, which catches multi-market campaigns.

Month and quarter boundaries matter where competitors have budgets that reset, because auction pressure changes when other advertisers' spending does.

Pay cycles matter in consumer categories more than most plans acknowledge.

None of these justifies elaborate scheduling. All of them justify looking at a calendar before choosing the dates, and noting anything that will make the flight's comparison period unrepresentative.

Do not change the total mid-flight if it can be avoided. A lifetime budget re-plans the remaining days against the new total, so raising it late accelerates spending into a short window and lowering it can halt delivery entirely.

Watch reach and frequency across the flight, not at the end. The end-of-campaign averages conceal whether the audience saturated in the first week.

Watch the last few days for the spend-down effect, and pace slightly conservatively if the strategy has shown it before.

Record the daily rate in the plan alongside the total, because the total is the number that gets agreed and the rate is the number that determines what it buys.

Before the dates are set, the useful question is whether the window is real.

A launch, a sale, an event, a season — these have genuine dates and the flight should match them. A campaign given a start and an end because a plan needed boxes filled has invented a constraint, and the constraint will show up as a higher cost per result than an always-on equivalent would have produced.

Where the window is real, the dates are a decision about how hard to buy. Where it is not, a longer flight at a lower rate is almost always the cheaper way to the same total.

A campaign that stops at the end of a flight and starts again for the next one is not resuming; it is beginning.

Automated strategies build their predictions from recent activity, and a campaign that has been dark for weeks restarts with a stale model and re-enters a learning period. On a short flight that period can consume most of the run, which means a seasonal campaign restarted four times a year pays for learning four times.

Two things reduce the cost. Keeping a low-weight version running between flights preserves the conversion signal so each burst starts from something rather than nothing. And extending the flight rather than repeating it — one longer period instead of two short ones — pays the fixed cost once.

Neither is available where the window is genuinely fixed by an external event, which is a reason to check whether it is.

A flight is worth documenting with three numbers rather than two, because the third is the one that explains the result.

The total, the dates, and the implied daily rate. The first two get agreed; the third determines what the money buys and is almost never written down.

Adding it takes a division and makes the comparison against the next flight meaningful: a campaign that performed worse at twice the daily rate performed exactly as it should have, and a plan that recorded only the total has no way to see it.