Media planning guides

A plan in Tempio is a working document. Its name, brief, window, currency and items are all free to edit, because none of them is a change to an ad account. Nothing in a plan spends, nothing in a plan is queued, and nothing in a plan has been approved by anybody. That is what makes it the right room for the awkward conversation about where next quarter's money goes.

What a plan is not is a forecast, and the distinction is worth holding onto. The document states an intent; the reporting side states an outcome; holding one against the other is still a thing a person does, on a rhythm they set. A planning surface that quietly implied it was watching the accounts for drift would be selling a different product than the one that exists.

It is also not a queue. A plan sitting in the workspace has drafted nothing and committed nothing. The moment it turns into work waiting for a decision is a separate and deliberate act, and it is the last section of most of the guides below, because it is the only step in planning that has consequences.

Strip a media plan of its formatting and five decisions are left. Every guide in this category is an argument about one of them.

How much, across what stretch of calendar. A budget with no window attached is a number with no meaning; the same figure is generous over a quarter and thin over a fortnight. The window comes first because every other decision is measured inside it.

In what currency. A plan carries exactly one, and every budget inside it is a whole number of that currency's minor units. One currency per plan is the reason the total at the top of the page is a real number instead of a sum of unlike things.

Across which platforms, in what split. This is the decision people spend the most time on and have the least evidence for. The honest version starts from what the accounts have already done, treats the split as a hypothesis with a review date, and writes down what would have to be true for it to change.

In what order, and with what overlap. Flights inside the window can run sequentially, in parallel, or staggered so a second concept starts while the first is still learning. The ordering is a real decision with real costs, and it is the one a spreadsheet of monthly totals cannot express at all.

Against what definition of working. A plan that does not say what counts as a result is a plan nobody can grade. The definition belongs in the brief, in a sentence, before the first line item exists.

One field is worth being blunt about, because it is the one people get wrong on their first plan. An item's budget is the whole-flight total, not a daily rate. A plan is a statement about a flight, and the daily figure a platform wants is derived from the total when the plan is turned into drafts, over the inclusive length of the flight.

Two consequences follow and both of them bite. The first is that an item's flight length changes its daily budget without the budget field being touched: shorten a twenty-eight day line to twenty and its daily rate rises by nearly half. The second is that whole-flight totals are what make a plan's own total honest, because the number at the top of the page is only a sum of comparable things when every line is stated on the same basis in the same currency.

Money is an integer in minor units throughout, which is the same convention the reading side uses. There is no rounding step hidden between what you planned and what gets drafted, and no rate applied anywhere, because no comparison in a plan crosses currencies.

Materialising a plan is the one place planning touches spend, and it is worth understanding before you rely on it. Every item becomes a create-campaign draft at the approval gate, grouped into one batch per connection, with the items keeping their board order inside each batch.

The step validates everything before it writes anything. Each item's flight has to sit inside the plan's window, each budget has to be a whole number of minor units at or above zero, each item has to resolve to an ad account, and that account has to bill in the plan's currency. Then the guardrails run, using the same functions the gate itself runs. A failure at any point drafts nothing at all, which is the only sane behaviour: a plan half-turned-into-drafts is worse than one that refused.

What comes out is a batch of drafts, not a running campaign. Creating something is an operation no rule may clear on its own, so a materialised plan always waits for a person, and every create lands paused on the platform, so approving the batch builds structure without starting delivery. Starting delivery is a second change through the same gate.

After that the plan freezes. Its name, brief, window, currency and items stop being editable, because the plan has become the record of what was drafted, and a record you can edit behind is not one.

The arrangement, stated once: your connections read your accounts, a change is written as a draft on the record with the state it would change from and the checks it passed, applying it is gated on approval — by you, or by a rule you wrote — and no platform write reaches a customer account until that platform has granted write access. Applying a cleared change to a platform is in private beta.

The difference between a plan written in a spreadsheet and one written here is that this one sits next to the numbers. Before the first line item exists you can ask what the last ninety days covered, which campaigns carried the spend, how concentrated that spend was, and what the mix of objectives looked like — in one exchange, across every connected account, in one place.

That matters most for the split. A platform-by-platform allocation argued from what the accounts have actually returned is a different document from one argued from a memory of last year, and it takes about a minute longer to produce.

They assume a workspace with at least one connection that has finished a sync, because a plan is easier to argue about with the account's own history in front of you. They assume nothing about how you drive it: the planner is the same surface whether you work through the pages, the in-app agent, or your own client over the programmatic surface, and the seven guides here name the tools rather than the buttons wherever the two differ.

They also assume you want the record. Ungated is not unrecorded: every write to a plan lands an audit row with the actor that made it, precisely because a plan is free to edit, and the log is what tells you who shaped the quarter and when.

A plan is the document every other subject either feeds or consumes.

• Agent setup comes first in practice: the ten minutes that turn planning from recollection into reading.
• Budget and bidding picks up where the plan stops — the plan states a whole-flight total, and everything about how that total is spent per day is the next subject.
• Governance and approvals owns what happens to the batch a materialised plan produces, and why a create always waits for a person.
• Measurement and metrics is where the split should have been argued from, and the reason a plan's numbers are worth checking before they are committed.
• Creative and testing is the half of a flight a plan does not size: the money is planned here and the variance is mostly there.