Read a blended report honestly

A weighted average across every channel in scope, weighted by spend, over whatever the mixture happened to be in that window.

That definition contains the entire problem. Move a third of the budget from an expensive channel to a cheap one and the blended cost per result improves without any channel improving. Move it back and the blend worsens without anything getting worse. In both cases the report is correct and the conclusion drawn from it is not.

Underneath every blend in Tempio is the per-channel breakdown, returned in the same answer. Reading the blend without reading what is under it is the mistake this guide exists to prevent.

Always first, and usually decisive.

Blended spend and results for the last thirty days, with the per-channel breakdown, and the same for the thirty days before.

Compare the spend shares across the two periods, not the results. If the shares moved materially, most of the change in the blend is composition, and there is nothing to explain at the channel level yet.

The version of this that catches people is a single dominant line. If one campaign or one channel carried most of the spend, the blend is that line's numbers wearing a disguise, and every conclusion about "the account" is a conclusion about one campaign. The cheap test is to look at what the blend becomes with the dominant line removed.

Spend adds up. Self-reported results do not, and treating them as though they do is how a blend becomes fiction.

Two platforms can each honestly claim the same sale, because each counts the conversions it can claim under its own window and its own definition of the event. The sum of what they claim can exceed what actually happened. A blended cost per result built from summed claims is therefore an optimistic number by construction, and the optimism grows with how much the channels overlap.

That does not make a blend useless. It makes it a consistent index rather than a ground truth: comparable with itself over time as long as nothing about the counting changed, and not comparable with your order data.

A scope covering one currency is exact and nothing is converted.

A scope covering several is converted through published rates into whichever currency carries the most spend, and the result is labelled an estimate carrying the date of the oldest rate it leaned on, so a stale leg cannot hide behind a fresh one. Where a rate is missing nothing is invented: the blended total is refused rather than estimated, and a list that cannot be ranked comparably is returned in its raw order with that fact attached — a raw order across currencies is not a ranking, and every surface showing one says so.

Comparing an estimate against an exact figure is not a comparison. Two windows compared against each other are pinned to the same currency, so at least a period-over-period change is never one currency divided by another. Currency and FX in ad reporting is the whole of that subject.

Three ways a window lies, all of them cheap to rule out.

A partial day. A range ending today includes a day that is not over.

A restatement. Platforms revise the recent past; each sync re-pulls the days a platform is still willing to restate, and once a week the full ninety days are re-read. A connection that rewrote a lot of stored history is marked restated.

A window chosen after looking. The most common and the least discussed. Fix the window in advance or the blend becomes a search for the period that supports the conclusion.

One more reading rule specific to blends: a ratio whose denominator is zero comes back empty rather than as nought, and the result says explicitly that some ratios are undefined. An empty cost per result inside a blend is not a free result, and averaging it in as a zero would drag the whole figure down for no reason.

Is the whole operation above the line. Blended return or blended cost per result against your break-even, computed from your own margin rather than a benchmark — break-even ROAS and break-even CPA take your figures and no account. This is a portfolio question and a blend is a portfolio measure, so the fit is exact.

Is the total trending. Over enough periods, with the mixture noted, a blend is a reasonable index of whether the operation as a whole is getting better or worse. The mixture note is not optional.

What a blend cannot answer is which channel to fund next. That question needs the parts, their own trends, and a view on how much each one is claiming that the others could also claim.

The arrangement the reading side sits inside, 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.

Total revenue over total advertising spend, taken from your own accounting rather than from the platforms, is the one figure no platform can inflate. It is blunt, it lags, and it does not attribute anything — and it is the only number in this guide that nobody is grading their own homework on. The MER calculator computes it from your own figures, and keeping it beside the platform blend is the cheapest sanity check available.

Build a weekly performance review is the rhythm that uses both, and Why platform numbers disagree is why the two will never meet.

The roundings that happen before a blend is even assembled are in The numbers your dashboard rounds off.