CALIBRA™
Volume III

Media.

Where the advertising money should go, and how you will know whether the answer was right.

The problem

Media budgets are inherited, not decided.

Most media allocations are last year's allocation with an adjustment. The split between channels was set by someone who has left, defended by whoever owns each channel, and never read as a single question. Meanwhile the measurement available inside the large platforms credits them with results they did not necessarily produce.

The result is a budget that looks deliberate and is not.

What the volume does

Seven channel classes, read as one decision.

Video. Display and programmatic. Paid social. Paid search. Audio. Out-of-home and contextual. Direct response and owned. Every channel a business can buy falls into one of these seven, which is what makes them comparable.

Each is read through the four lenses: does it pay, does it build recognition, does it open future options, and can this organisation actually run it. The reading then resolves to one of seven instructions.

The seven media instructions. Increase. Hold. Reallocate. Pause. Exit. Trial. Test. Each carries a timeline, a named owner inside the business, and a date when it will be checked.

How a reading works

How a media reading works.

You provide what already exists: the media plan, the budget, whatever channel results you hold, brand tracking if you run it, and time with the people who own the plan. The first job is a declaration: which audience the spend is aimed at, what behaviour it is meant to shift, and by what mechanism. Awareness is not an answer, and the reading does not start until there is one.

Every channel then gets read through four lenses: what it earns, what it builds in the buyer's memory, what future options it creates, and whether your business can actually run it well. Favourite channels do not skip the awkward lenses, and unfashionable ones do not skip the flattering ones. The four readings combine into one call per channel, weighted by evidence rather than seniority.

What arrives is not a report. Each channel carries one instruction from a set of seven, a stated level of confidence in that instruction, a named owner inside your business, and the date the call gets checked. Where the evidence is thin, the instruction says so, and says what would firm it up.

Six worked situations

What a reading actually produces.

Illustrative composites, drawn from the pattern of engagements rather than from any identifiable client. Figures are indicative.

Use case 1

The brand campaign that was working, in the wrong place

Reallocate
Situation
A national retailer, roughly $180 million turnover, running 70 per cent of its media budget in television because that is where it had always been.
They asked
Is television still worth it?
The reading found
Television was performing. The problem was next door: paid search was absorbing budget to capture demand television had already created, and being credited for it.
Instruction
Move 18 per cent of paid search budget into video and audio. Total spend unchanged.
Why it matters
Same money, same brand, a different split. The question the client asked was the wrong question, which is a common finding.
Use case 2

The channel nobody wanted to cancel

Exit
Situation
A financial services group with a long-standing out-of-home programme, renewed annually for eleven years.
They asked
How do we improve out-of-home performance?
The reading found
The programme had never been read against the other six channel classes. It survived because it had a champion and a rolling contract, not because it earned its place.
Instruction
Exit out-of-home at contract end. Redeploy across video and owned channels.
Why it matters
The reading did not say out-of-home does not work. It said it does not work here, for this business, against these alternatives.
Use case 3

The audio question nobody could answer

Trial
Situation
A subscription health business, strong in paid social, considering podcast and streaming audio.
They asked
Should we move into audio?
The reading found
The evidence was genuinely thin. The method declined to issue a confident instruction and said so rather than guessing.
Instruction
A time-bounded trial with named exit conditions and a stated measurement approach.
Why it matters
The most useful output was the admission of uncertainty. A framework that is confident about everything is telling you nothing.
Use case 4

The underfunded channel hiding in plain sight

Increase
Situation
A business-to-business software company spending heavily on demand capture and almost nothing on being known.
They asked
Why is customer acquisition cost rising every quarter?
The reading found
Acquisition cost was rising because the company was buying attention it had never earned. Nothing was building recognition ahead of the purchase.
Instruction
Increase video and contextual spend materially, funded from search.
Why it matters
Rising acquisition cost is usually a symptom. Treating it as the disease makes it worse.
Use case 5

The campaign that could not be measured

Pause
Situation
A consumer goods business running a large programmatic display programme through a walled-garden platform.
They asked
Which creative is working hardest?
The reading found
The attribution available could not separate the display programme's effect from everything else. The reading could not be made honestly at the resolution required.
Instruction
Pause the programme for one measurement cycle to establish a clean baseline.
Why it matters
Sometimes the instruction is to stop long enough to find out. This is unpopular and usually correct.
Use case 6

The channel that took credit for customers already on their way

Test
Situation
A homewares retailer, roughly $120 million turnover, put close to a third of its media budget into search advertising against its own brand name. Every quarterly report showed it as the best-returning channel in the plan.
They asked
Whether to increase the search budget, since it was the strongest performer on paper.
The reading found
The channel's numbers rested on last-click credit, where the final ad a customer clicks is credited with the whole sale. Most of those customers had typed the brand's own name into the search bar. They were already on their way. The reading could not separate sales the channel caused from sales it merely collected, and neither could anyone in the building.
Instruction
Test. Switch off brand-name search in a matched set of regions for one quarter, hold it steady everywhere else, and compare the sales lines. Owner named, check date set before the next planning round.
Why it matters
The best-looking number in a media plan is usually the least examined one. A channel that marks its own homework deserves the same scrutiny as a channel that cannot be measured at all.
For the academics

Marketing-mix modelling yields elasticities. Multi-touch attribution yields a narrative that the platform selling the media is delighted to supply. Volume III composes both into a single interval-scored instruction and publishes it before the money moves, which is the one step neither tradition has shown much appetite for. Attribution has spent fifteen years getting better at explaining the past to people who needed a decision about next quarter.

And commercially: you get an instruction, not another dashboard.

The check

Every instruction has a date.

Twelve months after a reading, the instruction is compared against what happened. The result is recorded whether or not it is flattering. For media the method is built to a design target of 0.79, drawn from a development cohort of thirty readings. That figure is a target, not a result, and it will be replaced by the observed number once the register is running.

Start a conversation

One sentence is enough.

A media reading settles one question: where next year's budget should sit, channel by channel, and how sure anyone should be about each move. Every channel leaves with one instruction, a stated level of confidence, a named owner and a check date. It starts small: you describe the decision in front of you, and Rob reads whether the method fits it. Send one sentence: what is the decision you are trying to make?