Why does marketing mix modelling not work on a 500,000 euro budget?

The budget is not the problem, the number of observations is. Weekly data over two years gives you about a hundred data points. With eight channels plus adstock, saturation and coefficient parameters per channel you are already past thirty parameters. Below five observations per parameter the model is not identifiable and produces estimates driven entirely by the priors.

Why is the language border not a valid test split in Belgium?

Flanders and Wallonia differ structurally in media consumption, competitive pressure, purchasing power distribution and seasonality. The gap you measure between them is therefore not attributable to your campaign. A valid geo experiment works with matched regions inside a single language region.

How long does a holdout test need to run?

The number of measurement weeks follows from W = phi times the square of the sum of z-alpha and z-beta, divided by the product of weekly conversions, the holdout share times its complement, and the square of the natural logarithm of one plus the lift. At a thousand conversions per week, a ten percent holdout and a five percent lift to detect, the best case is thirty-seven weeks.

What is a proof agreement?

One page that marketing and finance sign together before the budget year starts. It fixes what gets measured, at what confidence, what explicitly does not get measured, which decision rests on which level of evidence, and how often it is reviewed. It moves the burden of proof discussion forward, to the point where it can still be negotiated.

Marketing effectiveness

What can you actually prove in a quarter?

Your CFO wants evidence. The measurement chain is broken and it is not coming back. PROOF calculates which form of evidence is achievable at your volume, how long a test would need to run, and which effect you can demonstrate at all within your decision cadence.

You do not end with a score. You end with an agreement you can put on the table.

Start the diagnostic How we calculate

Three levels of evidence

ObservedYou watch something happen and lay it next to your spend. Costs almost nothing, proves almost nothing about cause.

ExperimentedYou deliberately withhold something and compare. Costs revenue and time, and proves causality.

ModelledYou build a relationship across all channels. Costs the most, and is only reliable when calibrated on level 2.

Nine questions

Tell us how your business works.

Everything stays in your browser. Nothing is sent until you submit it yourself.

Volume and model

How many conversions do you get per week?

Take the conversion you steer on: purchases, enquiries, sign-ups. A rough estimate is fine.

conversions per week

Which business model fits best?

Determines how much your weekly figures swing independently of media.

How long from first contact to conversion?

Determines how many maturation weeks come on top of your measurement period.

Media and history

What is your annual media budget?

Media buying only, excluding production and agency fees.

How many channels are you active on?

Count every channel you steer and budget separately.

How much usable weekly spend history do you have?

Usable means: per week, per channel, without gaps.

Decision and evidence

How finely can you steer your media geographically?

In Belgium this is often the decisive constraint.

How often do you have to account or adjust?

This is the window within which an answer is still usable.

Who is asking for the evidence?

Sets the tone of your proof agreement, not the calculation.

The reversal

variance factor for your model

weeks of maturation from your sales cycle

conversions per week

Module 1

What is achievable for you?

Five methods, tested against your volume, cycle, history and geographic steerability. Whatever shows up as not achievable belongs in your proof agreement, among the things you explicitly do not measure.

Module 2

How long does your test need to run?

Move the two dials you actually control: how much audience you withhold, and how small an effect you still want to be able to see.

Holdout10%

The share of your audience that gets no media. Larger is faster, and costs more revenue.

Effect you want to see5%

The smaller the effect, the longer you have to measure. This is the dial most plans break on.

Total runtime in weeks, maturation included. Green fits your quarter, orange fits your half year, red fits nothing.
Effect \ Holdout	5%	10%	20%	35%	50%

Module 3

What does being able to prove cost?

Evidence is not free. The largest item is not an invoice but unrealised revenue: the share of media pressure you deliberately withhold.

PAR is the scope unit used by IKAg: one PAR stands for roughly two agency working days in base year 2026. We deliberately express effort in PAR rather than in rates, so that you can compare setups without the conversation turning immediately to price.

Module 4

The proof agreement.

One page, signed before the budget year starts. This is the only moment at which the burden of proof can still be negotiated.

Method

How we calculate.

The audience is split into a test group and a control group. We treat conversions as counts. The logarithm of the ratio between the two groups has approximately a variance inversely proportional to the number of conversions, the number of weeks, and the product of the holdout share with its complement.

From that follows the number of measurement weeks you need to separate a given effect from chance. And in reverse, which is often more useful: for a given number of weeks it gives you the smallest effect you can still demonstrate.

We do not calculate with pure Poisson noise. Real weekly figures swing more than that, driven by season, promotions, stock and competitive pressure. So we multiply the variance by a factor that depends on your business model: the further the conversion sits from media exposure, the higher that factor.

On top of the measurement weeks comes a maturation period. With a long sales cycle, todayu2019s conversions belong to media from months ago, and that lag has to fit inside the window.

All calculations happen in your browser. The calculation core is the same one behind the other public instruments from IKAg, and it is verified by its own test suite.

How we calculate

What this instrument does not do

No scoreA mark out of a hundred would suggest measurability is a report card. It is a choice with a price.

No ratesWe express agency effort in PAR, not in euros per day. That is a deliberate limitation.

No model of your channelsWe say what is achievable, not what works. The second one needs your data.