Meridian: Impact Measurement - YourValidator: Scenario Planning

The distinction between impact measurement and deterministic planning

Comparison · Planning and measuring marketing budgets

Google Meridian is a scientific tool: it estimates from your historical data what each channel actually contributed, and states how certain that estimate is. YourValidator is a planning tool: it works out what would have to hold under your assumptions. In everything that concerns measurement, Meridian is superior. This page shows exactly where – and when a planner is still needed.

Measurement · the scientist

Google Meridian

Open-source marketing mix model from Google. Bayesian estimation of causal channel effects from your own historical data, optionally at the level of individual regions, with uncertainty estimates, calibration through experiments, plus budget optimisation and scenario planning.

For companies with several years of data history, relevant media spend across several channels and a statistically skilled team.

Planning · the planner

YourValidator

Deterministic budget and scenario planning in the browser. Works through your assumptions on reach, rates, potential and budget – without data history, without programming, in minutes.

For SMBs, agencies and consultants who want to make assumptions visible before budget is committed – or before there is enough data for a model.

Clear positioningYourValidator measures nothing and proves no effect. It shows which assumptions lead to which result. Anyone who wants to know what a channel actually delivered needs a measurement method such as Meridian. YourValidator is an independent product and has no affiliation with Google.

Two roles, two questions

The scientist asks: what worked? They need data from the past, test hypotheses statistically and state for every finding how robust it is. That is Meridian's job.

The planner asks: what would have to hold for this plan to work? They work with assumptions – often before there is enough data – and make visible what a result depends on. That is YourValidator's job.

Both questions are legitimate. But only one of them delivers proof.

Head to head: Google Meridian and YourValidator

Criterion	Google Meridian	YourValidator	Ahead
Question answered	What did each channel actually contribute?	What follows from my assumptions?	Meridian
Method	Bayesian model estimated from your own data; the result is a probability distribution of the parameters.	Deterministic calculation from entered assumptions.	Meridian
Causal statement	Aims to estimate the true causal effect.	None. Results are consequences of the assumptions.	Meridian
Uncertainty	Credible intervals for ROI and contributions.	No statistical uncertainty; only scenarios the user varies.	Meridian
Calibration through experiments	Results of incrementality tests and geo experiments can be included as priors.	Not provided; test results are entered by hand as assumptions.	Meridian
Regional depth	Geo models use data per region and thereby gain statistical information.	A regional factor as an assumption, no geo estimation.	Meridian
Reach and frequency	Can use reach and frequency data as input.	Reach as an assumption per channel; frequency is not modelled.	Meridian
Saturation and carryover	Saturation via a Hill function, carryover via adstock (geometric or binomial) – both estimated from data.	Saturation via the same function family (Hill), but with assumptions instead of estimation; carryover between channels (halo) and for actions as an assumption.	Meridian
Budget optimisation	Optimisation and scenario planning based on the estimated curves.	Scenarios, target back-calculation, marginal cost and break-even per channel – based on assumptions; no optimiser.	Meridian
Data requirement	Several years of weekly data across channels, control variables and the target metric.	No data history needed; your own figures replace starting values once available.	YourValidator
Time to first result	Weeks to months for data preparation, modelling and validation.	Minutes.	YourValidator
Skills required	Statistics and programming; the model has to be maintained and validated.	Marketing and business understanding; operated in the browser.	YourValidator
Cost	Licence-free. The effort lies in the team, data preparation and operation.	Calculating free; Pro from $17/month.	depends

ResultMeridian is ahead in every criterion that concerns measurement and proof. YourValidator is ahead only where getting started is concerned: data, time and skills. That is not a draw but a division of labour.

What Meridian can do – and YourValidator cannot

Prove impact. Meridian estimates the contribution each channel made to the result. YourValidator fundamentally cannot, because it does not analyse data.
Quantify uncertainty. Meridian does not just say "ROI 2.4", but also in which range the value probably lies. YourValidator calculates with point values.
Derive curves from reality. In Meridian, saturation and carryover are estimated from the actual history. In YourValidator they are the user's assumptions.
Include experiments. Results from incrementality and geo tests flow into Meridian as prior knowledge and sharpen the estimate.
Separate confounders. Control variables such as brand search interest help separate marketing effects from other influences.

A shared language, different sources of the numbers

Both tools describe diminishing returns with a Hill function: as the budget rises, each additional dollar brings less additional effect. The difference lies in where the parameters come from.

In Meridian, half-saturation point, slope and carryover are estimated from your data – with uncertainty. In YourValidator you enter what you assume: a potential per channel, the budget your values apply to and, optionally, what a doubled budget would bring. The curve is calculated from that.

That is why YourValidator's results are exactly as good as the assumptions entered. Meridian's are as good as the data, the model choice and the validation.

What YourValidator deliberately does not modelThe carryover of an ongoing channel's own budget is not calculated separately; with a constant monthly budget it is contained in the values entered. Frequency effects, control variables and statistical uncertainty are not included. These are exactly the points where a measurement model such as Meridian comes in.

How planner and scientist work together

Before the model: there is not yet enough data history.

Make assumptions visible and plan budgets – in YourValidator

While data grows: plans are implemented, results collected.

Compare plan vs. actual and adjust assumptions (Pro)

Deciding on an MMM: is the effort for team and data preparation worth it?

Work through the business case as a scenario

With a model: Meridian delivers estimated contributions and response curves.

Meridian is the basis; its results can serve as assumptions for planning scenarios

Transfer with careResults from Meridian can be carried over into YourValidator in spirit, for example as potential or as the shape of the curve. A one-to-one transfer of the model parameters is not possible, because the two tools represent budget, time periods and channels differently. For budget decisions based on measurement, Meridian offers its own optimisation and scenario planning.

When Meridian is the right choice

You need robust evidence of what each channel contributed.
Several years of clean weekly data on media, target metric and control variables are available, ideally per region.
Media spend is large enough and varies enough across channels and weeks for a model to separate effects.
A team with statistics and programming skills can build, validate and regularly update the model.
Experiments exist or can be run to calibrate the model.

How much data, for orientationGoogle's documentation itself calls its guidance "rough and directional". For a national model with 12 channels, 6 control variables and 8 knots, two years of weekly data give about 4 data points per parameter – too few for a reliable estimate. With three years and fewer parameters, directional insights may emerge; geo models contain more statistical information (Meridian: Amount of data needed).

When a planning tool is enough

The data history is too short, too patchy or not harmonised for a statistical model.
Media spend is too small or too evenly distributed for an MMM.
The task is a budget decision in days, not proof in months.
Stakeholders should be able to see and change assumptions together, without programming skills.
New channels or one-off actions such as a trade fair need to be planned, for which there is no data yet.

Summary

Google Meridian is the scientist: it measures from data what worked, and says how certain that is. In this role it is clearly superior to a planning tool.

YourValidator is the planner: it works through assumptions deterministically and transparently – where there is not yet data for a model, or where people without a statistics team need to understand and approve a budget. Those who can measure should measure. Those who cannot yet measure should at least make their assumptions visible.

YourValidator · browser-based planning calculator for marketing budgets. A planning layer, not a measurement tool and not a substitute for a marketing mix model built on your own data history. Google and Meridian are trademarks of Google LLC; YourValidator has no affiliation with Google.