In this article we will cover the Calibration process, its importance and what to expect during the Calibration.
NCRMNTAL works with aggregated data, learning from patterns and relationships across your marketing activity rather than user-level tracking.
This makes it essential that we have all necessary data inputs in place - complete, consistent, and aligned with your business logic - so that the models can learn effectively.
At the same time, we must be mindful not to pollute the models by injecting incomplete or noisy data too early in the process.
Calibration is the foundation for making INCRMNTAL insights accurate and actionable.
It ensures that the platform interprets your marketing activity correctly and reflects your real business performance. Calibration secures data accuracy and reliability, leading to more confident and informed decisions.
Collaboration During Calibration
Calibration is not a “black box” process-it’s collaborative.
Clients can and should use the INCRMNTAL platform during calibration.
If something doesn’t make sense, we investigate together:
“Why do you think that doesn’t make sense? Do you have a data point or test we can look into?”
With your real-world examples, we can verify whether the model’s functions behave as expected and explain the logic behind what you see.
If there are no examples, we’ll explore your preferred markets and clarify the observed patterns based on the data dynamics there.
When Calibration Is Needed
Calibration is not always necessary.
Once your integration is completed and we see the main data points injected, we’ll review the initial inputs together during the first onboarding call.
The purpose of this session is to verify:
- That the integration is complete and all data fields are correctly ingested
- Whether model calibration is required based on the initial behavior of your data
Calibration Process (If needed)
If calibration is needed, we will work together to make sure that we map your business logic and external context to the model to ensure the most accurate insights.
During this stage, we’ll discuss several key topics that can influence performance and model interpretation. These are not automatically added as calibration factors - each point is reviewed together with your Marketing Science experts to determine whether an adjustment is necessary and meaningful.
Questions to guide this step:
- What is your current paid vs. organic ratio?
- Do you have full visibility into all paid sources (including affiliates, influencer traffic, etc.)?
- Have there been any in-app events, sales, or promotions that could have temporarily impacted performance metrics?
- Do you have any referral programs or partnerships driving additional traffic or conversions?
- Have there been any market-wide factors such as press mentions, platform algorithm updates, or seasonal shifts that could explain recent performance changes?
These questions help us identify whether what the model is reading as a “signal” actually comes from your marketing activity or from an external influence.
It’s important to note that having one of these elements present doesn’t automatically mean it should be added to calibration. Instead, our team will review each case carefully, assess its potential impact, and decide whether incorporating it will enhance or distort the accuracy of the model.
The objective is to make sure calibration helps the model learn better - not to overfit or “correct” the data unnecessarily. By working together on these discussions, we ensure your INCRMNTAL insights remain both precise and trustworthy.
What to Consider Before Calibration:
To evaluate calibration needs effectively, during the integration process, in addition to the KPIs and Spend inputs, we might ask you to prepare the following data:
- External Activities sheet (i.e Promotions, Sport events, weather) depending on your business’ specifics.
- Referral channels data
- Etc
Read FAQ regarding the Calibration and data in this article.
For any question please contact onboarding@incrmntal.com , or open a support Ticket here