Kerzner International – Luxury Hospitality
Meridian MMM for
Smarter Media Spend
“We want to understand how every marketing dollar drives revenue, so we can plan and allocate spend with confidence.”
THE CHALLENGE
Kerzner’s marketing team lacked a reliable way to measure how media spend translated into revenue across 11 markets and a complex mix of digital channels.
Attribution relied on siloed, last-click metrics that couldn’t separate the revenue media actually drove from the baseline demand the brand would have earned anyway. This made it hard to compare channels fairly, justify budgets or plan spend in a consistent, data-driven way
DISCIPLINES
Bayesian Modeling, Machine Learning, Cloud Architecture, Data Visualization
OUR APPROACH
Modeling spend for
smarter decisions
Uptimal proposed building a Bayesian Marketing Mix Model on Google’s Meridian framework, using Kerzner’s weekly revenue and media spend data enriched with search demand, seasonality and macroeconomic signals to isolate the incremental revenue each channel truly drives.
The pipeline was designed to automatically ingest new data on every run, train probabilistic models on Vertex AI and surface results through interactive reports. Because markets differ widely in scale and behaviour, we trained a dedicated Meridian model for each market cluster, tuning its channel priors, controls, carryover and sampling to each market’s dynamics rather than forcing a single one-size-fits-all model. Every model was validated through Meridian’s diagnostic framework, including posterior predictive checks, ROI credibility assessments and contribution validation, ensuring recommendations were both statistically robust and economically plausible.
-
Data Exploration
We analyzed three years of weekly revenue and media spend across 11 markets, identifying seasonality, carryover and the drivers behind each channel. Search demand, holiday and pre-booking windows, FX and macro signals were engineered as controls to give the model richer context. -
Model Development
We built a flexible Bayesian MMM using Google’s Meridian framework on Vertex AI and GCP, dynamically applying the most suitable model to each market cluster. Channel-level ROI priors, carryover and saturation ensured accuracy and scalability across Kerzner’s diverse portfolio. -
Data Visualization
We built interactive reports surfacing model fit, channel contribution, ROI, marginal ROI and cost per incremental KPI. These tools let stakeholders compare channels, reallocate budget and make incrementality-based decisions with confidence.
“Working with complex, multi-market
data let us design a modelling system
that adapts to each market, combining
Bayesian machine learning with cloud
automation to isolate true incremental
revenue and ensure long-term scalability.”
Senior Data Scientist
RESULTS
Goals accomplished!
The final solution gave Kerzner an automated, incrementality-based view of marketing performance across every market by turning budget allocation into a clear, data-driven and scalable process.More than a one-off measurement exercise, the framework operates as an ongoing decision-support capability: fixed-budget optimization scenarios show how existing spend can be redistributed across channels to maximize incremental revenue, identifying channels with headroom to absorb further investment and channels approaching saturation where additional spend delivers diminishing returns.
0.94
geo-level R-squared model fit, with 0.85 at national level
3.7x
highest channel return on investment, at $3.66 revenue for every $1 spent
11
markets modelled through a single automated Bayesian MMM pipeline
geo-level R-squared model fit, with 0.85 at national level
highest channel return on investment, at $3.66 revenue for every $1 spent
markets modelled through a single automated Bayesian MMM pipeline
Get in touch
Questions? A project in mind? Talk to us.
- Analytics, tracking or data questions: real humans answer, fast
- Enterprise: audits, sGTM, GA4 360, BigQuery and MMM engagements
- Tell us your stack and goals, we reply within one business day