The Challenge: Seeing the Full Picture Across Paid, Owned, and Earned Media
A major telecom operator was investing across the full spectrum of modern marketing. Display banners, search campaigns, social media activations, TV, and radio were all running simultaneously. Yet the organization could not answer a deceptively simple question: how did these channels interact to drive sales?
The leadership team wanted to understand the complete customer path to purchase, accounting for the influence that paid, owned, and earned media exerted at every touchpoint. The objective was precise: measure the full impact of digital media investments, including display, search (both branded search and unbranded search), social media, and offline channels, on the business’s bottom line .
This challenge is increasingly common in complex, high-consideration categories like telecom. Consumers do not see channels in isolation. A TV spot sparks a branded search. A display ad reinforces a social message. Organic search captures demand that advertising created. Without a unified marketing measurement framework, these interactions remain invisible, and budgets are allocated based on last-click bias rather than true incremental impact.
The Proposed Approach: Nested Modeling for a Multi-Touchpoint World
To optimize the total customer experience, the telecom operator partnered with MASS Analytics to implement a Marketing Mix Modeling program that treated each media element as part of an interconnected ecosystem rather than a standalone silo.
Why Nested Modeling Matters
Traditional regression models assume that each channel independently contributes to sales. In reality, channels feed one another. TV advertising drives consumers to search. Social buzz amplifies display effectiveness. Search volume is not independent of TV spend; it is often a direct consequence of it. Statistically, this creates an endogeneity problem: if you simply dump search and TV into the same equation as independent variables, you misattribute credit and produce misleading ROI by channel.
To solve this, MASS Analytics deployed nested modeling, a technique where sub-models are embedded within a primary sales model. This hierarchical approach allows the model to account for both the direct indirect impact of each channel. For example:
The sales model (full model) captures TV, radio, and other top-line drivers.
A nested search model (reduced model) captures how TV and social activity drive branded and unbranded search queries.
The output feeds back into the full model, adjusting the sales contribution of each channel to reflect both its direct effect and its indirect influence on other touchpoints.
Nested modeling goes beyond simple channel reporting. It allows marketers to quantify the true relationship between paid owned earned media, adjust sales contribution by considering indirect impact, and gain a complete view of the brand’s ROI.
Data Scope: Three Years at Weekly Granularity
The model was built on three years of weekly data. A specific focus was placed on digital media, where data was collected and analyzed at the most granular level possible. This depth was essential because digital channels move fast. Weekly granularity allowed the model to capture short-term fluctuations in search behavior, social engagement spikes, and display impression delivery that monthly aggregations would smooth into irrelevance.
The MassTer Studio Platform
The entire project was executed within MassTer Studio, MASS Analytics’ end-to-end marketing analytics software. MassTer was configured to deliver three core capabilities:
Identify sales and performance drivers
Calculate the ROI for each marketing activity
Optimise budgets and run predictive analysis
Best industry practices were implemented to automate most of the modeling steps, turning what is traditionally a fragmented process into an efficient workflow.
Process Module: Interactive Modeling
The model was built using an interactive modeling approach within MassTer Studio‘s Process Module. This allowed the analytics team to construct the nested relationships visually, defining how each touchpoint influenced the next. The platform’s pre-programmed processors handled data transformation, ensuring that the raw weekly inputs were formatted correctly for the proprietary algorithms.
Results & Insights: What the Attribution Screen Revealed
Once the nested model was built, the team used MassTer Studio’s attribution screen to decompose the direct indirect contribution of every channel. The findings reshaped how the telecom operator thought about its media mix.
Digital Media Drove a 12% Sales Uplift
Digital media collectively generated a 12% sales uplift and paid back on investment. This validated the overall digital strategy, but the real value lay in understanding how that uplift was created.
TV and Search: A Powerful Interaction
The model revealed a strong interaction between TV and digital media, particularly in search. TV campaigns generated significant brand interest that consumers then expressed through branded search queries and visits to the brand’s website. In other words, TV was creating demand that search captured.
This is a classic demand capture dynamic. TV acts as the demand generator. Search acts as the demand capture vehicle. Without nested modeling, the search channel would have claimed full credit for these conversions, and TV would have appeared less effective than it actually was.
Display Underperformed Relative to Search
Display activity showed a lower ROI when compared to search. While display contributed to awareness and reach, its direct conversion efficiency lagged behind intent-based channels. This insight alone justified a strategic reallocation.
Social Media Generated Measurable Buzz
The social media campaign proved successful beyond simple click-through metrics. It generated significant buzz, leading to 30% more interactivity on the brand’s Facebook page. This earned media effect, while not directly attributable in a last-click model, was captured by the MMM framework as part of the broader media ecosystem.
The Optimisation: From Insight to Action
The results were fed directly into MassTer’s Optimization Module, which prescribed specific budget optimisation shifts:
Search budget was increased at the expense of display. Given search’s superior ROI and its role as a demand capture channel, shifting spend from display to search improved overall efficiency without increasing total investment.
Branded search budget was reduced during heavy TV flights. During periods of intense TV activity, the model showed that consumers were already motivated to search for the brand. Paying for branded search keywords simply captured traffic that would have converted through organic search anyway. This is a critical insight for marketing spend optimization: sometimes the most profitable budget cut is the one that stops you from paying for customers who are already yours .
Unbranded search budget was increased during TV peaks. Conversely, during heavy TV periods, the team increased investment in unbranded search. The logic: TV was generating category interest, and consumers were searching for generic terms related to the product. By increasing unbranded search presence, the brand could intercept this newly created demand before competitors did .
These moves represent sophisticated media mix optimization. Rather than treating each channel as an independent line item, the telecom operator used cross-channel synergy insights to dynamically adjust budgets based on how channels interact over time.
Why the Customer Journey to Purchase Requires Nested Thinking
The telecom operator’s experience illustrates a fundamental truth about modern marketing measurement: the customer path to purchase is not linear, and channels do not operate in isolation.
Direct vs. Indirect Impact
Every channel produces two types of value. Direct impact is the sale that happens immediately after exposure. Indirect impact is the influence a channel exerts on another channel or on future behavior. TV rarely generates an immediate click, but it fundamentally alters the probability that a consumer will search, click, and buy later.
Traditional attribution models, especially last click, systematically undervalue indirect channels and overvalue demand capture channels like branded search. Marketing Mix Modeling with nested modeling corrects this bias by statistically isolating both effects.
The Paid, Owned, Earned Media Triangle
The model also shed light on how paid owned earned media interact:
- Paid media (TV, display, paid search) created the initial stimulus.
- Owned media (the brand website, organic search presence) captured the demand.
- Earned media (social buzz, interactivity) amplified the message and extended reach at no incremental cost.
Understanding this triangle is essential for digital media optimization. If you only optimize paid media in isolation, you underinvest in the owned and earned infrastructure that converts and amplifies your spend.
Key Takeaways for Marketing Leaders
This case study offers several actionable lessons for organizations navigating complex, multi-channel environments:
Measure interactions, not just contributions
A channel that looks inefficient in isolation may be the engine that makes another channel work. Nested modeling is the technical capability that makes these interactions visible.
Separate branded and unbranded search strategy
Branded search and unbranded search serve completely different purposes. Branded search captures existing intent; unbranded search intercepts new intent. Their budgets should move in opposite directions based on upper-funnel activity.
Use econometrics, not just platform reporting
Platform dashboards tell you what happened inside a single channel. Econometrics tells you what happened to the business because of the channel. For strategic budget optimization, you need the latter.
Invest in granular data
The three-year weekly dataset was essential to detecting short-term interactions between TV and search. Monthly data would have missed these dynamics entirely.
Automate the workflow
By using marketing analytics software like MassTer, the telecom operator turned a complex, nested modeling project into a repeatable process. This speed matters when media plans need to adapt quarterly or monthly.
Conclusion
For the major telecom operator, the journey from raw data to optimized action required one critical shift: moving from channel-level reporting to ecosystem-level modeling. By applying nested modeling within MassTer, the team could finally see the complete customer path to purchase, quantify direct indirect impact, and make media mix optimization decisions that respected how paid owned earned media actually work together.
The results speak for themselves. A 12% digital sales uplift identified. Search and display budgets reallocated for maximum efficiency. Branded search spend cut during TV peaks to protect margin. Unbranded search increased to capture new demand. And a 30% social interactivity boost proving that earned media effects are real and measurable.
In an era where marketing measurement is under more scrutiny than ever, the message is clear: if your model cannot see interactions, it cannot optimize them. And if it cannot optimize them, you are leaving growth on the table.
Ready to talk about your business?
Whether you’re approaching MMM for the first time or looking to improve an existing measurement programme, we’d be glad to walk through what it would look like for your specific structure.