Incremental sales with Marketing Mix: how we knocked down last click bias in Financial Services

Executive Summary

At Bunker DB, we implemented Marketing Mix Modeling (MMM) with various financial institutions across Latin America and, each time we designed the statistical models, we observed that digital upper-funnel efforts are undervalued by legacy attribution systems. Compared to traditional last-click attribution models, Bunker DB found that TikTok, Meta, and YouTube ranked among the channels with the lowest incremental marginal CPA.

In this research*, we delve into how the financial services industry can use MMM to reallocate marketing budgets resulting in incremental sales.

The attribution challenges in financial services

The financial industry is one of the most competitive sectors, including banks, fintech, insurtech, lending, and cryptocurrencies. Executive pressure to demonstrate performance has led CEOs to demand measurable results from their CMOs. Therefore, marketing attribution has become critical for justifying ad spend and optimizing performance.

Attribution allows marketing teams to identify the campaigns or channels driving conversions. However, the fragmentation of the marketing ecosystem makes attribution and measurement across multiple online and offline channels significantly challenging.

In financial services, this challenge is compounded by long decision cycles, strict data privacy regulations, and the intangible nature of many financial products. Consequently, the financial sector has traditionally relied on limited attribution models—often last click or first interaction—which significantly understate the value of branding and awareness initiatives in the upper funnel.

This need has driven the adoption of more sophisticated methodologies like Marketing Mix Modeling. Increasingly, institutions turn to measurement partners who can help them attribute in an unbiased, agnostic manner.

The strategic role of MMM

Marketing Mix Modeling emerges as a particularly fitting solution for these challenges, offering a robust alternative to individual tracking-based attribution models. MMM enables financial institutions to overcome the limitations of individual tracking by analyzing aggregated patterns and statistical correlations.

MMM uses advanced statistical techniques to analyze the relationship between marketing investments and business outcomes, incorporating external factors such as seasonality, economic events, exchange rates, inflation, unemployment, and even competitive activities. Its holistic approach is especially valuable in financial services, where consumer decisions are strongly influenced by macroeconomic factors.

Our results reveal that effective implementation of Marketing Mix Modeling can increase conversions by 3% to 15% with the same advertising budget. For financial institutions with significant marketing budgets, this uplift can translate into millions of dollars in additional value. MMM provides insights across multiple actionable levers, offering improvement opportunities that go beyond mere budget reallocation between channels.

Modeling financial results in practice

At Bunker, we start by developing a theoretical design of the factors that influence the evolution of the target variable. We then proceed to collect the variables and construct proxies for those for which information is unavailable or whose measurement is partial or spaced out over time:

Footnote: Theoretical example of relevant variables and their availability

Having accurate data is essential to modeling incrementality

Defining which variables truly impact the primary outcome is complex. Macroeconomic factors influence card activations, account openings, or app registrations, but the causal relationship is difficult to isolate, has long-lasting effects, and data availability varies by market.

Often, the ideal variable does not exist or is only available with monthly granularity. To address this, we build theoretical models to identify the most relevant variables and then seek the best possible proxy.

Price in finance is not always easy to compare

While price is a key determinant in most verticals, in finance it is not always straightforward to compare. Including discounts, chargebacks, and other benefits significantly improved model performance.

Incorporating contextual variables

In addition to macro data such as seasonality, holidays, interest rates, and benefits, it is essential to incorporate contextual variables. Among the MMM implementations for financial institutions, we concluded it is best practice to include:

• Google Search SOV (Share of Voice): to estimate advertising competitive pressure.
• Google Trends GQV (Google Query Volume) weighted: to capture variations in aggregated service demand.

At Bunker Analytics, we connect all digital channels and retrieve historical data; we model aggregated media channels and, when they represent more than 1% of the mix, differentiate upper vs. lower funnel.

Understanding short-term vs. mid-term impact

Most digital spend is directed at short-term actions:

• Generate a lead
• Download an app
• Request a card

However, the consumer goes through multiple steps before switching their primary financial institution or applying for a loan. To capture these implications, we test hypotheses and apply Weibull adstock, enabling us to model the delayed impacts of media investment.

Footnote: modeled delayed impact of Investment in TikTok in incremental Sales. Spend (in red), estimated incremental conversion (in blue).

Key results and shareable learnings

Undervaluation of the upper funnel

MMM can reveal the long-term demand-generation impact of upper-funnel efforts. Institutions consistently investing in awareness building establish brand associations that influence future decisions. In all cases, upper-funnel efforts yielded incremental impact ratios of 2.5:1 to 5:1 compared to last-click (non-direct 30 days) or GA4 DDA attribution models.

Reallocating digital budget from lower to upper funnel

Opportunities in the upper funnel include optimizing media mix between traditional channels (TV, radio, outdoor) and digital channels (online video, social media). MMM can identify synergies among these channels to maximize awareness and consideration impact. Reallocate budget from the lower funnel (branded and unbranded search, remarketing, referrals) to upper-funnel channels to maximize conversion volume at lower CPAs given the current budget.

Incremental potential

Between 3% and 15% additional conversions at mid-term can be generated with the same budget by redistributing media allocation within reasonable thresholds (up to 30% variation).

Marketing contribution

The contribution of these models to total conversions in the short and mid-term (up to 6 months) ranged from 17% to 32%.

Model validation with incrementality studies

In 3 out of 5 cases, we conducted conversion lift experiments or geo experiments to measure the incremental cost per action, calibrate, and validate MMM results.

TikTok in the spotlight

In recent years, TikTok has emerged as an essential marketing channel in financial services in Latin America and has gained share in the marketing mix. In our analyses, TikTok represented 0% of the mix at the end of 2023, averaged 3% during the measured period, and reached up to 17% by the end of the measurement.

Footnote: estimated response curves for the simulated scenario

Average vs. Marginal CPA

Although TikTok’s average CPA ranked among the top five lowest CPA channels in all studies, it stood out even more for its marginal CPA, always appearing in the top three. Two factors explain this performance:

• The investment level did not show diminishing returns.
• Significant contribution from mid-term effects.

No other channel with less than 15% share demonstrated similar performance; only one other channel reached the top three for marginal CPA in a single study.

Conclusion

Marketing Mix Modeling emerges as the most suitable methodology to address the unique challenges of the financial services sector, including long decision cycles, strict privacy regulations, and the complexity of the omnichannel ecosystem. Our results reveal that effective MMM implementation can increase conversions by 3% to 15% with the same advertising budget, while providing actionable insights for strategic media mix optimization.

Full-funnel analysis uncovers significant opportunities at each stage of the customer journey. In the upper funnel, MMM quantifies the long-term value of brand-building investments—critical for emerging fintechs seeking to establish consumer trust. In the lower funnel, MMM identifies optimal touchpoint combinations to maximize conversions in complex omnichannel decision processes.

If you want to discover how to take your measurement strategy to the next level, contact our sales team to get started.

* Analysis based on 5 Marketing mix results from financial institutions across different Latin American countries executed during 2025.