By Amit Akhelikar & Darya Likhareva
It is October. Brand leads are meeting to agree next year's plan. The total marketing budget is fixed. Every brand has a case for more. The discussion runs on last year's spend, this year's forecast and who argues best.
The marketing mix modeling report arrives in February. It estimates how much advertising and promotion contributed to sales. The analysis may be careful, but it arrives after the allocation decision.
Marketing mix models should help decide how to allocate money between brands, as well as between channels. To do that, they need to reach the planning discussion before the budget is locked.
Consider an illustrative example. Brand A generated $3 in incremental revenue for every dollar spent, but its next dollar is expected to generate only $0.50. Brand B generated $2 per dollar, but its next dollar is expected to generate $1.80. Historical average return favors A. For additional investment, B offers the better revenue return, assuming comparable margins, timing and risk.
That distinction is marginal ROI: the estimated return from a small increase in spending. It can differ sharply from average ROI when a channel approaches saturation.2
Illustrative example: average ROI favors Brand A, but the next dollar works harder in Brand B.
At Lynx Analytics, this view comes from our work with leading pharmaceutical companies on marketing measurement and commercial planning. We build MMM into the marketing budget workflow, connecting field, HCP and consumer investment with portfolio decisions. The model informs the options; commercial leaders own the call.
MMM earns its keep when it tells you which brands deserve the next dollar, not just which channel did best last year.
A channel ranking can help a brand manager improve a campaign. A portfolio leader also needs to decide which brand should receive more funding and where that funding should come from.
Budget-setting habits can reinforce that gap. In the August 2021 CMO Survey, 41% of respondents reported setting annual marketing budgets from the previous year's expenses, with adjustments as needed; 10% revisited and potentially reset them monthly or quarterly.1 This is cross-industry evidence about budgeting practices, rather than proof of how pharma portfolios allocate funding.
In our work with commercial teams, three gaps can prevent that comparison: results arrive after planning, brand models use different financial measures, or recommendations rely on average returns. None is solved by producing another channel report.
Ask what a proposed budget increase would actually fund within each brand. Would it expand a productive activity, reach a new audience or add spend to a channel already delivering diminishing returns? Then compare that opportunity with the investment being reduced elsewhere.
This matters for new funding too. In a portfolio allocation exercise we supported, additional media budget was assessed centrally across brands rather than distributed according to historical budget shares. Each investment had to compete against the alternatives.
Putting brand-level ROI figures beside one another is not enough. Agree the comparison before the modeling begins:
The comparison should show each brand's expected return at different investment levels. A response curve describes how outcomes change as spending changes; each brand has its own curve. The portfolio decision combines those estimates with the activities that can realistically absorb the money.
Stability can be a rational choice. Field continuity, physician relationships, contracts and regulatory commitments may limit how quickly investment moves. Make those constraints visible rather than treating every dollar as immediately transferable.
A model delivered after the annual plan cannot inform that plan. Work backward from the decision date: allow time for data preparation, model review and discussion before leaders commit the budget.
Our implementation work has aligned quarterly model refreshes with annual planning, mid-year forecasts and launch reviews. We have built optimization and simulation tools so brand leads and finance partners can compare allocations, change assumptions and examine trade-offs during the discussion. Planners can select their decision window, set minimum and maximum spend by product and cost type, and preserve geographic commitments.
That approach starts with the business decision. Robyn's MMM guidance likewise treats business questions as an input to model design and includes forward-looking allocation among its applications.3
Four practices turn the analysis into a usable planning process:
The practical shift is to make MMM part of the marketing budget workflow. Updated response estimates feed the planning cycle; brand teams and finance compare the current plan with alternative allocations, agree where money should move and record the decision. Subsequent reviews revisit those assumptions and inform the next round of investment. The analysis becomes a recurring input to annual budgets, mid-year reallocations and campaign adjustments.
The question MMM should answer changes with lifecycle stage.
A launch brand has little history. That makes its response harder to estimate; it does not establish that the opportunity is small or large.
Where evidence is limited, use comparable brands to inform starting assumptions, then update them as the brand's own data develops. Compare plausible funding levels and show how sensitive the recommendation is to those assumptions.
Be especially cautious about predictions far outside observed spending. Meridian's guidance highlights that extrapolation risk increases as scenarios move further from historical experience.2 A staged investment with a defined learning objective may be more useful than a confident-looking point estimate.
For an established brand, start by asking whether the existing budget could work harder elsewhere. A channel can have a strong historical return while offering little return from additional spending.
One allocation decision we supported redirected roughly $35 million from linear TV toward channels with higher estimated marginal returns. The brand team used the model to move existing budget into more productive opportunities.
A further brand-level decision quadrupled investment in a third-party video channel where the model indicated room to grow. The decision was subsequently credited in client reporting with approximately $20 million in demand.
At portfolio level, a client estimated an approximately 29% improvement in media effectiveness, translating into an estimated $200–300 million in incremental revenue as spending shifted from linear TV and endemic display toward streaming TV, social, audio and programmatic buying. These are client estimates of impact; the $35 million figure describes budget reallocated. The examples illustrate individual decisions and program results, rather than aggregate outcomes across clients.
Challenging the allocation: four decisions from our work, each labelled by the type of estimate.
The decisions went in different directions because the opportunities differed. The common thread was using the model to challenge the existing allocation and give commercial teams a concrete alternative.
Testing also mattered. In one display program, vendor reporting suggested a 100% return while MMM estimated a return near zero. A follow-up test indicated that the brand was spending about twice what it needed. That led to a practical question: how much support should remain, and where could the rest work harder?
As a brand approaches loss of exclusivity or faces declining opportunity, reassess how much current promotion contributes. Separate estimated baseline demand from the incremental effect of promotion, while recognizing that this separation depends on model assumptions.
Baseline demand is not a guarantee that sales will persist after spending stops. Historical promotion can have carryover effects, and competition or market access can change the outlook.
Compare reduced, maintained and increased investment over a realistic horizon. Agree in advance what evidence would justify a reduction, what support must remain and when to revisit the decision. Finance and commercial leaders should set those criteria together.
Evidence after generic entry also argues against an automatic withdrawal. Research covering 72 brands, summarized by IESE in 2025, found average detailing budgets fell to 57% of their pre-entry level, while one in six brands increased them. Firms that redirected visits toward physicians with stronger brand preferences achieved greater ROI improvements.4 The finding concerns detailing, rather than portfolio MMM, but reinforces the need to assess both spending levels and targeting.
Where the expected return falls below a viable alternative, and commitments allow, release funding for that alternative. Lifecycle stage prompts the review; the evidence determines the allocation.
Take next year's plan and ask one question for each brand: what would additional investment buy here, compared with putting that money into another brand?
If your current MMM can answer on a common financial basis, bring it into planning. If the evidence is too old, the brand estimates are not comparable or the output stops at average ROI, address that gap first.
For a practical starting point, request a 30-minute portfolio diagnostic with Lynx Analytics. Bring your current allocation and an overview of your MMM outputs. Together, we will map which brands your current analysis can support, where evidence is missing and what needs to change before the next budget decision. The takeaway is a one-page summary of those gaps and priorities.
The Brand A and Brand B figures are illustrative, not client data. Client examples are drawn from Lynx Analytics engagements with pharmaceutical commercial teams; client details are anonymized, and results are client estimates of individual decisions and program outcomes, not aggregates across clients. This article is a perspective, not investment or regulatory advice.
© 2026 Lynx Analytics. All rights reserved.