Optimization: It’s About Time

Why Planners Must Understand Single-Period vs Multi-Period Optimization

A long-established principle in marketing science is that advertising works over time—not just in the moment it appears. Work as foundational as that of Simon Broadbent’s have been educating us about this since the 70s.   (The 4^th edition of Spending Advertising Money by Simon and Brian Jacobs (1984) covers this research and its impact on media decision-making at the time.)

An impression served today rarely produces all of its sales lift today. Instead, it generates a decay curve of effects that lasts for days, weeks, months, or even years depending on the channel, message, and category. These delayed effects interact with the lift created by subsequent ads, sometimes reinforcing them, sometimes diminishing them. What we observe as the “total effect of advertising” is really the accumulation of all these time-based interactions. This idea is supported by decades of research in advertising response modeling.

Models That Detect, Rather Than Assume, Time Effects

A well-designed model does not impose a time effect on the data. Instead, it identifies it empirically. For example:

• A brand TV campaign may show sales impact that persists for many months.

• A short-term digital promotion may decay in days.

In our experience, these general patterns are surprisingly consistent across brands and categories, although the scale of the effects may vary. But the model must be allowed to discover them. That is why we test multiple lag structures for every channel/content/creative combination until we reach a highly accurate model that also performs well in field validation.

Such a model forms the foundation for optimization algorithms used in planning.

But this is precisely where marketing leaders must exercise caution.

Where Planning Goes Wrong: The Hidden Assumption in Many Optimizers

Most advertising optimization tools optimize one period at a time. A “period” might be a week, a month, or a quarter—but the optimizer assumes your goal is to maximize results inside that window only.

This is a critical limitation.

Why it matters

Imagine a marketer planning the next quarter. The underlying model correctly predicts that certain channels—say, brand TV—will continue generating lift well into the following quarter (and beyond).

But a single-period optimizer ignores all lift that happens after the planning window.

As a result:

• Channels with long-term effects appear undervalued.

• Channels with short-term effects appear more attractive.

• Budgets unintentionally shift toward digital, promotion-heavy, or lower-funnel tactics.

Nothing in the model is wrong—the issue is the optimizer’s time horizon. Many planners do not realize this trade-off is happening under the hood.

This exact problem appears in the academic literature on dynamic advertising optimization, where single-period models are shown to bias allocations toward short-lived effects.

The Better Path: Multi-Period Optimization

A superior approach is a multi-period optimization algorithm—one that understands that advertising effects naturally spill over time.

Multi-period optimizers:

• Value both in-period and future-period lift.

• Produce allocations that reflect the true economics of advertising effects.

• Better match how real brands actually grow value across fiscal cycles.

When using these tools, it is essential to examine both:

1. In-period lift

2. Total lift across all relevant periods

The difference between these two can be substantial—and strategically meaningful.

A Practical Example: The End of the Fiscal Year

This dynamic becomes most apparent in Q4 planning.

Most marketers work within an annual budget and face a familiar dilemma:

• Maximize Q4 results (and hit internal targets)

• Or

• Invest in Q4 activity that also drives Q1 of next year

A single-period optimizer will aggressively favor Q4-only lift. A multi-period optimizer presents a more realistic picture:

• The true total return on each allocation

• The degree to which Q4 decisions influence Q1

• Where a marketer can “have their cake and eat it too” by hitting short-term targets and priming the next fiscal cycle

Accurate, lag-sensitive models empower better trade-off decisions.

Key Takeaways for Senior Decision-Makers

• Advertising works over time. Optimizers must reflect this.

• Many tools silently impose a single-period view of the world, skewing budget decisions.

• Multi-period optimization more accurately represents how brands grow and how advertising creates value.

• Reviewing both in-period and total lift supports better strategic judgment.

• This becomes crucial in high-stakes situations such as year-end planning.