Dr. Ramla Jarrar of MASS Analytics on performance marketing vs brand marketing, and how to split your marketing budget without cutting what pays back later.
Performance marketing vs brand marketing looks like a choice between two kinds of ads. It is really a budget question with a timing problem inside it. Performance spend pays back within days, but brand spend pays back over months.
So when one model scores both on the same short window, brand loses every time. Marketers then move money to performance, so the demand that brand would have built never shows up.
This article gives the method we use at MASS Analytics to split a budget between the two. It has three steps. Each step stops one specific mistake.
Performance marketing vs brand marketing: what each one does
Performance marketing buys a measurable action. For example, that action can be a click, a sign-up or a purchase. Paid search, paid social, display and affiliate are the channels most often used for it. You see the result within days, and you pay for it as it happens.
Brand marketing builds the reasons people choose you later. It also raises awareness, consideration and preference. TV, online video, out of home, audio and sponsorship are the channels most often used for it. The sale often arrives weeks or months after the ad.
Both drive sales. They work on different clocks. Neither one is better, so the real question is how much each should get.
One warning before the table. The channel does not decide which one an ad is. The goal of the ad does. A TV spot that pushes a promotion is performance marketing. A paid social video that builds awareness is brand marketing. So label each campaign by its goal.
Why one budget model scores brand and performance unfairly
Most planning models score each channel on the sales that arrive soon after the spend. That window suits performance channels, because their effect is almost immediate. It does not suit brand channels, because their effect builds over months.
The picture is skewed. So brand looks like the weakest line on the plan, and performance looks like the strongest. The next planning round then moves money toward the channel that looks better in the short window.
Two things also make this worse. Both make brand look weaker than it is.
First, brand sales hide inside base sales. Base sales are the sales a brand makes on its own. Say someone sees a brand ad today and buys three months later. A short-window model cannot link the two. So it files that sale as base, and the ad gets no credit.
Second, paid search takes credit it did not earn. Say the same person sees the brand ad, then searches for the brand a week later and clicks a search ad. The search ad gets the full credit because it came last. The brand ad created the wish to buy.
Our guide on how to measure brand correctly in your Marketing Mix Model covers the brand side in detail.
A fair split starts with a fair scoreboard. Marketing Mix Modeling (MMM) can separate each channel’s return into the sales that arrive in the first weeks and the sales that arrive later. Once that is done, both kinds of spend get judged over the same horizon.
Step 1: split the budget into two pools and optimize each one
Start with one decision that no optimizer should make for you. Leadership sets the split between brand and performance first. That decision depends on category position, brand strength and growth ambition, so a challenger brand and a market leader should not land on the same number.
Sort every campaign by its goal
Each pool follows the goal of the campaign. For example, a TV promotion goes in the performance pool. A TV spot that builds awareness goes in the brand pool. So if your model carries one TV line, split it by campaign type first. That needs spend data by campaign type, so ask your media team for it early.
Nest the brand KPI in the sales model
Brand does not drive sales in a straight line. Brand activity first moves a brand KPI, for example awareness or consideration. That KPI then moves sales, so brand reaches sales in two steps. Nested modeling puts this chain inside one model, with the brand KPI nested in the sales KPI. So the two pools are not unrelated budgets. They share one model, and a brand result shows up in sales.
Optimize each pool on its own KPI
Once the brand KPI is nested in sales, you optimize in two parts. You optimize brand activities for the brand KPI. You optimize performance activities for sales.
Brand activities work under reach and frequency floors. You judge them by whether the brand KPI grows. You also judge them by whether sales follow through the nest. Performance activities work under channel caps. You judge them by incremental revenue and blended ROI.
Why not one run for everything? Because one optimizer that ranks every activity on short-window ROI will always prefer performance. The math is correct. The strategy is wrong. So the nested setup stops brand and performance from competing for the same dollar on terms that favor one side.
Inside each pool, the optimizer also works on marginal returns, meaning what the next dollar earns. We explain that logic in our article on where the next marketing dollar should go. Budget optimization software that supports floors and ceilings per channel, such as MassTer Mind, can then run each step.
Step 2: protect a minimum brand budget the optimizer cannot cut
Even with two pools, pressure builds at every planning review. Brand spend is the easiest line to trim, because its cost shows up later. So we fix a floor before the optimizer runs.
A brand floor is a minimum spend that the optimizer cannot go below. The marketing leader sets it, so the algorithm never decides it. When an optimizer’s answer contradicts a known business commitment, the answer is wrong. The commitment is not.
To set the floor, run two scenarios in the model: one with no brand media and one with reduced brand media. Then read what happens to base sales over the following years. Set the floor above the level where base sales start to erode. The numbers also show what a cut really costs.
Add floors by channel where the business needs them, for example reach and frequency minimums for TV. Add caps where inventory or creative capacity runs out. The scenario planning work in MassTer Mind shows how to test a plan before you commit the spend.
Step 3: in year one, move no more than 20 to 25% of last year’s spend in any channel
A model learns from the spend levels it has seen. Inside that range, its response curves are reliable. However, far outside it, the curve extrapolates into territory with no data behind it.
So in year one, we cap the move. No channel changes by more than 20 to 25% of last year’s spend. This keeps the plan inside the range the model has seen. It also gives the team a result to check before the next move.
Here is a hypothetical example. Say TV took $4M last year. A 20% cap lets the plan spend between $3.2M and $4.8M on TV. A 25% cap widens the band to between $3M and $5M. The optimizer can find the best point inside that band. It cannot jump to $8M because the curve says so.
Once a year has passed, the model has seen the new spend levels, so the range can widen based on what the new data shows.
Proof: a consumer health brand that kept its $5M
We applied this thinking to a consumer health and wellness brand. Its previous model said TV, online video and out of home returned about $1.00 per dollar. The draft plan moved $5M out of brand.
The rebuilt model measured returns over the full period. Brand media returned $2.30 per dollar, so the brand kept the $5M at flat total spend. The numbers by channel are in our brand vs performance marketing case study.
Frequently asked questions
Performance marketing buys a measurable action, such as a click or a purchase, and pays back within days or weeks. Brand marketing builds awareness and preference, and pays back over months or years. For example, a search ad that sells a product is performance, and a TV spot that builds awareness is brand. Both drive sales, so the useful question is how to fund each one, not which one to choose.
No single ratio fits every business. The right split depends on your category, brand strength and growth goal. In our experience, brand has held between 30% and 40% of the media budget. Treat that as a starting range, then evaluate any split in a model that measures both over the full period.
Yes, if the model separates each channel’s return into early and late sales. Performance channels show most of their return in the first weeks. Brand channels show much of theirs afterward. Once both are compared over the full period, each channel gets a fair ROI. Marketing Mix Modeling does this.
Evaluate the cost first. Brand cuts save money now and show their effect later, because base sales erode. In our experience, stopping brand media can cost between 15% and 25% of base sales over time. Run a no-brand scenario in your model, then set a floor.
Set a brand floor and optimize each pool for its own KPI. The floor is a minimum spend that the optimizer cannot go below. You optimize brand activities for the brand KPI, nested in the sales model, and performance activities for sales. Set both rules before the run, because without them a short-window ranking steers money to performance.
In year one, move no more than 20 to 25% of last year’s spend in any channel. The model has learned from the spend levels it has seen, so large jumps leave its curves without data to support them. This also gives your team a result to check before the next move. After a year, the safe range can widen.
The question to ask before the next plan
For the full method behind this, from building the model to reading the curves, see our Comprehensive Marketing Mix Modeling Guide. To see how your own brand and performance budgets would split, book a demo with MASS Analytics.
Before you sign the next plan, ask one question. What time window did your model use to score brand?