Multi-funnel marketing: how I plan, split and measure TOFU, MOFU and BOFU

Performance teams love the bottom of the funnel because it is easy to measure. Brand teams love the top because it is easy to justify. In my experience, the businesses that grow sustainably do both on purpose: they give every stage of the funnel a job, a budget and a metric, and they judge the whole system rather than any single campaign.

Over roughly 20 years and about $100M in ad spend, I've seen the same pattern many times. Cut the top of the funnel and bottom-funnel ROAS looks great for a quarter, then brand searches fall and acquisition costs rise.

Fund the top without measurement and nobody can defend the budget. Multi-funnel marketing is the discipline of avoiding both traps. It doesn't need a huge budget or a complex attribution stack, just clear roles, separate reporting and the patience to judge each stage on its own terms.

The three stages, and the job of each

• TOFU (top of funnel): create demand. Reach new people who fit your audience, make the brand memorable and grow the pool of future searchers. Typical formats: video, social, display, programmatic, creators, OOH.
• MOFU (middle of funnel): build consideration. Re-engage people who showed interest, answer their questions and earn the click. Typical formats: retargeting, content, comparison and category search, email, lead magnets.
• BOFU (bottom of funnel): capture demand. Convert people who are ready to buy. Typical formats: brand search, shopping, retail and quick-commerce ads, dynamic remarketing, sales follow-up.

The mistake is to judge all three by the same metric. A TOFU video will never match a brand-search ROAS, and it isn't supposed to. Its job is to feed the stages below it.

Brand vs non-brand: the split that reveals the funnel

Search and marketplace data is where the funnel becomes visible. Brand queries are mostly BOFU: people already know you. Non-brand queries sit higher, where people are still choosing. On Blinkit for Priya Foods, brand keywords returned 33–46x ROAS while non-brand ran around 5x. Both numbers are healthy for their stage. Judging non-brand against brand would have cut exactly the spend that brings in new customers, and over time it would shrink the brand demand that produces the 33–46x.

That is why I always plan and report brand and non-brand separately, and why I treat growth in brand search volume as one of the clearest signals that the upper funnel is working.

How I split the budget across stages

There is no universal ratio; it depends on brand maturity, category and margins. I start from the business goal and work backwards:

1. Fund BOFU to capture the demand that already exists, up to the point where marginal ROAS falls below target.
2. Fund MOFU to convert the audiences you already have: site visitors, engaged viewers, past buyers, leads.
3. Fund TOFU to grow future demand, with a budget sized to your growth target rather than whatever is left over.
4. Review the split monthly, using stage-level KPIs and brand search trends, not last-click revenue alone.

A newer brand usually needs proportionally more TOFU to create the demand it will later capture. A mature brand with strong search volume can lean more on BOFU efficiency, but still needs enough TOFU to keep the pool from shrinking.

Measurement: one KPI per stage, one view of the whole

• TOFU: reach in the target audience, cost per thousand, video completion, engaged visits, and above all brand search volume and direct traffic over time.
• MOFU: engagement rate, return visits, add-to-cart or lead rate, cost per engaged user, lead quality.
• BOFU: conversions, CPA, ROAS, and new vs returning customer share.
• Whole funnel: blended CAC, contribution margin after media and incremental revenue.

Last-click attribution will always flatter BOFU, so I use it for in-stage optimisation but not for budget decisions between stages. For those I rely on a mix of tools: geo or holdout tests where volumes allow, time-series comparisons of brand search and conversion rates when TOFU spend changes, and platform lift studies as directional evidence. None is perfect, but together they're far better than pretending a single attribution model tells the truth.

What full-funnel looks like in practice

For Shin Chan with Sony Pictures × Ramoji Film City, the brief set a joint objective: lower acquisition cost and higher return on ad spend, without sacrificing real people through the door. The outcome was CAC down about 30%, ROAS up about 60% and footfall up 45%.

Media efficiency alone would not have been enough; the campaign had to create interest and then turn it into visits, which is a full-funnel problem by definition. The lesson I take from it is that the final business outcome, here footfall, should sit above every stage-level KPI, so no stage can look good while the result stays flat.

In lead generation the same logic applies, with lead quality as the bridge between stages. At Teleperformance, tightening the journey from first touch to qualified lead improved lead quality by 40% and shortened conversion cycles by 25%. For Margadarsi, combining wider reach with stricter qualification took search impression share from 12.8% to 25.1% while keeping junk leads at 12.9% and lifting hot-lead-to-sale conversion from 2.11% to 5.00%.

Making the funnel connect

• Build audiences from each stage for the next: video viewers and engaged visitors become MOFU audiences; MOFU engagers become BOFU remarketing lists.
• Keep the message consistent: the promise in TOFU creative should be the one the landing page and the checkout deliver.
• Use first-party data and server-side tracking so the signals that connect the stages survive browser restrictions and consent choices.
• Exclude recent buyers from acquisition campaigns and move them into retention, where the economics are different.

Signs your funnel is out of balance

• BOFU ROAS is high but flat, and spend can't scale without ROAS dropping: you're harvesting a pool that isn't growing.
• Brand search volume is declining while performance spend is stable.
• TOFU spend is rising but neither brand search nor MOFU engagement moves: the creative or targeting isn't working.
• Most conversions come from returning customers, and the new-customer share keeps shrinking.

Where to start

If you're not sure whether your funnel is balanced, start by splitting your current spend and results by stage and by brand vs non-brand. That alone usually shows where the next euro or rupee should go. My Media Plan + Audit does exactly that, and gives you a staged plan with KPIs for each layer.