Give dependable channels and experiments different contracts

A marketing portfolio needs reliable delivery and room to learn; separate their commitments, evidence standards and stopping rules before the quarterly allocation becomes a negotiation.

The established channel has a forecast. The new channel has a persuasive advocate. At budget time, both ask for the same treatment: confidence from management and protection from cuts.

The CMO needs a better distinction than old versus new. An established channel can become uncertain, while a new activity can test a well-understood mechanism. What matters is the job each commitment performs and the evidence behind it.

A practical portfolio gives dependable delivery, improvement work and exploration different spending contracts. This is a proposed management method, not a universal allocation formula.

Define the job of the money

Delivery funding supports activity with a reasonably understood operating model. The expectation is execution within agreed commercial and service limits. It still requires monitoring; previous performance is not immunity from changing costs or demand.

Improvement funding addresses a known constraint in that model. Perhaps acquisition is effective but onboarding loses suitable customers, or sales lacks evidence for a common objection. Its contract should specify the constraint and the improvement sought.

Exploration funding buys information about an uncertain opportunity. Its first deliverable may be a decision to stop, refine or expand, rather than a fully scaled revenue stream.

Write these roles beside budget lines. Without that distinction, an experiment is easily condemned for not behaving like a mature channel, while a mature channel can excuse poor delivery by calling everything learning.

Reject the automatic percentage split

A fixed allocation rule can start a discussion, but it cannot determine what your company needs. A business facing immediate liquidity pressure has different options from one with stable cash generation and a narrow future pipeline.

Assess exposure first. How much demand depends on one platform, one partner or one customer segment? How quickly could that dependence be reduced? What happens if acquisition costs rise or the channel becomes unavailable?

Then consider execution capacity. Funding ten experiments with enough staff to run three creates weak evidence and exhausted teams. An unstaffed opportunity is not diversified merely because it appears in the budget.

The Green Book’s appraisal guidance provides a useful general principle: compare feasible options and their risks. The commercial allocation method here is an adaptation, not government advice on marketing percentages.

Make exploration answerable

An experiment proposal should name the uncertainty, the decision it affects, the smallest useful test and the maximum commitment before review. It should also explain what result would make the team abandon the idea.

Suppose a hypothetical industrial distributor wants to test webinars for technical buyers. “Generate engagement” is too loose. A stronger question is whether a specific maintenance topic attracts suitable buying accounts that progress to a credible technical discussion.

The first test might examine audience fit and operational delivery. It may not establish incremental annual revenue. The proposal should say so, rather than adding a speculative sales multiplier to make the forecast impressive.

Evaluation guidance in the Magenta Book stresses examining the chain between an intervention and an outcome. For this webinar, attendance, suitability and progression are different links in that chain.

Create gates that release real resources

A review gate should make a resource decision possible. It may release the next production commitment, add a market, or stop further recruitment. A meeting that merely requests more slides is not a gate.

Before the first gate, fund only the work needed to reach it. Avoid committing the entire annual supplier contract and then describing monthly reporting as a flexible experiment.

Consider reversibility as well as size. A small activity that creates a long support obligation may be harder to unwind than a larger one-off research project. Include those commitments when comparing proposals.

For dependable channels, gates can be simpler: agreed operating ranges, exception triggers and scheduled reassessment. The objective is control proportional to uncertainty, not administrative theatre.

Keep the portfolio from grading itself

The person sponsoring an experiment should help interpret it, but should not be the only person deciding whether the evidence supports expansion. Include someone who understands the commercial outcome and someone who can question the measurement.

Do not reward teams solely for the number of tests declared successful. That encourages easy questions, forgiving thresholds and selective reporting. Reward useful decisions and reliable records, including well-supported decisions to stop.

Equally, avoid treating every stopped experiment as wasted money. If a modest commitment prevents a much larger poor investment, the learning can be valuable. The claim must identify the decision changed rather than assigning an invented financial value to every insight.

Review the mix as conditions change

At each allocation review, ask whether activities have changed category. An experiment may now have enough evidence for dependable delivery. A familiar channel may need renewed investigation because its economics have deteriorated.

Also ask which important uncertainty has no owner. Portfolios tend to accumulate visible campaigns while neglecting awkward questions about customer fit, service capacity or measurement quality.

The resulting budget should explain what the company is counting on, what it is trying to improve and what it is prepared to learn. That explanation is more durable than a fashionable percentage split and more useful when circumstances force the CMO to choose.