Nobody moves neatly from awareness to consideration to purchase. They loop, widen, narrow, and loop again. Budgets built on the old diagram pay for the wrong half of the journey.

In short

• Buyers alternate between widening their options and narrowing them, repeatedly, until something tips the decision.

• Most of a category is not in the market at any given moment, so a budget aimed only at buyers ready today is aimed at a very small room.

• The split between creating demand and capturing it is a budget decision, not a creative preference, and there is published evidence for where to start.

• The practical change is to brief campaigns against a loop rather than a stage.

Every marketing team has a funnel diagram somewhere. Wide at the top, narrow at the bottom, arrows pointing down. It is a useful teaching device and a poor description of how anyone actually buys anything.

Ask a colleague how they chose their last significant purchase, personal or professional. You will hear about a recommendation, then some reading, then a shortlist, then more reading that added an option back onto the shortlist, then a pause of several weeks, then a decision that was partly rational and partly a feeling that one option seemed safer. That is not a funnel. It is a loop with a way out.

What the research found

Google's behavioural research team studied this directly and gave it a name that stuck: the messy middle. Between the trigger that starts someone thinking and the purchase that ends it, buyers move between two mental modes.

• Exploration, which is expansive. The buyer widens the set of options and the set of things they think they should care about.

• Evaluation, which is reductive. The buyer narrows towards a decision and applies criteria.

  
  

The important finding is not that these two modes exist. It is that people cycle through them repeatedly, in no fixed order, for as long as they need to. A single new piece of information in evaluation mode can push someone straight back into exploration and add two options to a shortlist that was nearly closed.

The research also identified a set of biases that consistently shape which option wins in that loop, including category heuristics, social proof, authority, scarcity and the power of now. In simulations, a fictional challenger brand supported by those advantages took a substantial share of preference away from an established favourite. Not because it was better, but because it was easier to choose.

The winner of the messy middle is usually not the best option. It is the option that was easiest to justify choosing.

The second finding most budgets ignore

The loop describes people who are already thinking about a purchase. The harder question is how many people that actually is at any moment.

Research from the Ehrenberg-Bass Institute with the LinkedIn B2B Institute put a number on it that has been repeated ever since: roughly 95% of potential buyers in a category are not in the market today. They will be later. The category is bought infrequently, contracts renew on multi-year cycles, and most of the audience is simply not shopping.

Put those together and a familiar strategy starts to look expensive. A budget spent entirely on people showing buying signals today is competing, with everyone else, for the smallest and most contested slice of the category. Worse, by the time someone shows a buying signal, the loop has usually already run. The shortlist exists. You are either on it or you are paying to interrupt a decision that has largely been made.

So what is the right split?

This is where marketing arguments usually become arguments about taste. They do not have to be. Les Binet and Peter Field analysed roughly a thousand case studies in the IPA Databank and found that campaigns allocating around 60% of budget to brand building and 40% to activation produced the strongest long-run results. Their later work with business-to-business cases put the split nearer 46:54, reflecting shorter consideration and more rational evaluation.

Two things matter about those numbers. First, they are averages across a database, not a law, and the right answer for a specific business depends on how well known it already is, how often the category is bought and how aggressive the competition is. Second, almost nobody is anywhere near them. The typical split, when people actually audit it, is closer to 90:10 in favour of activation, because activation is what the reporting rewards.

How to have the argument properly

Three questions that settle the split faster than opinion:

1. How often does someone in our category buy? Long cycles mean a larger share of the audience is out of market at any moment, which raises the case for creating demand.

2. What does a customer pay us over their life with us? If the value is high and the cycle long, the payback window for demand creation is long too, and it should be judged over years rather than quarters.

3. Are we losing deals we never knew about? If sales regularly hears "we shortlisted three and you were not one of them", the problem is not conversion. It is that the loop ran without you.

Briefing against a loop instead of a stage

The practical change is smaller than the theory suggests. You do not need to rebuild your campaign structure. You need to change what each campaign is asked to do.

For exploration mode: be easy to find and easy to understand:

People in exploration are collecting possibilities and forming a rough sense of what good looks like. They are not comparing you to a specific rival. They are deciding which category of solution they are even in.

What works here is content and creative that defines the problem clearly, states what you do in plain terms, and is available on the surfaces where people widen their search: communities, video, peer conversations, review platforms and summaries. Creative quality matters more here than targeting precision, because you are trying to be remembered rather than to be clicked.

For evaluation mode: remove doubt, do not add persuasion

People in evaluation are narrowing and looking for reasons to strike options off. Every ambiguity is a reason to remove you. Pricing that cannot be found, a specification that is vague, a comparison page that only exists on a competitor's site, an unanswered question in a public thread.

This is where the unglamorous assets earn their budget: comparison pages, eligibility criteria, cost structure, implementation timelines, named references, and honest answers to the objection nobody says out loud.

For the loop between them: make yourself easy to come back to

The most under-built asset in most programmes is the bridge. Someone explored, narrowed, then paused. Weeks later something restarts the process. Whether you are still in the set depends almost entirely on whether they remember you and can find you again quickly.

That is what consistent presence buys. Not immediate response, but survival in a shortlist that reassembles itself later.

What this changes in reporting

A budget split across creating and capturing demand cannot be judged with one number, and this is where most attempts quietly collapse.

• Demand creation is judged on reach among the right audience, branded search trend and share of answer. Not on last-click conversions, which it will always lose.

• Demand capture is judged on cost per qualified enquiry and cost per closed customer, matched through the CRM rather than counted in the platform.

• The programme as a whole is judged on pipeline created and the trend in cost to acquire over a period at least as long as your sales cycle.

  
  

Set those separately and the perennial argument disappears, because nobody is asking an awareness campaign to prove itself with a conversion metric it was never designed to produce.

The uncomfortable part

Demand creation is harder to defend in a quarterly review and easier to cut in a difficult month. That is precisely why the average brand under-invests in it, and precisely why the brands that hold their nerve tend to compound an advantage that is very difficult to buy back later.

If you take one thing from the research, take this: the loop runs whether or not you are in it. Being present when someone is exploring costs less than trying to reopen a decision after it has closed.

Frequently asked questions