Demand Creation vs Demand Capture: Where Should Your Next Marketing Dollar Go?

One of the easiest ways to misallocate marketing budget is to confuse capturing demand with creating it.

Both can generate revenue, but they do different jobs. Performance marketing is often exceptionally good at finding people who are already close to buying and making it easier for them to convert. Demand creation operates earlier, increasing the probability that future category buyers will know, consider and ultimately choose the company when they enter the market.

The distinction becomes important because the two activities are not equally easy to measure. Demand capture usually happens close to the transaction. A customer searches the brand, clicks a paid-search advertisement, returns through retargeting or responds to a conversion campaign and purchases shortly afterwards. The interaction is visible, the revenue arrives quickly and the attribution system can connect the two.

Demand creation is less convenient. Advertising may influence somebody months before they are ready to buy, and its effect may be expressed through familiarity, preference or consideration rather than an immediate transaction. When that person eventually enters the market, the final sale may be recorded against search, direct traffic or another channel positioned much closer to the purchase.

A measurement system built primarily around observable conversions can therefore make the activity harvesting demand look considerably more productive than the activity responsible for increasing the amount of demand available to harvest. If those reported returns are then used directly for budget allocation, the business can gradually shift capital towards capturing an existing pool without noticing that the pool itself has stopped growing.

The useful question is not whether demand creation or demand capture is more important. A healthy growth system needs both. The allocation question is whether the next marketing dollar creates more value by improving the conversion of existing demand or by increasing the number of future buyers willing to choose the company.

The demand available to performance marketing is finite

I find it useful to think about the immediate commercial opportunity as three components:

Available prospects = target customers × proportion currently in market × proportion willing to buy from you

The first component, the total addressable customer population, usually changes relatively slowly. The second is determined largely by the category and its purchase cycle. Only a fraction of potential customers will be actively buying at any particular moment, especially for products purchased infrequently.

The third component is where marketing can make a substantial difference: among the people who eventually enter the category, how many are willing to buy from your company?

Performance marketing mostly operates against the available pool created by those three factors. Search reaches people expressing intent. Retargeting concentrates on customers who have already interacted with the business. Conversion campaigns use behavioural signals to identify people with a relatively high probability of purchasing.

These activities can be highly productive, but increasing their budget does not automatically expand the number of equally valuable prospects available to them. As spend rises, the platform still needs somewhere to deploy the additional money. It can raise bids, broaden targeting, increase frequency, move into less productive inventory or concentrate more heavily on customers already likely to convert.

This is why diminishing returns are a normal feature of performance marketing rather than evidence that the channel suddenly stopped working.

The early portion of investment may reach customers whose behaviour is genuinely changed by the intervention. Additional spend can still create value, but eventually the marginal opportunity deteriorates because the business is attempting to harvest more demand than currently exists at the same quality.

At Growth Dynamics, we describe the flatter part of this response curve as the Death Zone. Spend continues to increase while the additional effect on topline revenue becomes progressively weaker. Attribution can continue reporting conversions because the advertising remains present around customers who buy, but an increasing share of those customers may have converted without the final layers of investment.

This is one reason average ROAS can be so misleading for allocation. The productive early spend and the weaker marginal spend are blended together, allowing the channel average to remain attractive after the economics of the next pound have changed materially.

Demand creation increases the pool available to harvest

Demand creation is sometimes reduced to awareness, but awareness alone is not particularly valuable. A person can recognise many companies without being willing to buy from any of them.

The commercially meaningful change is in preference and consideration. When somebody eventually enters the category, is your brand among the options they are prepared to choose?

That willingness can be influenced by accumulated experience with the company, advertising, reputation, product quality, distinctive creative, recommendations, physical availability and many other factors. Much of this develops before the active purchase journey begins.

That creates an important measurement asymmetry.

Imagine somebody is exposed repeatedly to advertising over several months, develops a preference for the brand and eventually enters the market. When they are ready to purchase, they search the company's name, click a paid result and buy.

Paid search has clearly played a role in the final transaction. It may even have incremental value by making the purchase easier or protecting the customer from competitors.

What it probably did not do is create the entire preference that caused the branded search.

If the measurement system assigns the commercial value almost entirely to the final observable interaction, the business can end up funding the mechanism that captures demand more aggressively than the mechanisms responsible for replenishing it.

This does not create an immediate crisis. The existing pool can often be harvested efficiently for a considerable period. The problem becomes visible later as performance budgets need to work harder to produce the same level of growth.

Why attributed ROI naturally favours demand capture

This measurement bias is not necessarily caused by a bad attribution model. It is partly a consequence of observability.

Activity close to the transaction leaves clearer data.

A branded search click happens seconds or minutes before the purchase. Retargeting operates against identifiable visitors. Conversion campaigns can report the customers who purchased after being exposed to the campaign.

Demand creation can happen much earlier, across multiple channels, devices and offline environments. Its contribution may appear gradually through stronger consideration, more branded demand or a greater willingness to choose the company later.

Attribution therefore tends to have a much easier time describing demand capture.

The organisational consequences can be significant when attributed ROI becomes a management target. Marketing teams respond rationally to the metric they are given. Capital moves towards channels that can demonstrate conversions clearly and quickly, while activity whose contribution appears earlier or over a longer horizon becomes progressively harder to justify.

Brand search, retargeting and high-intent conversion activity can then absorb more of the budget precisely because they are closest to customers who already possess substantial purchase intent.

The numbers can improve while the marketing system becomes more dependent on an existing pool of demand.

Eventually the company reaches a position where additional performance investment produces less incremental growth. If leadership responds by demanding even higher attributed efficiency, the incentive to concentrate further on existing demand becomes stronger.

The issue is therefore not that performance marketers are behaving irrationally. They are often behaving entirely rationally inside a measurement system that rewards the easiest revenue to claim.

Brand and performance work as one economic system

For this reason, I do not find the traditional brand versus performance debate particularly useful.

A business that creates preference but fails to convert it efficiently wastes valuable demand. A business that becomes excellent at conversion while doing little to increase future preference eventually restricts the pool available to its performance system.

The economics of the two are connected.

If brand activity increases the proportion of future category buyers willing to choose the company, performance marketing operates against a better pool. More people search the brand. Conversion rates can improve. Retargeting encounters customers with stronger prior preference. Acquisition becomes easier because the company is no longer trying to persuade every buyer from a standing start.

This interaction can make channel-level measurement difficult to interpret.

Suppose a period of effective brand investment increases branded demand significantly. Paid search subsequently reports more conversions and stronger revenue because more people are actively looking for the company. An attribution system may reasonably record those transactions against search, even though the increase in search activity was itself partly produced by marketing elsewhere.

Search has captured the demand effectively. That does not mean it created all of it.

Understanding where the next dollar belongs therefore requires examining the state of the whole system rather than ranking channel ROAS in isolation.

You can observe more of this than marketers often assume

Demand creation is frequently described as difficult to measure, which is true if measurement is defined as assigning a precise incremental ROI to every individual activity. It does not follow that nothing useful can be observed until an MMM provides a coefficient.

There are several ways to examine whether the underlying demand pool appears to be strengthening.

Share of Search is useful because it introduces competitive context. Absolute branded demand can increase simply because the category is growing. If the company's share of relevant search demand is increasing relative to competitors, the evidence of strengthening preference becomes more interesting.

Branded organic search provides another useful view because it is not contaminated by the decision to purchase the click through paid brand search. Direct and organic revenue can help show whether the business is monetising more demand without requiring an attributable paid interaction immediately before conversion.

Buyer evidence provides a different perspective. Post-purchase research, loss surveys, sales calls and reviews can show whether people were aware of the company before entering the active purchase journey, what created consideration and which factors ultimately influenced choice.

None of these measures proves the incremental ROI of a particular campaign. They should not be represented as if they do.

Their role is to establish whether the underlying conditions for growth appear to be improving and to create hypotheses about what may be driving that movement.

The same principle applies to demand capture.

If performance spend has increased materially, examine what happened to total revenue and commercially meaningful mid-funnel activity. Did more customers enter the purchase journey? Did the pool expand in proportion to the spend? Did branded demand strengthen, or did the additional budget mostly increase activity around customers already showing high intent?

If the relationship between spend and business outcomes begins to flatten, the next step is to understand why. Saturation may be the explanation, but so might poor creative, low-quality placements or changes in campaign mechanics.

Observed evidence will not settle the causal question, but it can identify where the economically important questions are.

Incrementality becomes valuable when the decision needs stronger proof

When the amount of capital at stake becomes large enough, descriptive evidence may no longer provide sufficient confidence.

If a company believes several million pounds of conversion investment has moved beyond the productive part of its response curve, leadership may reasonably want to know what revenue is genuinely at risk before removing it.

That is where incrementality testing becomes useful.

By deliberately changing the intervention and constructing a counterfactual, the business can estimate how much the commercial outcome changed because the treatment changed.

The important point is that incrementality does not transform a channel into a binary classification of working or not working. The result depends on the amount of investment being tested.

The first portion of a performance budget may be strongly incremental while later spend produces progressively smaller additional effects. Brand search is a useful example. In a competitive environment, a core level of paid coverage may protect genuinely valuable demand, while higher levels of investment increasingly pay for customers who would have reached the company organically anyway.

A test showing that brand search has incremental value therefore does not prove that every existing pound of brand-search spend deserves to remain invested.

The business is ultimately interested in the shape of the response: how much additional value is created at different levels of investment, and when the marginal return becomes less attractive than an alternative use of the capital.

Designed experiments can provide stronger evidence about that relationship. They still need to be interpreted in context because the effect applies to the treatment, markets and conditions actually tested, and the platform may not deliver the intervention identically everywhere.

Incrementality is stronger causal evidence, not a universal truth detached from the experiment that produced it.

Demand creation is harder to identify because the signal is often weaker

The causal measurement problem becomes more difficult for activity intended to create demand over longer periods.

Consider two marketing interventions that eventually create the same amount of incremental revenue. One produces most of the effect within a week. The other spreads the effect gradually over six months.

Economically, the total effect can be identical. Statistically, the second is much harder to identify because each week's contribution is smaller relative to all the other variation affecting the business.

Revenue changes because of seasonality, price, distribution, competition, promotions, economic conditions and numerous other factors. A small marketing effect spread across many periods can disappear inside that noise even when it matters commercially.

This is where econometric modelling becomes useful, but also where its assumptions become more influential.

An MMM can represent delayed effects using adstock, priors and other modelling structures. What it cannot do automatically is create information that was absent from the historical data. If several long-term explanations are compatible with the same revenue history, decisions about lag structures, priors, controls and the baseline can materially influence the decomposition.

The longer and weaker the claimed effect, the more important it becomes to understand whether the data is genuinely identifying the result or whether modelling choices are carrying a larger proportion of the answer.

For demand creation, this makes triangulation particularly important. Changes in relative preference, buyer-reported evidence, mid-funnel behaviour, creative quality and experimental findings can all provide independent information capable of supporting or challenging what the econometric model concludes.

The answer to a weak long-term signal is not necessarily a more complicated model. Often it is a stronger body of evidence around the model.

Creative makes channel-level allocation more complicated

Demand creation also exposes another limitation in the way marketing budgets are often discussed.

The channel is not the asset producing the entire effect.

Media determines where and how widely the advertising is distributed, but the creative determines what people actually experience. Its quality can therefore materially change the economics of the media investment.

This matters when historical channel performance is used to plan future budgets.

A two-year coefficient for television or paid social is an estimate of the historical portfolio that ran during those two years. The creative changed, as did audiences, placements, bidding systems and media costs. The coefficient does not describe a permanent property of the channel independent of what was put through it.

If next year's creative is substantially stronger or weaker, the response to additional media may differ from the historical average.

This is why I would be cautious about treating the allocation problem purely as a question of moving pounds between channel labels. The treatment needs to be understood as the combination of the media, the creative and the execution through which it was delivered.

Increasing demand-creation investment behind weak advertising may simply buy more distribution without creating proportionally more demand.

Understanding creative quality therefore belongs inside the allocation discussion rather than being treated as a separate optimisation exercise after the budget has already been decided.

Deciding where the next dollar belongs

There is no permanent ratio between demand creation and demand capture that can be applied across businesses.

The appropriate allocation depends on the state of the marketing system.

A company with strengthening preference, growing branded demand and significant untapped conversion opportunity may have more to gain from improving demand capture. Another business may already be investing heavily in performance while its marginal returns flatten and its relative preference weakens. In that situation, pushing more money into conversion activity may simply make an already saturated system more expensive.

The evidence needed to make that judgement comes from several places. Observe whether preference is strengthening relative to competitors. Understand what customers say drives consideration and choice. Examine how meaningful customer behaviour changes as performance investment changes. Look inside the media execution to determine whether apparently weak economics are caused by saturation or by poor creative, placements and campaign mechanics.

Where the allocation decision is large enough, use designed causal evidence to test the important hypotheses. MMM can then help connect the wider portfolio once there is sufficient evidence underneath the model to challenge its assumptions.

This is a more useful approach than starting with a predetermined brand-performance split because the allocation can change as the business changes. The productive level of demand capture depends on the size and quality of the demand pool available at that moment, while the return from demand creation depends partly on the company's existing strength, competitive position and quality of execution.

The ratio is an outcome of the economics, not a rule to impose on them.

Demand creation and demand capture should be measured together

The distinction between demand creation and capture is useful because it prevents the final observable interaction from receiving credit for the entire customer decision. It becomes less useful if the organisation then manages the two as unrelated systems.

They are economically connected.

Demand creation increases the number of future buyers willing to choose the company. Demand capture converts that preference efficiently when those buyers enter the market. The performance system becomes easier to scale when the available pool expands, while strong conversion ensures that the value created by preference is actually monetised.

A useful marketing measurement system should therefore show whether the pool is growing, how efficiently it is being harvested and where each system begins to encounter diminishing returns. It should also make clear which conclusions are observed, which have stronger causal evidence and which depend substantially on modelling assumptions.

Growth Dynamics uses that evidence to make an allocation decision rather than to defend a predetermined philosophy about brand or performance. In some cases the next dollar should go into capturing existing demand more efficiently. In others, the greater constraint is the number of people willing to choose the company in the first place.

The purpose of measurement is to understand which constraint the business has now, and where additional investment has the best chance of changing it.