Acquisition playbooks: distinguish new demand from credited demand

A non-branded search does not contain the focal brand's name. “Cold” commonly describes an audience without a known prior relationship. These concepts overlap in planning, but they are not identical. An existing customer may use a generic search, while a first-time customer may search for the brand after hearing about it elsewhere.

Start with the need. Organise an acquisition hypothesis around the problem being solved, the product's suitability and the reason to choose it. Generic category searches, competing products and relevant audience signals can represent different discovery opportunities. Their value depends on the actual offer and economics, not on a universal targeting hierarchy.

Use precise internal definitions. Separate the focal brand, competitor brands and genuinely generic terms where the reporting supports it. Record the classification rule so that changes in categorisation do not masquerade as changes in performance. Targeting settings and the search terms or placements that actually deliver activity should not be assumed to be identical.

Model the commercial boundary. In a simplified pay-per-click model, expected contribution before advertising per click equals purchase probability per click multiplied by expected contribution per purchase. Subtract cost per click to estimate contribution after advertising per click. The model requires aligned periods, suitable order and revenue definitions, and a representative product mix.

For illustration only, a 10% purchase probability and £12 pre-advertising contribution per purchase imply £1.20 expected contribution per click before advertising. At £0.90 cost per click, that leaves £0.30 before fixed costs. This is not an estimate for any real campaign. Observed averages can change as bidding, placement and audience composition change, and attribution is not proof that every purchase was incremental.

Leadership implication. Create a bounded learning budget with a hypothesis and decision rule. Do not scale indefinitely because traffic is labelled “cold”, or stop automatically because its attributed efficiency is below that of warm audiences. Compare the additional commercial value supported by the evidence with the additional resources required.

Branded activity can help people find an offer they already know. Retargeting seeks to reach audiences associated with earlier behaviours or relationships. The precise signals, exclusions and lookback periods depend on the product and platform. Neither method tells you with certainty what the customer would do without the advertisement.

The selection problem. People selected because they have already searched for a brand or visited a product may have a higher underlying likelihood of purchasing. Strong observed results may therefore reflect both the effect of the advertisement and the characteristics of the selected audience. This is why attributing a purchase and causing a purchase are not interchangeable claims.

That does not mean branded advertising or retargeting is inherently wasteful. They may support navigation, communicate a relevant offer, help comparison or influence choice. The right conclusion is conditional: evaluate the role and evidence rather than declaring these activities always essential or always redundant.

Design a fairer question. Instead of asking only “How much revenue was credited?”, ask “What changes when this activity is reduced or withheld under a credible comparison?” Google's conversion-lift documentation illustrates treatment/control measurement, though the tool and eligibility described there are specific to Google Ads. 1

For any experiment, define the treatment, business outcome, observation window and relevant guardrails in advance. Consider exposure overlap, changes in price or availability and whether the sample can support the decision. A simple campaign pause followed by a sales comparison is vulnerable to seasonality and other changes. It may be informative without being a clean causal estimate.

Leadership implication. Manage overlap across campaign roles and avoid allocating all investment to the activity with the most flattering attributed return. Conversely, do not remove a potentially useful activity solely because some purchases would happen anyway. The decision concerns the incremental contribution of the activity relative to its cost and alternatives.

Amazon's sponsored-ads documentation defines a purchase as new-to-brand when the customer has not purchased from the brand within the preceding 12-month lookback window. A new-to-brand detail-page view uses the corresponding browsing-history definition. These are different events; a new detail-page visitor is not necessarily a new purchaser. 2

The documentation reviewed for this guide states that sponsored-ads NTB metrics are available for Sponsored Brands and Display, and are not available for Sponsored Products in that reporting context. Availability and individual metrics can vary with campaign type and history. Check the current documentation and the account's actual reporting rather than filling missing data with an assumed value. 2

What NTB does not mean. It is not proof of the customer's first-ever interaction with the brand. A customer may have purchased outside the stated window or through another channel. Nor does the classification establish that advertising caused the purchase. It describes observed history within a defined reporting scope.

Keep volumes and percentages together. An NTB purchase share can rise because NTB purchases increase, because other purchases fall, or both. A percentage alone cannot tell you which happened. Also distinguish purchases, units, sales and unique customers: these are not automatically interchangeable denominators.

Illustrative reporting discipline. If a report shows £1,000 spend and 50 attributed NTB purchases, the ratio is £20 spend per attributed NTB purchase. Do not relabel that result “incremental customer acquisition cost” unless the evidence supports both the incremental and customer-level claims. Multiple purchases and identity definitions can matter.

Leadership implication. Use NTB alongside contribution, observed repeat behaviour and the strength of the causal evidence. It is a valuable lens on acquisition, not a substitute for assessing whether a customer relationship is new to the whole business or economically worthwhile.

Separate activity by commercial role where this makes the evidence clearer: discovery, comparison, branded demand and re-engagement. Reconcile those views to total business outcomes rather than summing overlapping claims.

At each review, record what is directly observed, what is inferred and what remains unknown. “The campaign received purchase credit” is an observation under a reporting model. “It generated additional profitable customers” is a stronger inference requiring further evidence.

Our proposed leadership standard is to make budgets conditional on a clear learning or commercial objective, a finance-approved loss limit and a review point. There is no universal sample size, test duration or NTB target that fits every business.

Conclusion. The purpose of an acquisition playbook is not to make every campaign look efficient. It is to identify where the business can develop additional, economically supportable demand, while remaining honest about what the measurements can and cannot establish.