How Agencies Can Test Strategy-Deck Recommendations Before a Pitch
On this page
An agency can strengthen a strategy pitch by showing which proposed action earns a better modeled response, alongside the facts supporting the recommendation. Subconscious compares controlled alternatives and estimates choice effects with uncertainty. The useful deliverable is one recommendation slide the client can inspect and accept, revise or hold.
For River City Federal Credit Union, a study with 140 simulated respondents identified the weakest staff-proposed slogan. The same slogan later performed worst in live Facebook and Instagram ads. That agreement supports testing a message before committing spend. The campaign comparison does not isolate the causal effect of the simulation or slogan. The public River City case records the comparison and its scope.
Separate the strategic premise from the proposed action
A response comparison can support a recommendation without proving the premise behind the entire strategy.
Client records, category data and qualitative research can establish different parts of the argument. A catalogue can establish what the client sells. Interviews can explain the problems particular participants describe. Neither a preferred headline nor generated comments establish how common a problem is across a market.
Keep three statements separate:
- Premise: the independently sourced client or category fact that makes an action relevant.
- Comparison: the response to specified alternatives in the defined study.
- Recommendation: the action the agency proposes, under an agreed acceptance rule.
A recommendation can also follow directly from established evidence. A verified incorrect product description needs correction. An existing study that covers the audience, offers and outcome can answer the pitch question. Commission another comparison only for a material gap.
Test a feasible recommendation, not an abstract promise
Consider an illustrative pitch for a printable household-maintenance calendar. The agency is deciding whether to recommend an action-led headline instead of the current product-name headline. The example is a proposed design, not a reported client result.
Assume the seller confirms one complete offer: USD10 total, including applicable taxes, for a downloadable PDF. The file contains twelve undated monthly pages in US Letter format, with blank spaces for tasks and dates. Printing is the buyer's responsibility. The purchase permits personal household use, includes no subscription or automatic reminders, and provides a full refund on request within fourteen days. The seller must verify the file, fulfilment and refund terms before fielding.
Compare current headline A, “Household maintenance calendar,” with proposed headline B, “Keep the next maintenance check visible.” Show the same product image, facts, price and terms in both conditions. The outside alternative is to continue the existing household list and skip the purchase.
Define the intended audience as US adults who maintain a paper household-task list and are considering a printable calendar. Inspect the actual population construction, available fields, selection and counts against those criteria. A requested behavioral criterion is not proof that the modeled population contains that group.
Define the contrast and the action rule
Randomly assign modeled respondent units to A versus the outside alternative or B versus the same outside alternative. Estimate the difference in modeled purchase-choice probability, B minus A. Report the actual unit counts, estimator, baseline and confidence interval. Repeated choices from the same modeled respondent require dependence-aware uncertainty; repeated answers are not independent new respondents. The methodology explains controlled designs, baselines and uncertainty.
Before seeing results, the agency and client marketing owner should agree the minimum useful choice gain. The rule must account for actual creative, production and placement costs. Recommend B only when the estimated contrast and uncertainty support that rule. Retain A when the evidence supports retaining A; mark the choice unresolved when uncertainty prevents either conclusion. The largest point estimate alone is insufficient.
The estimate concerns the headline's effect on modeled choice for this offer and audience. It does not measure completed sales, task completion, demand for household maintenance, or the psychological reason for a response.
Put the recommendation and its evidence on one slide
The slide should let the client follow the inference without searching the appendix. For the illustrative calendar decision, use this structure and insert only actual evidence:
| Slide element | Inspectable content | Client action |
|---|---|---|
| Established premise | Seller-confirmed printable calendar, offer terms and source references; separately labeled evidence about the audience's problem | Accept the factual basis or name the unsupported premise |
| Defined comparison | Exact A/B stimuli, common outside option, actual population construction and counts, assignment and estimated contrast with uncertainty | Check whether the comparison answers the proposed headline decision |
| Recommendation and owner | Adopt B, retain A or hold under the prior gain-and-cost rule; named client marketing owner accepts the action | Approve the scoped action, request a revision or identify the exact missing evidence |
Keep the tested wording and offer version with the slide. A recommendation for this landing-page headline does not justify changing the entire brand position, category strategy or product promise. A broader recommendation needs evidence that covers the broader action.
The appendix can retain the complete design, source records and analysis. Distinguish simulated responses from direct customer testimony. Label the agency's interpretation and any unsupported assumption explicitly. A client can then challenge the premise, method or action separately without discarding useful evidence elsewhere in the deck.
For deadline planning across the whole deck, use the agency strategy-deck delivery guide. This page addresses the evidence behind the recommendation itself.
Bring the recommendation the client must accept
Bring the proposed recommendation slide, existing client evidence, complete alternatives and accepting owner to a Subconscious decision review. Agree whether existing evidence answers the question or which bounded comparison would change the recommendation. Leave with a defined action and the exact evidence needed to support the client sentence.