Test the Customer Claims in Your Investor Deck Before the Meeting
On this page
A founder preparing an investor deck needs a customer claim that can survive a source check and guide the next product commitment. Subconscious connects controlled comparisons on modeled populations to estimated choice differences, uncertainty and a bounded recommendation. The useful output is a supported customer decision and an accurate claim beside the evidence.
Start with the consequential claim, the existing records and the action the founder can change. Adequate evidence may already settle the question. Commission a comparison when a missing answer could change the product, offer or message; keep investor judgment separate from the measured customer response.
The claim that actually gets challenged
A pitch deck may claim how many customers have a problem, what buyers will pay, which feature matters or how the company compares with alternatives. Each claim needs evidence appropriate to the population and outcome. Investor feedback can improve presentation without establishing those customer facts.
Subconscious's question-to-decision workflow names the owner, feasible alternatives, population and acceptance rule before the result. Review the complete choice task and attributes, then inspect estimates, uncertainty and relevant segment differences. Agree the configuration and supported study type before treating a specific claim test as delivered.
Finta shows a customer decision informed by that kind of work. The company adopted a $950 annual subscription and reintroduced a $99 monthly plan after its pricing study. The CEO reported five monthly-plan customers and continued sales at the old $348 annual price. The evidence supports pricing adoption and reported closes, not an isolated causal sales increase, unchanged demand or a fundraising outcome.
Choose a design that answers the actual claim
Representative descriptive measurement addresses problem prevalence. A randomized comparison can estimate how an assigned feature, message or price changes choices within a defined task. Price sensitivity requires an appropriate design and monetary scale; a modeled or stated choice is different from an actual payment.
An interval describes uncertainty under the study's assumptions. Repeated responses from the same simulated respondent are not automatically independent people. Inspect the population, measurement, assignment, analysis and relevant validation alongside the estimate. Aggregate method evidence does not establish every claim in a founder's deck.
Work one feature claim into a usable decision
Consider a proposed study for fictional ProjectHarbor, browser software for US adult sole proprietors managing client projects. The founder wants to decide whether CSV export belongs in the first paid package. The existing spreadsheet is a feasible alternative. This planning example contains no executed study or result.
Specify the complete comparison
Offer A provides one user, five active client projects, tasks, due dates and email support for $12 per month. Offer B provides exactly the same package and terms plus CSV export. Both versions are feasible to supply. No project limit, task wording, product image, support promise or other feature changes.
Both offers include tax, have no setup or extra usage fees, renew monthly at $12 and permit cancellation before the next charge. Access continues through the paid period; no partial refund is offered. At the paid period end, both versions stop edits and show existing records read-only for 30 days before deletion. CSV export is available only in B during the paid period. Review every stimulus before assignment so the export contrast is the only intended difference.
The proposed modeled population matches the stated geography and work context. Confirm coverage and relevant reference evidence; generated characteristics do not establish recruitment of verified current customers. Randomly assign complete A or B within the configured population. Ask whether the respondent would select the presented paid offer or continue with the current spreadsheet without buying ProjectHarbor.
Record the exact question, assigned offer, modeled choice and assignment counts. Estimate B minus A in paid-offer choice, with uncertainty that accounts for the design and repeated responses. The estimate concerns the export addition in this offer and task. The comparison does not establish how many businesses have the problem, real purchasing or a preference for every possible feature bundle.
Set acceptance before seeing the estimate
The founder specifies the smallest useful modeled choice gain, tolerable uncertainty and maximum implementation and support cost before the run. The product owner checks delivery feasibility; adequate existing customer and validation evidence enters the rule. The study may support including export when the supported difference justifies the commitment. Retain A when the gain cannot justify the cost, or leave the decision unresolved when uncertainty or conflicting evidence could reverse the choice.
The output is an exact package recommendation, its evidence and the affected product brief. No numerical threshold or winner is invented here. A favorable modeled contrast alone cannot support “customers bought more” or a revenue forecast. Those claims need the corresponding observations and financial inputs.
Put the result and source trail beside the deck sentence
If a completed study meets the prior rule, the deck can state the bounded finding using actual values. A sentence template is: “In the specified modeled US sole-proprietor choice task, adding CSV export changed paid-offer choice by [estimate] percentage points, with [interval]; the recommendation is [action].” Replace every bracket from the approved analysis and preserve the modeled label, task and uncertainty. The template is not a reported result.
If the rule is not met, remove the unsupported claim or state that the feature decision remains unresolved. A weaker result is still useful when the result stops an unsupported product commitment. Do not convert the sentence into “the market demands export,” an observed sales claim or a prediction about an investor.
Keep a source card beside the slide: exact study question and version, completion date, population and setting, complete offer texts, assignment and response counts, estimate and interval method, relevant validation and remaining limits. Name the analyst, accepting founder and product owner. Preserve links to the design and analysis artifacts so a reviewer can trace the sentence. The founder approves the words; the product owner accepts the associated package brief.
Resolve the remaining claim-specific evidence gap
Inspect existing calibration and observed records before adding another study. Where a consequential gap requires human or observed-outcome evidence, confirm recruitment and measurement arrangements and preserve the decision question while adapting the task. Check important rejected claims as well as promising ones. Agreement on a human choice task does not itself establish actual purchases.
Keep specialist, legal, customer and private-data restrictions attached to the affected claim. The source card should distinguish completed evidence from a proposed validation step. A meeting deadline cannot turn an unresolved assumption into an established customer fact.
Keep broader diligence and customer measurement distinct
Affinity's due-diligence guide covers financial, legal, market, product, business-model and team questions. Customer-claim testing supplies evidence for part of that assessment. The process does not provide a warm introduction, explain why a particular fund passed or negotiate a term sheet.
Before the next meeting, match each consequential customer claim to the appropriate evidence:
- Problem prevalence: representative descriptive evidence about the defined population, rather than an intervention effect.
- Price tolerance: modeled or stated sensitivity within the specified price task; actual paying requires consequential purchase evidence.
- Feature priority: choices among complete feasible alternatives, with generated responses labeled as modeled and the tested difference named.
- Positioning: rankings or choices in the specified context, with relevant evidence for transfer to the actual buying setting.
Book a Subconscious decision review with the exact deck claim, current evidence, feasible alternatives, intended population, meeting date and accepting founder. Name the product or offer commitment the answer can change. The review should settle the appropriate comparison, prior rule and source-backed claim packet before the founder uses the result in the deck.