Adding Causal Testing to an Existing Client Scope: A Decision Guide
On this page
Add research to an existing client scope when a specific decision can still change and the current evidence leaves material alternatives unresolved. The agency needs to define the missing evidence, incremental work, and client's approval before commissioning a study.
The revenue argument is real, but it points at the wrong question
The Business Research Company's September 2026 market report forecasts growth in research and analysis services. That publisher's forecast does not establish demand in a particular agency account. Start with the client's decision and budget.
A summary of opinions may help define alternatives without estimating the effect of an action. The risk is selling that summary as proof that a proposed campaign, price, or offer will change the endpoint the client cares about.
What evidence does the decision require?
Separate the response source from the comparison design. A human survey can include randomized offer tasks, while a simulated study measures generated responses. Neither stated nor generated choice alone establishes actual purchases.
| Research approach | What it answers | Best fit | Where it fails |
|---|---|---|---|
| Unrandomized survey or qualitative exploration | Reported needs, reactions, and hypotheses | Defining alternatives or understanding context | Does not isolate the effect of an untested intervention |
| Randomized human stated-choice study | A tested contrast on human choices within the task | Comparing specified offers with relevant participants | Hypothetical choices may differ from purchases |
| Randomized generated-choice study | A tested contrast within the configured simulation | Screening defined alternatives when matched validation supports that use | Model bias, population mismatch, and transfer to human behavior remain |
| Randomized live test | A contrast on measured behavior, such as conversion | Assessing a deployable change with eligible traffic | Results apply to the tested setting and intervention |
For a budget decision, identify whether the client needs stated preference, observed conversion, or another endpoint. Existing interviews or surveys may already supply part of the required evidence.
Audit the account before pitching anything
Skip the general pitch. Look at each client's near-term calendar and ask, for every candidate decision:
- Is there real budget or reputational cost attached to the outcome (media spend, a launch date, a repositioning campaign)?
- Are there two or more concrete alternatives already defined, rather than an open-ended "what do you think"?
- Would the client change what they do if the evidence pointed a different direction than their current plan?
These answers establish whether a comparison can inform the decision. They do not select the response source. A small live test or human choice task may be more direct than a simulated screen, depending on access and the endpoint needed.
Building the case with one client first
Choose one client decision with unresolved alternatives. Agree a pilot scope before running the comparison: versions, audience, endpoint, estimate and uncertainty, delivery date, fee or pricing basis, approval owner, and the existing work it supplements or replaces. Share the result as a decision input, including an inconclusive outcome.
Evaluate whether the pilot changed the decision or reduced a named uncertainty. Use that evidence to discuss future studies and their costs rather than assuming that one successful comparison establishes a recurring package.
Where does a causal experiment fit, and where does it not?
Subconscious uses controlled choice studies on generated responses to compare defined alternatives. Its public replication evidence evaluates agreement with human estimates in prior study designs. It does not establish that a new client market is validated or that generated preference predicts the client's actual campaign performance.
Plan a human comparison with aligned alternatives and endpoint definitions when the decision requires it. Record fielding changes and let conflicting human results change the recommendation. How Subconscious works can help frame the proposed study.
The agency still owns strategic interpretation and client context. Verify any applicable human-participant requirements. A controlled comparison needs relevant alternatives and a defined endpoint; when those are missing, begin by clarifying the decision and customer evidence.
Common objections, answered by scope rather than by discount
"We already do research." Identify which endpoint the existing work measures and whether its design already compares the alternatives. Add a study only for a specific evidence gap.
"Can't we do this ourselves?" Running a comparison is one step. Defining the right alternatives, framing the outcome that matters, and translating the result into a recommendation the client will act on is the work an agency is already being paid for.
"We don't have budget for another line item." State the incremental fee, timing, and deliverable. Identify what existing work could be reduced or replaced, then let the client approve the tradeoff.
"Is this reliable enough to act on?" Explain the response source, design, validation, and remaining uncertainty. A generated screen can miss useful options; define which human or live evidence is needed before the client commits.
Review the aggregate validation evidence, or discuss a client decision with its alternatives, budget, approval owner, and required endpoint.