B2B Intent Creative Tests: Cohorts, Outcomes and Uncertainty

Current MarketPulse research workflow

BrandWell MarketPulse: intent data starting at $500/mo. Find people researching topics relevant to what you sell, apply saved ICP filters, and review available contact details before taking action.

MarketPulse plans are starting at $500/mo. Verify current package scope and offers on the product page.

MarketPulse standard monthly plans

Plan	Topics and refreshes	Monthly exports
Starter	1 topic, weekly	25,000 contacts
Growth	3 topics, daily	100,000 contacts
Scale	10 topics, daily	Unlimited contacts

Swipe the table to see all columns.

A pull retrieves available data from the previous 24 hours for one topic. Daily use across all included topics provides 90 Growth pulls or 300 Scale pulls in a 30-day month. Weekly Starter pulls vary by calendar month. Available audience size and contact fields vary, and audiences may overlap. Refreshes stop at the plan limit without overage charges.

• Find and qualify: search the topic catalog, save your ICP, and review the people and companies that match.
• Act on the research: export qualified contacts or prepare an Outreach or Direct Mail workflow. Connected products, channel usage, and fulfillment may cost extra.
• Get help from AIMEE: Growth and Scale include AIMEE for the initial user to research accounts and prepare the next step through connected tools.

Agencies and other service businesses can build client deliverables around their available Intent Scans and exports. Keep each client’s scope, permissions, and service fees clear. A subscription is not a promise of a white-label platform or exclusive access to a market.

Explore MarketPulse and watch the demo or compare plans starting at $500/mo. Any early-access discount is a separate, time-limited offer shown to an eligible recipient.

Define one message decision

Begin with a decision the team will make after the test. Examples include whether to lead with a problem, proof, comparison, implementation plan, or risk-reduction offer. Then write a falsifiable hypothesis: accounts showing recent research on implementation may respond better to an operational guide than to a category-awareness message. State the alternative and the practical threshold that would change the creative plan.

Define the cohort by fit, topic or stage evidence, freshness, source, identity level, exclusions, and minimum size. Intent is probabilistic evidence. It does not prove that a particular person researched a topic, requested personalization, or is ready to buy. The creative should reflect a useful job-to-be-done hypothesis, not reveal a surveillance-like detail.

Pre-register the primary outcome and window. Click-through rate may diagnose attention, but qualified pipeline or a well-defined intermediate event should drive the business decision. A hypothesis without a decision rule becomes post-hoc storytelling.

Freeze cohorts and actual assignment

Freeze cohort eligibility before reading results. Create mutually interpretable groups, document overlap with retargeting and customer audiences, and choose assignment at the account or person level appropriate to the platform and action. Keep creative variants identical except for the message element under test. Balance placements, frequency, budget, bid strategy, landing page, and timing where possible.

Set minimum exposure and outcome windows, but do not copy a universal sample benchmark. Use baseline event rates and the smallest practical effect to plan feasibility. Establish stop conditions for privacy, platform policy, tracking failure, extreme imbalance, or harmful creative. Creative and legal owners approve variants before launch; an analyst owns the readout.

Verify the actual experiment eligibility, assignment unit, split, exposure, outcomes and stopping mechanics for the chosen destination. A platform report describes that implementation; it does not establish that every cohort or creative comparison is causal or available.

Test seven message hypotheses

1. Problem framing by research topic

Test whether a problem statement tied to a broad validated topic outperforms generic category language for the eligible cohort. Keep the offer and format constant. Limitation: a topic can contain several buyer jobs, so one message may misclassify the audience.

2. Stage-appropriate offer depth

Compare an educational guide, diagnostic, calculator, or demo invitation according to a documented stage hypothesis. Measure qualified next steps, not only clicks. Failure mode: using “late-stage” intent to push an aggressive demo can reduce trust if the signal is wrong.

3. Proof type by risk profile

Test operational proof, methodology, customer evidence, security detail, or implementation clarity against the same core claim. Use only verified proof. Limitation: different proof assets may vary in quality and format, confounding the intended message test.

4. Outcome versus process message

Compare a buyer outcome with a concrete description of how the work happens. This can show whether the cohort needs aspiration or implementation confidence. Failure mode: an unsupported outcome claim may attract clicks while creating regulatory and trust risk.

5. Category versus alternative framing

Test a category explanation against a fit-based comparison to a status quo, manual approach, or adjacent solution. Keep competitor names out unless the campaign has a verified and approved comparison. Limitation: alternative framing can change audience intent rather than merely creative response.

6. Creative format within one message

Hold the claim and offer constant while comparing a simple static visual, document, short motion asset, or text-led execution. Failure mode: placements may distribute formats differently, so delivery imbalance can masquerade as creative preference.

7. Intent-cohort versus broad-cohort interaction

Run the same creative variants in an eligible intent cohort and a comparable broad or fit-only cohort to test whether the message effect depends on timing evidence. Limitation: cohort size, auction conditions, and identity match may differ, requiring careful interaction analysis rather than two raw CPA comparisons.

Compare broad, cohort and dynamic tests

Compare actual delivery and uncertainty for each design. A broad audience may provide more observations but can conceal differences in buyer questions. A narrow cohort can test a specific hypothesis while limiting sample size and usable exposure. Dynamic delivery adds separately verified routing, QA, fallback and interpretation work. None is a universal winner.

Choose a broad, cohort or bounded dynamic test against the actual decision, eligible volume and implementation constraints. Preserve a generic fallback where context is missing. Verify current permissions and QA rather than assuming a static or personalized design is inherently safer or better.

Budget the full creative test

Budget includes intent data, identity matching, platform media, creative strategy, production, trafficking, experiment configuration, analytics, CRM joins, monitoring, and readout. Estimate eligible accounts, matchable accounts, reachable members, exposure per cell, event rate, and outcome lag. If each variant receives too little exposure, fewer variants or a broader cohort may produce more useful evidence.

Price the test around the decision and effort, not a promised result. Separate setup from recurring cohort refresh and reporting. Include the cost of a null result and the value of avoiding a poor rollout. Do not publish universal creative testing benchmarks; audience value, platform, event, and sales cycle change the economics.

Report outcomes and uncertainty

Start with assignment and delivery: eligible size, match rate, reach, exposure, frequency, spend, placement, and imbalance. Then report primary and secondary outcomes by assigned group. Include counts, rates, absolute difference, relative difference only when useful, uncertainty interval, missing joins, and the pre-registered decision threshold.

Attributed conversions and CTR differences are not automatically incremental lift. A lift design requires a valid control, assignment, exposure definition, and contamination assessment. Qualified pipeline can be directionally useful but delayed and affected by sales execution. Report “no clear difference” when supported. Practical value asks whether the observed effect is large enough to justify production, governance, and media complexity.

Check actual sample feasibility

Good candidates have distinct buyer jobs, sufficient eligible and matchable volume, stable creative operations, and a downstream event that arrives within a useful window. An established B2B SaaS team may test implementation versus comparison cohorts. An agency may test report offers by topic across several clients only if each client remains isolated and separately consented.

A test is a poor fit when intent evidence is stale, the audience is tiny, the platform cannot prevent overlap, or the team changes bids and landing pages simultaneously. Do a feasibility calculation before production. If the expected event count is too small, test a larger message family in a fit-only audience or use qualitative research instead.

Keep creative useful without surveillance

Create a message map with five columns: eligible topic or stage, likely buyer job, helpful claim, appropriate proof, and safe offer. Add an explicit “do not say” column. The creative should be understandable to anyone in the target market and should never state or imply that the advertiser observed an individual’s browsing.

Keep actual source, recipient, purpose, jurisdiction, channel, destination and permissions under named review. A topic record is not automatically first-party customer information or an eligible ad audience. Preserve necessary fields and resolve missing rights before activation.

Control overlap and repeated testing

Maintain an audience matrix across campaigns and exclude treatment groups from competing paths where the design requires it. Monitor cross-device and account-level contamination, creative fatigue, frequency, placement mix, conversion lag, and identity-match changes. Do not stop when a preferred variant briefly leads. Follow the decision window unless a pre-registered safety or tracking condition triggers.

Minimize data and retain cohort membership only as long as needed. Avoid sensitive topics and protected inferences. Human approval is required before customer-data uploads, audience activation, material spend changes, creative publication, and client-facing causal claims. Keep an audit trail and rollback path for every automated recommendation.

Report evidence before recommendations

A useful agency report states the decision, hypothesis, cohort definition, exclusions, assignment, variants, exposure, outcome window, primary measure, uncertainty, contamination, cost, result, and next action. It shows delivery before performance and limitations before a recommendation. Package feasibility, creative production, media operation, measurement, and learning as separate workstreams.

AIMEE can prepare a review brief through configured connected tools. Keep proposed actions separate from actual approvals and destination acknowledgements. Confirm the implementation and permissions before execution.

Use a recurring experiment register so the client can see decisions accumulate. Each row should contain the buyer question, message element, cohort rule, exclusion rule, assignment unit, variants, owner, start condition, stop condition, primary outcome, practical threshold, result, limitation, and next decision. Link the final creative and exact audience definition. This prevents teams from retesting the same idea under a new campaign name or cherry-picking an attractive secondary metric.

The service cadence can include a feasibility review, creative approval, launch QA, exposure audit, outcome maturation check, and decision meeting. If delivery is badly imbalanced, pause interpretation before asking design for another variant. If the cohort is underpowered, consolidate it or change the research question. If a result meets the predeclared evidence rule but is too small to justify operational complexity, document the learning and keep the simpler creative.

Client dashboards should never hide counts behind percentages. Show assigned accounts or people, reached members, primary events, qualified outcomes, missing joins, and exclusions. Put the conclusion beside the evidence standard: adopt, reject, iterate, or inconclusive. That record preserves which message works for which buyer job and what evidence would change the recommendation.