Skip to content
October 28, 2026

Evidence Review: Turn Campaign Ideas into Testable Decisions

Evidence Review: Turn Campaign Ideas into Testable DecisionsAIM ENGINECREATE · AUTOMATE · AMPLIFYCAMPAIGN STRATEGYEvidence Review: Turn CampaignIdeas into Testable DecisionsEVIDENCE · IDEAS · RESULTSAIM ENGINE JOURNAL59

Before generating another campaign idea, write down the decision it needs to support. Then record the evidence, its limitations and the result that would justify taking action. This gives an alternative a clear purpose without presenting it as a proven strategy.

AIM Engine’s evidence review evidence review method describes seven lenses for developing alternatives after useful evidence exists. It also distinguishes generating candidates from measuring whether they work. The practical question is how to move between those stages.

One proposed approach is an Evidence-to-Test Card: a short review record connecting a business decision, supporting observations, a proposed alternative and a test. The workflow below is a recommendation, not a verified AIM feature or an approach with demonstrated results.

Start with the decision, not the idea

Define what someone will decide after reviewing the test. “Improve the campaign” is too broad. “Decide whether to test a shorter enquiry form more widely” identifies a specific choice.

Before ideation, record the current baseline, the primary measure and the threshold that would support proceeding. Also specify what would count as an inconclusive result or a reason to stop. Where a baseline is missing, collecting it may be the next useful action rather than launching a comparison.

Choose thresholds appropriate to the decision, available traffic, commercial constraints and measurement quality. The supplied source does not establish universal thresholds or a testing duration.

Build an Evidence-to-Test Card

A document or shared worksheet is enough to prototype this approach. There is no need to assume an integration exists. Include these fields:

  • Decision and owner: What choice will this inform, and who is responsible for reviewing it?
  • Evidence source and date: Where did the information come from, and when was it collected?
  • Observation: What does the evidence directly show?
  • Limitations and gaps: What is missing, uncertain or restricted to a particular context?
  • evidence review lens and alternative: Which lens prompted the proposed change?
  • Hypothesis: Why might the alternative help, and which assumptions does that depend on?
  • Test and decision rule: What will be compared, measured and reviewed before a decision?
  • Status: Is the card a draft, approved for testing, inconclusive or supported within the tested context?

Keep observations separate from explanations. “Visitors left the form before submitting” is different from “the form was too long.” The latter is a possible explanation requiring investigation, not an observation established by abandonment alone.

Use the seven lenses to create distinct alternatives

Apply whichever lenses are relevant; completing all seven is not a requirement. The aim is to identify useful alternatives, not fill a content quota.

  • Substitute: Replace a weak input, message, channel or measure. For example, propose a more decision-relevant measure instead of relying only on clicks.
  • Combine: Bring complementary evidence together. A campaign evaluation and an approved knowledge reference could sit beside each other in the same review record.
  • Adapt: Borrow a pattern from another context while stating why it might transfer—and why it might not.
  • Modify: Change the timing, format, scale or sequence of an activity to create a focused test.
  • Put to another use: Consider whether an approved campaign insight could inform a sales-message review, subject to relevance and permission checks.
  • Eliminate: Propose removing a step, output or metric that appears unnecessary, while checking what purpose it currently serves.
  • Reverse or rearrange: Challenge the existing order. Define the decision and measurement requirements before developing the alternative.

Hypothetical example: reviewing an enquiry form

This example is fictional and illustrates the worksheet; it does not describe an AIM customer or a measured outcome.

A campaign manager has a dated evaluation showing that some visitors begin an enquiry form but do not submit it. The evaluation does not explain why. The business decision is whether to make selected questions optional.

Using the Eliminate lens, the manager proposes removing the requirement to answer non-essential questions. The hypothesis is that this could reduce friction. An explicit assumption is that those questions are not necessary for assessing enquiry quality.

The card proposes comparing the existing form with the alternative, measuring completed enquiries alongside an agreed quality measure. Before starting, the team must establish its baseline, define acceptable quality and choose a decision threshold. The owner must also check whether the comparison can produce interpretable evidence with the available traffic.

If submissions increase but enquiry quality falls below the agreed requirement, the change would not automatically qualify for wider adoption. If the evidence is insufficient, the card remains inconclusive rather than becoming a success story.

Select candidates without mistaking confidence for evidence

Review proposed alternatives for traceability, relevance, feasibility and testability. Can the reviewer find the supporting evidence? Is the assumption explicit? Can the team run the test responsibly? Would the result inform the stated decision?

A candidate can be worth exploring despite weak evidence, provided that uncertainty is visible. Conversely, a polished idea should not receive stronger status simply because its explanation sounds persuasive.

There are trade-offs. Detailed cards take time, while very short cards may hide important assumptions. Manual records avoid making integration a prerequisite, but still need clear ownership and consistent review. Start with the minimum fields needed for a defensible decision.

Validate the worksheet before expanding it

The supplied material does not establish whether this card improves campaign work, whether intended users understand its terminology or how it would fit existing interfaces and permissions. Those remain research gaps.

A practical pilot could ask five intended users to complete equivalent tasks using the existing guidance and the proposed card. Record completion, elapsed time, reviewer corrections and whether observations remain distinct from hypotheses. A proposed acceptance criterion is four unaided completions out of five, with no unlabelled outcome claims across completed cards. These are pilot targets, not achieved results.

Use the findings to simplify unclear fields and refine review responsibilities. The next step is a small, evidence-labelled test—not treating the worksheet, or the ideas it produces, as proven.

JOIN THE EVIDENCE

Add an example, conclusion or question.

Short, specific responses help AIM Community improve future articles and product decisions.

Your response