Set the test before launch
Use the run and decision weeks to carry out one test as planned, then make a deliberate choice from what the test produced. Treat the hundred-dollar budget as a planning scenario, not a promise of traffic, conversions or income. The purpose is to learn how a specific setup behaves while keeping spending and decisions under control.
- [ ] Write down the test question in a sentence. Make it specific enough that the outcome could inform a decision.
- [ ] Record the single variable being tested, such as a creative, audience, landing page or campaign setting.
- [ ] Note what will stay unchanged. This makes it easier to understand which part of the setup the test actually examined.
- [ ] Set the budget limit, run period and decision point before launch. Use the amount and timing that fit the plan; do not treat any example budget as a recommended spend.
- [ ] Decide what evidence would support continuing, changing the tested variable or stopping.
A test plan is useful only if it matches the campaign that goes live. Keep a copy of the approved setup and compare it with the live version before spending begins. If the campaign differs from the plan, record the difference and decide whether it is still a valid test. Avoid quietly changing the question after seeing the first results.
Check the account and campaign setup
The account is both a product and a working tool. Before launch, confirm that it is under your control and appropriate for the platform and task. A purchased account should be used within the platform’s terms, with accurate information and ordinary account-security practices. Do not treat account ownership as a reason to bypass verification, conceal activity or work around platform checks.
- [ ] Confirm that you can access the relevant advertising account and the connected assets needed for the campaign.
- [ ] Check that the campaign, billing setup, tracking and destination are the ones intended for the test.
- [ ] Review the ad and landing page for consistency: the destination should match what the ad presents.
- [ ] Confirm that tracking records the events you intend to evaluate, and note any limits or uncertainty in that measurement.
- [ ] Save the launch configuration in the diary, including the platform, account type, campaign objective and tested variable.
Keep the setup check practical. The goal is not to search for a way around a platform’s process, but to make sure the campaign can run as intended and that you can observe it. If access, billing, tracking or policy status is unclear, pause and resolve the issue through the platform’s normal processes before treating the campaign as a test.
Launch once, then verify
When the plan and setup agree, launch the single test. A clean launch makes the later decision more useful: the diary can connect the recorded activity to the setup that was actually running. If the campaign fails to deliver, sends people to the wrong destination or records events unexpectedly, that is a setup issue to document, not evidence about the creative or audience.
- [ ] Check that the campaign is active in the intended account and uses the planned settings.
- [ ] Confirm that the ad displays and that its destination opens correctly.
- [ ] Check that the relevant tracking appears to be functioning, without making unplanned campaign edits just to create activity.
- [ ] Record the launch time and any differences between the planned and live setup.
- [ ] Note any review, delivery or account issue and follow the platform’s standard support or troubleshooting route.
A verification check is not an invitation to optimize immediately. Distinguish between a fault that prevents the planned test from running and an ordinary early fluctuation in delivery or response. Fix a genuine fault carefully, document the change and consider whether the original test remains interpretable. If the change alters the test, do not pretend that the original plan was followed unchanged.
Keep a daily diary
Record spending and results each day during the run. The diary should be factual and consistent, not a stream of guesses about what the campaign “means.” Use the same definitions and reporting view throughout where possible, and note when a figure is missing, delayed or affected by a tracking problem.
- [ ] Enter spend and the relevant results shown by the platform.
- [ ] Record the date of the observation and the reporting window or view used.
- [ ] Note delivery status, account notices, tracking concerns or interruptions.
- [ ] Mark any change made, who made it and why.
- [ ] Separate observed facts from interpretations. For example, write “delivery paused” as an observation and keep possible causes in a separate note.
The diary is especially valuable when the result is inconclusive. It can show whether the campaign ran as planned, whether spend was distributed as expected within the setup, and whether a technical issue affected measurement. Avoid filling gaps with estimates presented as facts. If platform reporting later changes or attribution remains uncertain, preserve that uncertainty in the record rather than smoothing it away.
Resist changing several things at once
During the run, keep the test stable unless there is a clear operational, policy or safety reason to intervene. Changing the audience, creative, bid approach and landing page together may make the campaign different, but it also makes it harder to tell which change mattered. The purpose here is not to keep every setting frozen at any cost; it is to make each intervention explicit and understandable.
- [ ] Before editing, ask whether the issue prevents the planned test from operating.
- [ ] If an edit is necessary, change only what is needed to address the issue.
- [ ] Write down the setting before and after the edit, with the reason.
- [ ] Reassess whether the test can still answer its original question.
- [ ] If multiple changes are unavoidable, label the result as a changed or interrupted test rather than a clean comparison.
Do not chase every daily movement by adjusting the campaign. Daily notes are for observing the run, not for turning each new data point into a new strategy. If the campaign is operating within the planned setup, continue collecting evidence until the pre-set decision point. If it is not, address the specific problem and document what that means for interpretation.
Wait for the pre-set decision point
Judge the test at the point chosen before launch, using the evidence and criteria written in the plan. Do not move the decision point simply because early results look encouraging or disappointing. Nor should you assume that a particular amount of spend or elapsed time automatically makes data conclusive. The scenario and the platform’s reporting behavior determine what can reasonably be assessed.
- [ ] Check whether the campaign reached the planned decision point.
- [ ] Review the diary for missing observations, interruptions, edits and measurement issues.
- [ ] Compare the evidence with the criteria set before launch.
- [ ] State what the evidence supports and what it does not establish.
- [ ] If the planned data point was not reached, record why and decide whether the test is incomplete.
This is not a general exercise in interpreting every early signal. It is a discipline for honoring the test plan and avoiding a decision based on whichever moment feels most persuasive. If the evidence is too limited or compromised, say so directly. An inconclusive result is different from proof that a setup works or fails.
Choose one outcome
At the end of the planned period, choose one of three outcomes: continue, change one variable or stop. Make the choice from the test question, the recorded evidence and the campaign’s practical constraints. A decision does not need to imply certainty; it needs to be clear enough to guide the next action.
- [ ] Continue when the evidence supports running the same setup further and the plan still makes sense. Record what additional question the continued run is meant to answer.
- [ ] Change one variable when the test suggests a specific next adjustment. Name that variable and keep the rest of the setup stable where practical.
- [ ] Stop when the evidence, risk, compliance concern or remaining budget makes continuation inappropriate. Record the reason without dressing it up as a broader conclusion than the test supports.
Do not combine outcomes into a vague decision such as “keep going but change a few things.” If you continue, preserve the setup. If you change one variable, define the next test before editing. If you stop, pause the campaign through the normal platform controls and record the final status. These are decisions about this test, not universal verdicts on a platform, account type or traffic source.
Write down what was learned
Close the diary with a short review, whatever outcome you chose. The point is to retain useful knowledge for the next plan, including limits and mistakes. A stopped test can still clarify that a setup was unsuitable, that measurement was unreliable or that the original question was not answered. A continued test does not prove that the next stage will produce the same pattern.
- [ ] Restate the original test question and the setup that actually ran.
- [ ] Summarize the recorded spend, results and any interruptions without adding unsupported estimates.
- [ ] Note the decision and the evidence behind it.
- [ ] List what remains unknown, including tracking or attribution uncertainty.
- [ ] Write the next action, or state that no further test is planned.
Keep the conclusion narrow. For example, describe what happened under the recorded conditions rather than claiming that a creative, audience or account will always perform a certain way. Save the notes with the campaign records so the next test can build on them. The run and decision weeks are complete when the outcome and the lesson are both written down—not when a particular result appears.