Start with a customer, not an asset
A creative test asks whether a particular message matters to a particular customer. Start with the persona, the benefit they care about, and the way you will show it. Adsmith’s nano targeting explores combinations of personas, stories, and styles; Creative Exploration turns those ideas into keywords, captions, images, and videos.
Make the comparison specific
For an illustrative coffee subscription, you might compare “a better morning ritual” with “fresh beans without another errand.” Keep the offer and destination consistent where possible. Write the question before generating variations: does this audience respond more to the ritual or the convenience? This is an example, not a reported Adsmith result.
Change enough to learn something
A new audience, headline, offer, and landing page in the same test can produce a useful campaign, but it will be hard to attribute a difference to one change. For a clearer comparison, vary the idea you are investigating and keep other conditions reasonably comparable. Automated platform delivery can still change who sees each ad; a performance difference alone does not establish causation.
Judge the outcome your business needs
Clicks help explain attention. Purchases, qualified leads, acquisition cost, and customer value tell you more about whether the advertising is worthwhile. Confirm that those events are recorded correctly. Choose the primary outcome in advance and record the spend and comparison period alongside it.
Use creative breadth as a starting point
TravelJoy’s recorded Google PMax story versions with impressions increased from four to ten between June and July 2026. That shows a broader set of ideas reaching audiences; it does not itself show higher revenue. Read TravelJoy’s story for the customer’s experience of scaling the work without more headcount.
Turn the result into the next question
If the convenience idea earns better qualified leads, explore ways to make that benefit more concrete. If it attracts clicks without useful leads, investigate the promise and destination before making more variations. Keep a short record of the question, creative, period, outcome, and next decision. See how to use results in the next experiment.
Publish through supported accounts
Adsmith connects creative with campaign management through your existing Google PMax or Meta account. The advertising platform determines delivery within the campaign settings. Comparing ads receiving different audiences or delivery conditions is not the same as a controlled randomized experiment.
Use portfolio results to guide allocation
Portfolios connect campaign management and budget sharing. Optimized Rebalancing uses performance to guide where the budget goes, while the learning loop informs subsequent creative experiments. You choose your budget and review settings before publishing unless you explicitly enable auto-publish. See portfolio optimization.
Learn within your own experiments
The Adsmith Brain connects results to combinations of personas, stories, styles, and placements in your experiments. A result can suggest a related angle or an idea to retire. Do not assume that one ad’s performance proves why it worked, or that a promising idea will remain effective forever.
Judge automation by the work it improves
Gitar’s CEO reports that customer acquisition cost was cut in half through continuous experimentation and optimization. This is a customer-reported result without a specified comparison period, rather than a guaranteed outcome. Read Gitar’s experience and evaluate Adsmith against the outcomes that matter to your own business.