The Simple 3-Step Way to Test Ad Ideas (Most Brands Skip It)

The most common thing a marketer hears from a brand owner is: "can we just try a bunch of ads and see what sticks?" It's a reasonable question. It's also one of the fastest ways to burn through a budget without actually learning anything.

Ad testing at most brands goes wrong in one of two ways: either they test too many things at once (15-20 versions with nothing to compare them against), or they play it too safe (tweaking a headline when they really need to test a completely different style). The brands that grow steadily use a third approach: testing in order, one step at a time, separating "what do we learn" from "what do we do more of."

The Three Steps

Brands that do this well move through three distinct steps. Most brands squash them into one, and that's where the money gets wasted.

Step 1: Find What Basic Style Works (20% of budget)

Goal: figure out which basic style of ad connects with people. This is where you test format (a photo vs. a video vs. a slideshow), angle (leading with the problem vs. the benefit vs. proof it works), and show it to a wide, untargeted group of people.

• Run 4-6 genuinely different styles, not 15-20 small wording tweaks
• Keep who you're showing it to BROAD at this stage — don't narrow in on specific interests yet
• Spend enough on each version to get 1,000-2,000 people to actually buy — that's roughly the minimum needed before you can trust the result
• Run it for at least 7-10 days — it takes time for real buying behavior to show up
• What success looks like: 1-2 styles that beat the rest by 20% or more in how many people click or buy

Example: a supplement brand might test (1) a customer testimonial video, (2) a before/after slideshow, (3) a photo focused on ingredients, (4) a benefit-focused video, (5) a doctor-endorsed style, (6) a problem-first video. Spend $800-1,200 on each. After 7 days, drop the bottom two, and carry the top two into step 2.

Step 2: Sharpen It And Find Who It Works For (40% of budget)

Goal: take the winning styles from step 1 and sharpen them by trying different wording and messages, and testing which groups of people respond best. This is also where you test different audiences (people actively searching, people similar to past buyers, past buyers themselves).

• Take the 1-2 winning styles from step 1 and create 3-4 different wording versions of each
• Test each against a few different audiences: people actively searching, people similar to your best customers, past customers, and people who share relevant interests
• Keep the winning style, but change the story, which benefit you lead with, or the proof you show
• Spend enough to get 2,000-5,000 purchases per audience — you're now confirming who it works for, not just what style works
• What success looks like: knowing which audience gets you the best return, and which message lands best

Example: the supplement brand's winning style was the testimonial video. In step 2, they'd test: (1) a version emphasizing ingredient quality, (2) one emphasizing results, (3) one emphasizing scientific backing, (4) one emphasizing value/price — each shown to 2-3 different audiences. Total spend: 2-3x what step 1 cost.

Step 3: Spend More On What Works (40% of budget)

Goal: put real money behind what's working. Only small tweaks now — the core approach is settled.

• Use the winning style + message from step 2 as your foundation
• Increase spend gradually (25-50% a week) and watch for it losing effectiveness as you go
• Only small changes from here: new product photos, minor wording tweaks, seasonal updates
• Refresh the ad every 2-3 weeks with variations on the proven winner (same style, different scene or different testimonial), so people don't get tired of seeing it
• What success looks like: keeping your return steady while spending more

Example: the supplement brand doubles down on the testimonial style with ingredient-quality messaging. They increase spend from $5k/day to $15k/day over 3 weeks. Their return per dollar spent holds steady at $2.80. They rotate in new testimonial videos every 2 weeks so people don't get tired of it. The ad keeps working for 3-4 months before it wears out.

Why Testing In Order Beats Testing Everything At Once

Running 15 versions all at once only teaches you what worked *for that one specific split of budget*. You learn "we made $2.10 back for every dollar this week," but you don't learn *why*, or whether you can repeat it.

Testing in order teaches you the actual reasons something wins. After step 1, you know testimonials beat plain photos. After step 2, you know leading with ingredients beats leading with benefits for this audience. After step 3, you know you can spend more on this and keep making $2.80 back per dollar.

A Realistic Timeline

1. Step 1 (Find what works): Days 1-10, 20% of weekly budget ($1-2k for most brands)
2. Step 2 (Sharpen it): Days 11-28, 40% of weekly budget ($2-4k), running alongside step 1's winners
3. Step 3 (Spend more): Day 29 onward, 40% of weekly budget, based on step 2's winners
4. Month 2: Refresh the winning ad (same style, different scene or proof), keep the audiences that worked
5. Month 3: Spend more on winners, cut the weak performers, start a fresh round of step 1 testing for the new season

Common Mistakes

• Testing too many things at once in step 1 — creates noise, nothing gives you a clear answer
• Spending more on step 1's winners before confirming them in step 2 — a step 1 winner doesn't always hold up once you test it properly
• Stopping a test too early — most winners need 7-10 days to show a stable result
• Testing wording before testing style — the overall style drives most of the result, wording tweaks only add a little
• Narrowing your audience too soon in step 1 — that only teaches you about one small slice of people, not the basics that matter to everyone

“The ads that keep making money the longest at every brand we've worked with share one thing: they weren't winners on day one. They were styles that survived several rounds of testing and held up across different audiences and messages.”