Growth Newsletter #064

Good creative is the biggest driver of ad success in a post-iOS 14 world.

Your ads need to resonate with your audience. No amount of sophisticated targeting or optimization tweaks will save your campaigns—that's what makes creative testing so crucial.

Here’s a look at Thesis’ creative testing methodology. Consider using it to find learnings faster and protect your core campaigns from creative flops.

Step 1. Use a simplified account structure. Three campaigns, 3-4 ad sets per campaign, with 3-6 live ads in each. The campaigns:

• Campaign 1: Creative testing. Isolate creative testing into a separate campaign to ensure live tests won't impact your core campaign performance. This also allows you to force spend to drive faster learnings and curb creative fatigue.
• Campaign 2: Prospecting. Move winners from your creative testing campaign into a separate prospecting campaign. This campaign contains only your best-performing ads.
• Campaign 3: Retargeting. Again, move winners from your testing campaign into a separate retargeting campaign. You can test the prospecting creative as is, only changing the specific offer or discount for your retargeting promotions.

Step 1a. Allocate ~20% of your budget to creative testing. Use your CPA target and this formula to calculate (approximate) starting daily spend:

• First, calculate your weekly budget by multiplying your target CPA by 50 (minimum weekly conversion threshold needed to exit the learning phase).
• Then, divide your calculated weekly budget by 7 to arrive at your daily budget.
• Here's a hypothetical example using a $35 CPA:
• $35 x 50 = $1750
• $1750 / 7 day = $250 daily budget

Step 2. Set up a creative test. Use broad targeting—it's the most scalable (and often the cheapest). Each creative concept gets its own separate ad set containing up to six variants.

Only test elements of the ad unit itself (e.g., ad formats, new images or videos, thumbnails, copy, or CTAs). Here's an example of a net-new video test:

• Create a new ad set for your video test.
• Choose one element to test. If the video is untested, start by testing different hooks or altering the first three seconds of footage.
• Launch test.

Step 3. Run creative tests for at least three days, then make a call. After about three days of running a new test, you'll typically run into one of the following scenarios:

• Results are excellent: CPA is lower than average. At least 1-2 variants show signs of traction. Start scaling spend by ~20% every three days directly in the creative testing campaign. If you have the budget, you can increase spend by 50%-100% to drive learnings even faster.
• Duplicate the winning ad into your core prospecting and retargeting campaigns.
• Results are average: Only a few purchases are generated, falling within 10%-20% of your CPA targets. Start optimizing at the ad level and turn off worst performers to give other ads more spend.
• If an ad reaches 2X your average CPA without a purchase, turn it off.
• If no winners are found in the test after 5-7 days, turn off the ad set.
• Results are bad: CPAs are high (2X normal or greater) across the board, engagement rate is poor, and there are little to no purchases generated from the new creative test.
• Follow the same optimization process from the previous bullet point.

Don’t turn off any ad set that's performing well during your creative tests. Keep it running in your testing campaign as long as results remain strong.

If you have an ecommerce site, best practice is to keep your menu (navbar) as simple as possible.

That’s not to say that every ecommerce brand should remove its informational pages—e.g., blog, About Us, FAQ—from their menu. For some, those links will help conversion, not hinder it.

A general rule of thumb for you to consider:

• If your brand has a relatively low average order value and you’re not selling a complex, ultra-specialized, or mission-driven product, move your informational pages and blog out of your menu and into your footer.
• For higher-priced, sophisticated, or story-oriented products, content marketing and info pages could increase conversion. These can factor into the consideration, intent, and evaluation stages of a buyer’s journey. Your navbar might be a good place for them.

Examples: Allbirds has a page about sustainability in its navbar. That's a core value that many shoppers will connect with and support. Judy puts FAQs in its navbar, since the decision to buy a disaster prep kit brings a lot of questions with it. On the other hand, the navbars for Nomatic and Clevertify focus on their relatively straightforward products (backpacks and baby clothes).

This boils down to a simple question to ask about any page: Will it help prospects convert or distract them from converting?

How much should you charge, and how should you charge (e.g., subscription, usage-based, flat rate)?

When establishing your business model and answering those questions, be sure to factor in friction.

Your product’s friction should align with your model’s friction. A low-friction product should have a low-friction business model. A high-friction product should typically have a higher-friction business model.

• A low-friction product is easy to use to accomplish a product’s job-to-be-done. It’s easy to get started in and stick with. Examples: TikTok, Gmail. Those have low model friction, too—they’re free. Free trials, freemium, or just plain free often align with simple products.
• A high-friction product has a more complex onboarding and use process. Experiencing full product value and forming a product habit take longer. Examples: Salesforce, Palantir. They have high model friction, too, such as higher pricing, add-ons, and variable pricing.

When product and model friction don’t align, there’s a risk that your product’s value won’t get realized, your unit economics (CAC and ARPU) won’t work, and growth potential will be stymied.

• Low product friction and high model friction: Not competitive. You’ll lose out to competitors who make it easier to pay or offer a more affordable solution. Because of the pricing barriers to entry, you’ll limit the number of users who experience your product’s job-to-be-done—and limit growth. Hypothetical example: if Instagram were to start charging a monthly fee.
• High product friction and low model friction: If you have a highly complex product, it probably can’t be learned during a free trial or freemium use. Users wouldn’t get the maximum value from your product. If you were to offer onboarding services to help free trialers get the most out of your product, your CAC would go up and could become unsustainable.

Because of the importance of aligning product and model friction, successful low-ARPU products tend to be low-friction (like social media apps), and successful high-ARPU products tend to be high-friction (like enterprise B2B SaaS).