Lessons From Spending $3M on Ecommerce Facebook Ads


Running Facebook ads can be expensive. Some even think that the platform is now saturated. Despite that, Facebook remains extremely popular among marketers. Even with YouTube and TikTok grabbing a larger piece of the pie, there’s no Facebook exodus on the horizon.

In this post, we share the most important takeaways we learned from our partner agency – Sales Genomics.

In 2019, their total Facebook ad spend was over $3M, which is enough to validate their experience.

Whether you’re scaling your own Facebook ads or have someone else helping you, these lessons will keep you up to date with the competitive Facebook advertising market.

Lesson 1: Creatives are King

Even with a good product, you can still screw up the targeting, optimization and even bidding. But if you have really good creatives, you’re pretty much safe.

Sales Genomics invest a lot in the creatives when doing Facebook ads. In fact, their creative team is the largest department in their company. They continuously look for creative experts with unique ideas that stand out, stop the scroll, grab attention and generate desire.

Regardless if you’re just starting your FB ads or not, make sure you invest a lot in your creative designs. Constantly test new creatives, either weekly or even more often.

Don’t compare apples to apples. Test vastly different creative angles, testimonials, memes, influencer endorsements, user-generated content.

Video ads and Instagram Stories are NOT the new norm

Don’t get me wrong, they’re great for some products and businesses. They definitely work, but having worked with lots of clients, Sales Genomics saw that:

1. Stories work great with stores that have a small catalog or where there’s a strong best seller product. (Surprisingly, we’ve seen stories work great even in 50+ demographics! Never assume…)

2. It’s better to remain agnostic about creative styles and test it all. Test images, carousels, video collections, stories, and canvas instead of jumping to a conclusion about what will be the winner.

3. Facebook expects you are hoping to get better results from video AND when a user is watching your video, they could alternatively scroll through 10+ image ads/posts. So, Facebook will charge you more for impressions resulting in higher CPMs. Images seem to get better results than videos for most of the accounts that Sales Genomics manages.

4. Videos are not great when your ad strategy relies on dynamic ads and catalog sales, because they take away from the benefits of dynamic personalized product recommendations. Collection ads are a good way to get the best of both worlds, but also not always effective.

Lesson 2: Look at the whole funnel. Don’t get fixated on CPA (Cost Per Acquisition)

This one is pretty basic and cliche, but nevertheless important.

When you’re running ads, it’s easy to get fixated on the CPA. More often than not, you’ll find yourself worrying about how much money you spent without considering the bigger picture.

But one thing you need to remember. When you scale, it becomes harder and harder for Facebook’s algorithm to find ready-to-buy audiences. This ultimately leads to a higher Cost Per Acquisition (CPA).

Does that mean your Facebook ads aren’t working? Not necessarily. Instead of checking CPA daily, divert your attention to Upsells, Lifetime Value, Returning Customer Rate, and Referrals.

PRO TIP: If you can, build a subscription into your product to exponentially and predictably increase your revenue every month. Don’t let increasing CPA on the frontend stop you from scaling your profits big time!

Lesson 3: Let the algorithm do its work

Facebook is becoming smarter and smarter with Artificial Intelligence (AI). New Campaign Budget Optimisation feature works well, be it for testing, scaling or remarketing (although to a smaller extent here).

Although it hurts the ego of some ad buyers, AI is getting better than any human in making decisions. Learn to work with it and give it more control: broader audiences, longer decision times, larger budgets, and more auto placements.

Use AI to your advantage, so you can focus on the creative, on-site experience and lifetime value of your customers.

Let the algorithms do the work!

Lesson 4: No one-size-fits-all scaling method. Test it all.

There’s no single magic bullet scaling strategy.

From spending $3M on Facebook Ads, Sales Genomics found that you can never rely on a single strategy when scaling.

Sometimes “lookalikes” work better, sometimes it’s broad interests or open targeting. In some cases, it’s manual bids, but quite often automatic.

PPE objective used to work well for Sales Genomics, but now they’re using 90% conversions.

Test it all for yourself. Don’t be rigid, don’t be a know-it-all. Things change, things evolve. Change is awesome.

Lesson 5: Learn to navigate data attribution and embrace some data uncertainty

This is somehow related to Lesson #2. Many business owners look too closely at their Ad Manager dashboard and are convinced they’re seeing the whole picture. If the business owner is not fixated on the CPA, they’ll fixate on what they see on their dashboard.

100% Measurable campaign performance is a myth.

Some of Sales Genomics’ best clients, who have scaled multiple businesses, understand that as the ad budget scales there is always a halo effect of additional sales due to buyers using ad blockers, referrals, cross-device conversions, increased brand awareness and more.

Always triangulate data between Facebook Ads Manager, Google Analytics and store reports and embrace some degree of uncertainty.

Lesson 6: Don’t rely on Facebook advertising courses too much

If you have ever watched a Facebook ads course, you will have a weirdly skewed perception of how it works.

A disproportionate amount of the material you’ll find in most FB ad courses is dedicated to the technicalities of the platform, such as bidding, scaling strategies, Campaign Budget Optimization (CBO) tricks, hidden targeting options and other stuff that makes typical media buying analytical geeks salivate.

It’s easy to think that Facebook campaigns depend so much on the technical stuff. But in reality, here’s the order of importance Sales Genomics have seen across the board of ecommerce niches:

1. Brand – It’s an unpopular thing to say among bragging marketers, but the largest determinant of campaign success is how successful a brand already is. It is not to say that new brands have no chance, but brand building clearly pays dividends.

2. Product – What’s your products’ utility, uniqueness, emotional and visual appeal? Marketers rarely have control over it, but it’s important to stay in close cooperation with your agency/ads partner on the evolution of the product to provide valuable feedback from on-going campaigns.

3. Offer – How attractive is your offering compared to it’s perceived value.

4. Funnel and user experience – Using conversion rate optimization experts and UX scientists to get your site before launch and every few weeks, allows you to collect feedback.  Use Hotjar polls, heat maps, recordings, user surveys, monitor ad feedback, comments and periodically review competitors’ movements. Ideally, you can run A/B tests on the website too. With this data, you can impact conversion rates, Average Order Values, and Returning Customer Rates and make sure the advertising strategy is always congruent with the funnel.

5. Targeting – Simple stuff works well when starting out. Don’t over complicate it.

Distilling a year’s worth of tests into a single post wasn’t easy but we hope it was helpful. Comment below if you have any questions or if you found your experience to be similar.

More to read



TikTok Marketing: How to use TikTok for Your Ecommerce Brand

June 23, 2022 - Holly Cardew



Automatically Add Alt Text to Your Shopify Listings

June 13, 2022 - Holly Cardew



Shopify Image Sizes for Ecommerce Store

December 13, 2021 - Holly Cardew


Proudly made remotely from