214: The Right Way To Run Facebook Ads For Physical Products With Andrew Foxwell

Podcast: Download (Duration: 54:07 — 74.6MB)

Video ads generally do not work for prospecting physical products on Facebook, despite everything the information-product marketing world says about them. Static images and carousel ads drive cold traffic to ecommerce sites more reliably, and Andrew Foxwell has tested this across accounts at serious scale.

Andrew has run Facebook ads since 2008, starting with congressional campaigns before Facebook Pages even existed, then leading social at 3Q Digital. He and his wife Gracie run Foxwell Digital and have helped over 250 businesses manage more than $10 million in Facebook ad spend, working almost exclusively with physical product ecommerce.

That specialization matters, since most Facebook consultants work with digital products where margins are enormous and mistakes are cheap. This episode covers the average order value threshold where Facebook stops working, the social proofing technique that turns your ad into its own landing page, what belongs in each funnel stage, and why broad audiences now beat narrow targeting.

Key takeaways

• Video ads underperform static images for cold prospecting on physical products, despite conventional wisdom to the contrary.
• Sub-$50 average order value is hard to make work at scale unless the product has genuine mass market appeal.
• Social proof an ad by running it as page post engagement for 48 to 72 hours, then use that post ID in a conversion campaign so the comments carry over.
• Target break-even on prospecting over a 30-day window and make your profit on remarketing.
• Broad audiences with no targeting now work well once your pixel has seen roughly $10,000 to $20,000 in spend.
• Optimize for add-to-cart rather than purchase on products over $100, so the ad set reaches Facebook’s 50-conversions-per-week learning threshold.
• Broad match dynamic product ads need 30 to 45 days before you judge them.
• Build a loyalty audience for past purchasers. Andrew sees it missing in nine out of ten account audits.

What average order value do you need for Facebook ads to work?

Products under $50 are difficult to scale profitably on Facebook and Instagram unless they have broad mass market appeal. Pura Vida Bracelets and Blenders Eyewear work at lower price points because everyone wears sunglasses and bracelets have universal appeal.

Below that threshold the math gets tight fast. On a $25 average order value with a $10 to $15 acquisition cost target, you have almost no room for a multi-touch journey.

The workaround at low price points is capturing emails first rather than pushing for the immediate sale. That is a far more reasonable ask from cold traffic, and you convince them over time through email.

Does your product need to solve a problem for Facebook ads to work?

Your product does not need to solve a problem, and it does need a compelling pitch or story. There are a hundred thousand watch companies online, and the ones that work on Facebook have a hook beyond the product itself.

Andrew’s examples make the point concrete. Flex Watches ties each watch to a cause, so buyers connect to something beyond the object. Rareform turns old billboards into bags, which is an origin story people remember.

Look for a story before the product exists or after the purchase happens. How the product is made, or what gets donated when someone buys, both give you material a plain product photo cannot.

The critical addition is having several pitches rather than one. Andrew worked with a handheld air purifier where the first value proposition they were confident about was not the one the audience responded to.

Plan for a testing period. Committing to six months of trying different hooks and value propositions is a realistic expectation.

Why do video ads fail for physical product prospecting?

Video fails at the top of funnel for physical products because cold viewers scroll past before watching enough of it to matter. Andrew reviews roughly 30 accounts a week and has seen this pattern repeatedly at significant scale.

The comparison he draws is with information products, where video genuinely does work. An inspirational video makes someone want to join a movement, and a video about a physical object mostly gets scrolled.

If someone asked him to make an account profitable within 30 days, he would not suggest video. Static images in the right formats get there faster.

Video belongs in the middle and lower funnel, where you are furthering a brand proposition to people who already know you. It also builds video view audiences you can remarket to, which is a legitimate reason to run it even without immediate return.

Animated GIFs are the useful middle ground. Adding movement to a product image stops the scroll without requiring viewers to commit to watching anything.

What ad formats should you start with for physical products?

Start with three to five square 1000×1000 images for photo posts and carousel ads, plus three to five 1200×628 images for link posts. Those two ad units are the most proven drivers of prospecting traffic to ecommerce sites.

The first creative question is contrast. Set the product against a background that makes it stand out in a crowded feed.

The second is movement. Convert those images into GIFs or short video files so something in the frame changes as people scroll.

Native-looking content produced on an iPhone often outperforms polished studio work. Brands without budget for a second brand video should lean into that rather than reusing an expensive video that is not converting.

What is Facebook ad social proofing and how does it work?

Social proofing means running an ad as a page post engagement campaign first to accumulate likes and comments, then using that same post ID inside a conversion campaign so the engagement carries over. The ad effectively becomes its own landing page.

Andrew runs the engagement phase to a wide domestic audience matching his core demographic, excluding fans and all past website visitors. Beyond age and gender he adds essentially no targeting, letting Facebook find people likely to engage.

That phase runs 48 to 72 hours. He looks for a click-through rate north of 2% and a relevance score of seven or eight.

The payoff is what happens in the comments. People considering a purchase read comments looking for validation, and buyers commenting that they received theirs validates the product before anyone reaches your site.

Running the engagement phase in cheap foreign countries is a common variant Andrew has tested and does not recommend. The engagement quality is poor.

Note that a high relevance score does not guarantee conversion. Andrew has run well-socially-proofed posts in conversion campaigns that attracted excellent traffic and never sold, which is still acceptable if your remarketing return covers it.

Should you use broad audiences or narrow targeting on Facebook?

Broad audiences with minimal targeting now outperform tightly specified ones, provided your pixel has enough history. The shift comes from Facebook’s auction and algorithm maturing to the point where its data beats your guesses.

The threshold Andrew cites is roughly $10,000 to $20,000 in spend through the pixel. At that point Facebook understands the difference between people who view content, add to cart, and actually purchase for your specific store.

Below that threshold you still need to supply targeting. Build lookalikes from purchasers and email lists, add interest-based targeting built from audience insights, and add behavior-based targeting such as credit card usage patterns.

Separating interests rather than grouping them is how you find scale early. Testing whether fans of one personality convert better than fans of another gives you signal you cannot get from a blended audience.

What lookalike audiences do high-spend accounts actually use?

The common pattern across accounts spending $20,000 to $200,000 a month is a 1% lookalike of highest lifetime value customers, a second 1% lookalike built from a different engagement signal, a 3% to 5% lookalike for scale, and a wide untargeted audience.

That second lookalike source varies by business. It might be people who engaged on Instagram, two-time page viewers in the last 60 days, or add-to-cart events, depending on which signal predicts purchase for you.

The fifth component is a broad match dynamic product ad campaign under product catalog sales. It lets Facebook decide which catalog products to show which people based on prior purchase patterns.

How long do broad match dynamic product ads take to work?

Broad match dynamic product ads typically take 30 to 45 days before showing a return, so patience is the requirement. Andrew describes a workshop attendee whose first two weeks produced three sales, and by early the following month the campaign was running at two and a half to three times return on ad spend.

The latency is built into how the system learns. Judging these campaigns on a normal 48 to 72 hour testing window guarantees you kill them prematurely.

What return should you expect from prospecting campaigns?

Aim to break even on prospecting over a 30-day window, then make your profit on remarketing. Andrew acknowledges that plenty of store owners find that target too modest, and it reflects what taking someone from never having heard of you to buying actually costs.

The combined math is what matters. Break even on prospecting plus a 3x return on remarketing produces a blended return around two to two and a half.

Your cost structure sets your real number. One client selling witty t-shirts needed at least 1.5x over 30 days on prospecting given their cost of goods, so that became the target.

Higher average order value products make this considerably easier. Spending $200 to acquire one $200 sale is a much better position than doing the same on a $25 product.

Should you optimize for purchases or add to cart?

Optimize for add to cart rather than purchase on products above roughly $100, because you need volume to feed Facebook’s learning phase. Facebook wants around 50 conversions per week at the ad set level, and a high-priced product on a cold audience will not produce that.

Two or three purchases over seven days is a weak signal. Switching the optimization event to add to cart gets closer to the threshold and stabilizes performance.

Should you offer a discount in top of funnel Facebook ads?

A 10% to 15% discount on cold traffic is the reliable default, paired with visually interesting creative. The incentive gets people to click even when redemption rates are modest.

The exception is brands where the message itself is the hook. Andrew has a client selling handmade organic baby clothing whose cold ads carry no discount at all and instead talk about slowing down and caring for your child.

That approach targets a specific mindset rather than a price-sensitive shopper. It also reads more like something a friend shared, which feels more native in the feed.

Where you send the click depends on complexity. Products needing explanation go to a page with context, and a watch or sunglasses go straight to a best sellers page so people start browsing.

What belongs in the middle of a Facebook ads funnel?

The middle of the funnel is where you hit engaged non-buyers with several different objectives, since people respond within different objective pools. Andrew runs page post engagement, lead ads for email signup, conversion campaigns, and website click campaigns at the same audiences.

His current example is repurposing a client’s Instagram content as a page post engagement ad, refreshed every two days, targeted at a 21-day top 25% website custom audience who have not purchased or added to cart. It carries no discount and mostly announces what is back in stock.

Objective diversity is the point. Some people engage with posts, some watch videos, some click through, and running only one objective misses whole segments.

The middle of the funnel should still roughly pay for itself, typically at 1.5x to 2x. Andrew’s shift over time has been away from the simple lookalike-offer-then-dynamic-product-ad model toward storytelling content that builds an actual relationship.

Discounting in the middle is reasonable on lower-priced products. On higher average order value items, brand content often works better than deeper discounts because you are building toward advocacy rather than one transaction.

How do you structure bottom of funnel remarketing?

Run dynamic product ads across multiple time windows, typically 90 day, 30 day, 7 day, and 1 day, with a different creative format in each. A collections ad works well at 90 days because you want people browsing the catalog again.

Shorter windows get more specific. A single product image suits the one-day window since the person likely viewed one item, and a carousel suits seven days.

This is where the discounting belongs. A 20% offer at the bottom of the funnel is normal, and Andrew’s challenge to clients who want to offer 10% is whether that discount would actually make them buy.

Run a 30-day website custom audience remarketing campaign alongside the dynamic product ads with fresh creative, watching frequency stays under three. A review-based dynamic product ad using something like Yotpo adds another layer of validation.

Why should you build a Facebook loyalty audience?

A loyalty audience of past purchasers is the most commonly missing piece Andrew sees, absent in nine out of ten accounts he audits. Sending new product announcements to previous customers regularly returns two to three times ad spend.

The setup excludes recent purchasers, typically the last 30 days, so you are not annoying people mid-transaction. The message is what is new rather than a hard pitch.

One account spending roughly a quarter million dollars in a week runs identical creative to prospecting and to its loyalty audience. Loyalty customers post photos of themselves in the comments because the ad asks them to, which turns the prospecting ad into social proof.

The high-effort version is worth knowing about even if you never do it. One client runs a dayparted Messenger campaign to past purchasers from 9am to 3pm local time simply asking how things are going, with a human answering.

Another client surveyed their Messenger audience about which products to launch next and got 5,000 responses. Treating impressions as people rather than numbers is what makes that possible.

Where do Messenger ads fit in an ecommerce funnel?

Messenger ads work best for lead qualification, contests, and loyalty conversations rather than as a primary sales driver. Andrew uses them mostly in B2B contexts, launching at a 30-day website custom audience and inviting people to start a conversation.

Lead costs are low and open rates are extremely high, which is the genuine advantage. The ecommerce sales use cases were less proven at the time of this conversation.

One case that did work is a deep discount offer. A CrossFit brand ran a $40 product at $10 through a Messenger subscribe-and-buy flow and saw two and a half times return on ad spend on prospecting while also building their broadcast list.

Andrew’s caution is that Messenger risks becoming what video became: a tactic the industry over-promotes because it works well for information products and lead generation rather than for physical product direct response. Messenger marketing has since changed substantially as Facebook tightened its policies, and the underlying caution about adopting tactics from the info-product world still applies.

How long should you let a Facebook ad run before judging it?

Give an ad 48 to 72 hours before deciding whether it is working. That window applies to most creative and audience tests.

The exception is a low cost per click combined with a high relevance score. Andrew will let those run longer even without a sale, out of curiosity about what develops.

Broad match dynamic product ads are the other exception and need 30 to 45 days. Applying the standard window to them will kill campaigns that were about to work.

Frequently asked questions

Do video ads work for selling physical products on Facebook?

Not for cold prospecting in most cases. Static images and carousels drive traffic more reliably, and video works better in the middle and lower funnel where people already know your brand.

What is the minimum average order value for Facebook ads?

Around $50 is where it gets difficult at scale, unless the product has broad mass market appeal. Below that, capture emails from cold traffic instead of pushing for an immediate sale.

What is social proofing a Facebook ad?

Running an ad as a page post engagement campaign for 48 to 72 hours to accumulate likes and comments, then using that post ID in a conversion campaign so the engagement carries over and validates the product.

Should you optimize Facebook ads for purchase or add to cart?

Optimize for add to cart on products over roughly $100, so the ad set gets closer to Facebook’s 50-conversions-per-week learning threshold. Low-priced products can usually optimize directly for purchase.

How much do you need to spend before broad targeting works?

Roughly $10,000 to $20,000 through the pixel. Below that, Facebook lacks enough data about your specific customers and you should supply lookalike, interest, and behavior targeting.

What return on ad spend should prospecting campaigns hit?

Break even over 30 days is a realistic target, with profit coming from remarketing. Your cost of goods determines your specific number.

How long do dynamic product ads take to become profitable?

Broad match dynamic product ads typically need 30 to 45 days. Standard remarketing dynamic product ads work much faster since the audience is already warm.

What is a loyalty audience in Facebook ads?

An audience of past purchasers, usually excluding the most recent 30 days, targeted with new product announcements and special offers. It commonly returns two to three times ad spend and is missing from most accounts.