Scaling Facebook ads like Casper, Everlane, Nurx, Roman/Hims and other ecommerce companies is the million dollar question on every marketer’s lips.
The revenue growth Facebook has seen over the past few years continues to increase, with the clear majority of dollars coming from ads.
Up until around 2012, Facebook wasn’t quite a force when it came to revenue generation. It just wasn’t Mark Zuckerberg’s calling to create a billion dollar revenue company.
But he had no choice – in order to continue toward his vision of connecting the world through meaningful, digital connection he needed to fund it. And boy did they ever.
Facebook is the colossus of social networks, finalizing 2016 with a cool $27.64 billion in earnings, leaving the likes of Twitter and LinkedIn in the dust.
Although still 3 times smaller than Google in terms of revenue, Facebook has yet to diversify income away from ads, something which Google has done over the years — in addition to Adwords, they added mobile phones, the Android operating system, YouTube and more..
Around the end of 2016 there were around 4 million advertisers on the platform. Now it has surpassed 5 million. But what does this mean for you as an advertiser? Is there still room for growth?
Right now, if I was to target everyone in the world 18+, I’d have around ONE BILLION monthly active people I can serve an ad to, across Facebook, Instagram and Messenger.
The US alone caters for between 100-200 billion users and the largest by size is India with around 400 million monthly active users, with Brazil also in the top 3 with 100-150 million active users.
For those interested in commercial gain, the US and other western economic powerhouses still represent rich pickings. However, there is a big limiting factor at play here, and it has to do with the number of available placements and the number of ads competing for those slots.
CPM (cost per thousand impressions) has been on the rise in particular over the last 2 years as Facebook has been on a drive to get more advertisers onto the platform.
At the same time, a surge of people have flocked into online business models given the low barrier to entry. Facebook ads for ecommerce, drop-shipping, online courses, multi-level marketing programs and affiliate marketing have all been feeding those online businesses.
Fast growth countries like India, Brazil, Indonesia and Mexico might be adding hundreds of millions more users onto the platform, but their economies are not yet fully developed enough for advertisers to take advantage. For example selling into countries like Brazil and India come with custom and local delivery challenges. For local businesses however it is a huge opportunity.
So how do you win with Facebook ads in countries with rising competition and rising cost on the Facebook ad platform?
The following tips are based on helping a small ecommerce company scale to over $26 million recently in worldwide sales via Facebook ads. I’ll share some of the key areas which helped us the most.
How I scaled an ecommerce company to over $26 million using Facebook Ads
When they first started out, this business I worked with spent around $3k per month (around $100 per day) on Facebook ads. Not a huge sum, but they burned through a few months of testing at this level, at best breaking even (making $3K back in sales on $3K ad spend).
Now considering the cost of goods and advertising, some may have decided to quit after taking a fair loss as a small company. But the founders were keen in ensuring they gained as many learnings as possible and to continually try to crack Facebook ads.
By the time I joined the team, we were spending around $7K and made about $13K back – it was a big stepping stone. But it wouldn’t have been possible to achieve that within a month without having the learnings and data to do so. They had collected numerous email addresses and had a good amount of visitors to the site for us to be able to determine where the opportunities were.
One way we did this was by creating more focused lookalikes of their buyers (e.g. buyers in the last 3 months as opposed to all buyers ever), which made a big difference.
If you’re looking to scale up, know that it comes with the risk of losses. The amount you scale up will align with your appetite for controlled risk (as far as being able to switch off poor performing ads for example).
How fast you scale often comes down to your tolerance for taking risk, but there are some controls you can put in place to make sure its not too too risky. I line out some of these controls below.
1. Place Facebook ad testing at the heart of your growth
It goes without saying that as with any growth initiative, testing as much as you can on the Facebook ads platform is essential to growth. At the early stages you want to focus primarily on audiences and creative.
Audiences in Facebook are a vast array of innumerable possibilities combining demographic, interest and behavior based targeting to allow you to be broad or deep with your targeting. Use Facebook’s Audience Insights tool (soon to be removed) to play around with targeting options and see what’s available (as per the screen above).
There are a few schools of thought here: the first is, if you’re targeting something like a hobby, to drill down to the nth degree to pinpoint your ideal target.
For example, rather than chasing people interested in a specific sport, within which your niche hobby falls into, drill down to a specific team or even a specific person associated with that sport that those die-hard hobbyists will likely know. Whilst Facebook’s complex algorithms generally work better with huge audiences into the millions, there are cases when you’re targeting a specific niche, that a smaller audience can pay bigger dividends.
On Facebook, there are a variety of things you can test:
-Dynamic product ads
Intra-platform ad variants:
-Right hand side ads
-Copy below the creative itself
-Text overlay on videos (which I recommend as most will view your video with sound off)
-CTA (call to action) button at the bottom right
For every ad format and Ad placement, there are mobile and desktop variants as well.
Facebook has a built-in test feature, available when creating new campaigns, called the Split-Test tool.
This is a more scientific way of testing within the platform, however not many use it as it doesn’t allow you to attach a test to an existing campaign. The advantage is that Facebook will slice your audience and ensure that a certain portion of your audience only ever gets to see one version of your ad, much like an AB testing tool might do.
Read more about Facebook’s Split Test feature here.
How to set up Facebook ad testing
The most common way to test Facebook ads is to create multiple ad sets and ads and let them play against each other in a campaign. One potential issue with this is that users may see more than one version of your ad – the only way to be prevent that from happening is by using the split test tool that Facebook offers.
However, a commonly missed part of testing is actually once the person clicks off of Facebook and into your funnel, whether that’s onto your website or an app. I’d go as far as arguing that targeting and getting the user to click your ad is easier than landing them on a page or app store and getting them to convert. In ecommerce it’s common practice to test landing pages and track the experience through checkout to see how a specific audience is reacting to an offer or creative.
We placed CRO (conversion rate optimization) at the heart of our growth strategy.
Both within the ads platform and on the website, we tried to tweak and improve our CRO at every opportunity.
The key to high converting landing pages
One of the key tests was placing social proof on our landing pages – as an unknown ecommerce retailer with a global footprint, we had to convince people to put their trust in us. We gathered social proof such as the volume of our sales, testimonials and media mentions. They played a big part in improving conversion rates.
2. Know your customer
It may sound simple, but how well do you actually know your ideal customer?
Selling a product aimed at children of reading age we knew that parents, in particular mums were a key market for us. And we did pretty ok selling to them too.
But what really lit a fire under our rocket was a survey sent to customers asking them who they were actually buying the product for. Surprisingly to us, the majority were buying it as a gift for other people’s children.
We then realized that by talking to what we thought was our ideal customer was only part of the equation – knowing what they were buying this product for was the other part; gifting! We began changing our copy, still predominantly targeted towards mums but with a focus on gift giving, and saw a big uplift in sales.
But what was even more interesting was that as we studied the additional data we were getting back, we saw the age of the children dramatically changed. Where we were selling a product aimed at kids 3 to 8 we were actually now selling it to gifters buying the product for newborns – it had become far more appealing as a keepsake product for newborns than an educational product for kids at a reading age as it was a personalized product.
Despite starting with what we believed, on the face of it, being quite a clear customer persona, we soon realized our customer persona was the thing holding us back.
This highlights one of the crucial aspects of our product marketing development which was focusing on customer feedback.
Why did they buy?
Was it as they expected it?
Would they recommend us to friends and family?
Would they buy again?
Is there anything missing that they’d have liked to have seen?
The list could go on and needs to be tailored for your business but uncovering the customer needs, wants and desires is a crucial part of making any marketing campaign work for you. This includes going through pre-purchase enquiries.
This will help you with retargeting messaging and improve your landing page experience by answering the most common issues.
3. Keep your creative fresh
As you start to scale up spend significantly, it’s important to keep your creatives fresh. Having advertised on the platform since 2012 with spends over $15M in that time, it’s far more common for the creative to have ‘fatigued’ (that is, having been served too many times) than it is for an audience to burn out.
This isn’t a hard and fast rule, but the creative refresh is often the first port of call for a waning ad set.
What does this mean?
Keeping an eye on your main goal: usually ROAS (return on ad spend) and CPA (cost per acquisition for ecommerce) to ensure you’re tracking ad performance.
When you’re running lets say 3 to 4 ads per ad set, if one them drops in performance it’s more than likely ad fatigue.
Another common reason for a sudden drop is also if you’ve had a negative, detrimental comment on an ad, which is why it’s important for someone (eg a customer services representative or your social media manager) to keep a regular check on comments.
We ran fortnight ‘design sprints’ – meaning every 2 weeks our designers would come up with new versions of ads for us to test.
In return we’d feedback performance from the last 2 weeks of testing, creating a revolving feedback loop where the designers would get to see and learn from the creatives we were testing and in return they’d either iterate or innovate from the previous designs.
As best practice, depending on audience size we’d aim for 3 ads running in a given ad set, but we’d also ensure we had ads lined ready and waiting if we needed to switch them on, for example if previous ads began to drop in performance. The worst thing is to lose momentum on an ad set waiting for new ads to be delivered.
Why 3 ads? Because you want to give enough options for Facebook to be testing with an ad set and because 3 gives the algorithm to run the majority of traffic toward one and test 2 others. In practice you might find the 2 ads that have fewer impressions are just not getting a fair run – however if your main ad drops in performance you have one of 2 backups to pull in and then replace the low performing ad with another.
For large accounts, spending upwards of $10k per day I’ve run 4 and maybe even 5 ads in an ad set. In other accounts I’ve only run one ad but then duplicated the ad set. They’re different ways of pushing your ads to your audience and is something to test as the right approach is dependent on your setup.
4. Quality lookalike audiences
Facebook’s lookalike audience (LLA) tool is a ground-breaking form of targeting; taking a ‘seed’ audience, which could be as low as 100 people, but ideally thousands if not tens of thousands, you ask Facebook to create a larger audience similar in attributes to this smaller audience.
Let’s take the example of this ecommerce store: we were selling to gifters who were generally mums with kids up to 8 years old.
Extracting a list of emails from our sales database we could target every person on our list.
We also pulled out high value customers; those that had bought more than 5 items over a period of time (12 months in our case). Now this could have been 5 in one order or 5 over a period of time, we didn’t differentiate.
This seed list proved to be far more profitable than simply putting all purchasers in. This is because those customers had attributes about them which made them spend more with us; Facebook’s amazing ability to find more people like that meant this new lookalike audience was of really high value.
To share another example, we used another seed audience based on Promoters from our Net Promoter Score (NPS) survey. NPS is a form of customer experience feedback, with a simple, single question of ‘would you recommend this company to friends or family’. The rating, between 1 and 10 decides whether they were promoters or detractors and anyone over an 8 is a promoter. This was one of our most profitable lookalikes we ever run and continues to be a high value audience for all ecommerce companies I’ve run NPS lookalikes on.
You can create lookalikes from people that have visited your website, viewed you cart, viewed certain categories, engaged with your Facebook fan page or Instagram page, of the most engaged people on your website and many more. The key here is to test with as many data points as you can to find the best seed audience for your campaigns.
5. Analysis without the paralysis
The Facebook Ads Manager tool is a useful way of reporting on data, by selecting columns of data, such as spend, clicks, and purchases. However, to really get deep into analysis, you also need to split by the various segments Facebook offers you.
The reason this is important is that you may find your CTR (click through rate) is far worse for example for a particular demographic, or placement, than other ads. There are many reasons why ads work or don’t work, and some of them could simply be that they’re more effective on Facebook than Instagram.
However for ecommerce, its important to track both the key Facebook metrics as well as the funnel metrics; that is, of those that clicked, how many added to cart and how many converted once they got to the cart.
It’s in particular important for ecommerce as there are factors involved in site conversion and average basket value that are driven by the store itself in the form of stock availability (or the lack of) and on-site promotions. For example if there’s an on-site promotion running, the discounted rate may turn into a lower ROAS, impacting your advertising effectiveness.
The table below is my preferred format for quick analysis before deep diving into optimization opportunities within ecommerce.
What we’re looking for here are the trends denoted by the color gradients and relative performance. Here for example we can see % ATC (percentage of people that add to cart) has declined in the week however the Cart CR% (how many have converted from the cart page to sale) has increased. However looking at ROAS, which for this ecommerce store is our main objective you can see that the ROAS peaked on Thursday at 181% driven by Sales CR% (click to purchase conversion rate), high funnel conversion and importantly a decent AOV (average order value).
The other hidden factor here is that since this ecommerce store has a maturing performance over 7-10 days, the ROAS will tick over to 200%+ courtesy of the way Facebook reports conversion data by default, which is attributing sales to a given day when the conversion may have happened days later.
However the Sunday is a clear anomaly here with low Cart CR%, lower ROAS and low AOV which might be an area of investigation.
The important part of analysis like this is spotting trends and reacting. The inner workings of what to look for and how to deep dive are too deep for this post to go into, but needless to say for ecommerce you’re generally tracking ROAS and CPA at the end of your funnel, and looking at the metrics in between as above, and also on CTR and CPC at the top of your funnel to identify where the opportunities to improve are, within your ad and/or within your website.
6. Focus on funnels
When it comes to Facebook advertising, each ad and interaction should nudge that user towards you. Where many fail at Facebook advertising is trying to convert someone that’s seen your ad for the first time, to buying from you.
It’s like going on a first-date and asking the other person for their hand in marriage! You might get away with it if you’re already rich and famous plus perhaps a smooth talker, but for most of us we need to do the whole courting thing.
The same goes for your Facebook ads: court your audience. Below is a typical funnel I use when creating ads for ecommerce.
There are 4 levels of campaigns that I run, each aimed at serving a specific type of audience.
At the top you have cold audiences, meaning that they’ve not seen your ads before and are not customers. You generally want to attract them into your product using a range of targeting options to entice them in.
For ecommerce, category specific videos work best. For example if you’re selling children’s clothing, then running a video ad to a cold audience focused on winter jackets, during the winter season is going to resonate and perform better than a general clothing store ad with no consideration to the season.
There are some stores that will do better at converting cold traffic straight away than others and these are mainly those that have lower ticket prices, lets say $20 and below where the consideration of parting with that amount of money isn’t too stressful. Where your item’s pushing $30-50 and upwards, then you need to play the courting game harder.
With this campaign structure above, we have a ‘warm’ audience targeting pot where we’re targeting those that have interacted in some way – whether by liking our fan page, engaging with ads or viewing a video. We know they know about the brand/product so we can communicate differently at this stage to hook them in with customer benefits and testimonials.
The next level down is our hot prospecting list of people that have visited the site, and ideally those that have added to cart. The types of ads you offer here are more product focused, more sales focused and simpler in messaging, often using DPA ads (dynamic product ads) – you know the ones, that seem to follow you around showing the exact product you just browsed.
Now, most ecommerce stores stop at this point, once the sale has been made. However, whilst many stores will employ email campaigns to get customers to repurchase after their first purchase;
Using Facebook ads after they’ve purchased is an important way of increasing the customer’s lifetime value at a lower cost per acquisition resulting in a higher ROAS over time.
Using the full funnel approach, which is something to be tested and adapted for each store is a good approach at combining the business objectives of strong ROAS from your customers with taking those customers on a journey with you, priming them, enticing them and then closing the sale.
7. Scaling with manual bidding
One of the power tools within Facebook is manual bidding. Those from a paid search world might consider it strange that Facebook marketers find manual bidding so fascinatingly difficult to use and yet it’s amazing how powerful automatic bidding can be.
When entering an auction to have your ad displayed, Facebook is calculating the best way to use your budget based on:
-how it is allocated (eg daily or lifetime)
-how much there is
-your optimization goals (eg to get people to purchase)
-your ad set history
Amongst many other factors.
Autobidding allows Facebook to flex your bid to try and get you the best results for your budget. For the most part, and for those at the beginner to intermediate stage of learning this, more often than not it delivers compelling results.
However to scale faster and harder, pushing more and more spend often requires you to take control through manual bidding. Whilst Facebook is good at generally getting you into the right auctions at the right price, there comes a time when you cannot increase reach without hitting better quality members of your audience.
Using an ocean analogy, manual bid is like pushing your rod deeper than the surface level and grabbing the juicier fish that are harder to get. Or in reality, bidding higher to reach users that are likely more expensive than your auto bid is prepared to go to.
It comes down to risk and reward and much of scaling Facebook ads falls into mindset. What level of risk are you willing to take in order to generate growth? Let’s say your ecom website has a $20 CPA (cost per acquisition). If you want to scale more aggressively, you might increase budget and leave the bidding to auto, or duplicate the ad set and add new budget there as mentioned earlier. Or you might actually put a manual ‘cap’ in place, lets say $20 to ensure your CPA doesn’t go higher than that. However the reality is that you don’t really know what auto bids Facebook are putting in – that’s because the bid you place, doesn’t accurately reflect the CPA that churns out at the bottom of your funnel.
The bidding system Facebook uses is based on the following graphic in its simplest sense. The important elements are Estimated Action Rates and User Value. Both of which you cannot see but will have an idea about.
The lowdown is this – Facebook will use the conversion rate of your desired outcome, such as purchases and the likelihood of a given user to take that action and also add an additional score based on the relevance of your ad (calculated from the ad relevance score) and other things like: Do users go beyond your landing page? Do they browse the website in general? Are they engaged with your website? etc.
The benefits of manual bidding
For ads that have a poor relevance score, manual bidding ups your chance of winning the auction and will often result in a higher CPM and CPC. However, the offshoot could be that your CTR and your website conversion rate also increase because you’re now reaching a customer that’s far more relevant to you. Conversely a high relevance score gives you by default a higher chance of winning an auction and in either case, manual bidding will bump you higher in the auction.
When scaling hard in the US for example, we were chasing a broad “mother” audience, we placed a manual bid in the region of $40 for a target CPA of around $15. How does this make sense? Because although the CPA did increase marginally to around $18, so did both the website conversion rate and average order value for these users because they were higher quality. I’d only advocate manual bidding higher when you have a proven audience and good creative already in place.
Conclusion on Facebook Ads & Ecommerce
Ultimately with ecommerce the challenge with Facebook ads is in maintaining a positive customer lifetime value. While there are many ecommerce stores that are able to generate ROAS of 250-500%, as you start to scale your spend,the auctions become more expensive. It becomes even more important to make more money on back-end, once you have the customer, to justify the rising costs on Facebook.
Key takeaway: Facebook ads are a great way to drive targeted traffic to your site. As the cost of Facebook ads rise, focus on the entire funnel to increase customer LTV, not just the Facebook ads platform.