As we approach peak season for eCommerce, SMB owners are looking at where their budget should be allocated to maximise Q4.
Increased buying propensity across Black Friday and in the lead-up to Christmas will be a welcome change after the economic climate we’ve seen in the UK in 2023.
Whilst there is opportunity, Facebook ads and paid social can generally be more challenging to crack.
Without the huge budgets and brand familiarity of the big guys, it can feel like a bit of a lost cause at times.
It is possible to get real results, though, and even with a limited budget, drive commercial value without breaking the bank.
There are three common challenges that SMBs run up against:
● Limited budgets/shorter duration campaigns
● Increased competition/higher CPCs
● Finding audiences that are ready to buy
Each of these limits performance as the amount of positive feedback (conversion data) coming back into the algorithm is reduced.
You don’t need to become a data scientist to understand how machine learning drives performance in ad campaigns.
All you need to know is that the conversion data is feedback that’s essential for the campaign to find more users who are likely to become customers for you.
It will take the contextual pattern of data around a user who made a purchase and then find it elsewhere around other users on the channel.
This means that the more sales that occur from the campaign, the better predictions can be made about entering other auctions and as a result, further sales occur.
The challenges are overcome far easier with larger budgets or being a household name brand. Sales occur at volume, quickly in those instances, and the campaign is then able to exit the learning phase, finding those likely to be in-market for the products being advertised, even in a very broad audience.
Getting to this place actually supports cost efficiency in terms of getting a higher ROAS or lower cost-per-acquisition.
Targeting broader audiences allows you to go into less competitive auctions with lower CPCs, all the while with that very well-defined predictive model that the user is likely to be in-market for what you sell—not so broad in reality!
What we need to do
Limited budgets/shorter duration campaigns
Optimising for sales makes sense on the face of it—it’s what you want. However, it’s one of the common configurations that blocks smaller brands from seeing sales or sales at target cost in Facebook ads.
To support campaigns with a limited budget or those that only run across a short time period (such as 3-5 days for Black Friday weekend), it’s advisable to change the conversion action that your campaign is optimising for.
You should still utilise the conversion objective, as others, such as traffic or brand awareness, aren’t looking to drive commercial outputs—even if the audience itself has buying propensity.
A traffic campaign, for example, is looking for those in your target audience who are most likely to click, not those who are most likely to click and then also become a customer.
Optimising for an event earlier in the conversion journey, such as add to cart rather than purchase, enables the campaign to get feedback from those who have taken more steps towards providing a valuable action to your business without limiting the number of signals that the campaign receives back as positive feedback.
You can still measure success based on sales, revenue and ROAS by customising your reporting table.
If there are not enough add-to-cart events to exit the learning phase (typically 50 are needed across a 7 day period), it could be an indication that your budget isn’t high enough to generate enough clicks to allow a statistically significant amount of add to carts to occur or that there are fundamental issues with your proposition.
If this is the case, you may want to reconsider using paid social at this moment in time.
If this was a campaign that was going to become ‘always-on’ rather than just a seasonal push, you would look to move the conversion event back to purchase once enough sales were being driven consistently.
Budgets themselves are a tricky one; you need to spend enough for there to be a chance for sales to occur, and above that, it should be based on performance and results. If the campaign is converting at a profitable ROAS, then the budget shouldn’t be restricted.
The more likely factor for a smaller business to consider is the minimum likely to be enough to see meaningful data.
If your average CPC is £0.90, your daily budget is £20, and your conversion rate is 4%, statistically speaking, it’s correct that you shouldn’t see a sale (22 clicks/4% = 0.88).
Of course, this is far too black and white from a numbers-only perspective.
Perfection is very often the enemy of good. Getting the campaign on and seeing what happens can be a much faster, more effective way to learn.
That being said, I find it helpful to frame expectations in this way. It’s not ‘wrong’ if the campaign didn’t drive a sale under those conditions; it would be more surprising if it did (statistically speaking).
Machine learning is extremely competent at providing the desired outputs we need in both PPC and paid social. When the output isn’t there, looking at the conditions we’re asking it to perform under is where we will find scale and profitability as business owners and advertisers.
Increased competition/higher CPCs
Social channels get very noisy during this time, not just with ads but also with organic content.
Standing out is all about being clear and concise with your offer and messaging.
You’re not just competing with your competitors but also every other business vying for users’ attention.
Keep your offer as simple as possible with clear information on how to redeem. If it’s a discount, is there a code, and if so, how do they get it?
Introducing any additional points of friction or uncertainties will lower the efficacy of the campaign and uptake of the offer more broadly.
Think about the questions that would arise in your head if you saw the ad yourself and hadn’t previously engaged with the business.
Good creativity hooks engagement because it appeals to our system 1, the fast emotional response system.
Copy needs to address the pragmatic, analytical system 2, our thinking, reasoning system. This is the internal questioning that comes afterwards;
“If I order it, how long will it take?”
“What if it’s the wrong colour? Can I send it back?”
“Does it come with a warranty?”
“How do I get the 10% off they’ve mentioned?”
To support being prioritised in the auction, it’s worth making sure the website is in order.
Website experience is a factor that is considered in terms of who can show in the most competitive placement and how much you pay.
You can’t optimise the click costs down below the baseline of the auction, but you can pay less than a competitor by providing a good user experience.
Review your website’s key pages, landing page, product page, add to cart and checkout on a mobile device as if you were a user.
Often, pop-ups, misplaced banners, simple tweaks to layout or improving header messaging can remove uncertainty and confusion when someone comes through via an ad.
You’ve done all this hard work to get your campaign in front of the right potential customer; don’t lose it now they’re on the website.
Improving here has a compound effect on campaign performance. It both increases your sales volume and lowers your acquisition costs at the same time.
Finding audiences that are ready to buy
Paid social audiences are typically more those who are “interested” in a given criteria rather than actively looking for the product or service.
This is a key point of difference between search and social advertising.
Someone actively searching for “xyz product” in Google can be deemed in-market, and, as such, there typically is a higher efficacy when they’re then shown a search or shopping ad for the product.
We can get around this by the guidance put into the first recommendation. Once conversion data is coming in, the audience goes from;
Interested in home furniture
Interested in home furniture + has exhibited a vast number of data signals similar to those who have bought from us previously
We can support getting this to where it needs to be by the way we set up our campaigns and blend our audiences.
Creating one campaign, enabling budget optimisation at the campaign level and then having multiple ad sets for different audiences will allow us to help teach the algorithm faster. This
happens through conversion data coming into the campaign quickly via remarketing, this is then available to the other broader ad sets to use to further guide targeting.
This way, the campaign can choose to spend where it’s most likely to deliver the commercial results you’re asking it for.
It avoids over-segmentation, which further spreads lower budgets too thin and reduces management time—great as an SMB.
My suggested structure is:
Ad set 1 – Purchase lookalike
Ad set 2 – Relevant interest audiences
Ad set 3 – Remarketing (add to cart or product page view)
What you will typically see is that initially, the remarketing ad set will be prioritised with budget as they are users who have already visited the website, so are more likely to buy.
There’s a natural ceiling to this ad set, though, as it’s constrained by website visits.
As the conversion data comes in, the other more broad audiences will start to be prioritised as they’re scalable and are likely to be lower cost.
If this was an always-on campaign, you would also want a very broad ad set, so no audience targeting added, the same situation occurs.
Having remarketing in the same campaign enables the fast data to teach the campaign and then unlock new user acquisition, which is far more important to the business (obviously with returning users excluded from new user audiences to avoid overlap.
You don’t want to rely on this, though, as remarketing generally reports inflated performance due to many users being likely to have come back and converted anyway.
This approach is very much how a Performance Max campaign works in Google Ads with audience signals. We’re giving guidance and supporting the algorithm in learning with data.
Having the audiences separated means that you can still tailor your copy.
What you say to the new user who hasn’t visited the website and what you say to the user who has visited but chose not to purchase should be different, right?
Addressing these three areas of Facebook ads should allow your SMB to get better performance this peak period!
I would implement the audience structure from the get-go.
You can always start by optimising for purchase and then change to add to cart if you see that you’re not exiting the learning phase once the campaign is on.
The basics can’t be underestimated.
Go through the website on your mobile device with your key stakeholders, and get others who aren’t close to the business to do so also.
There is nothing more valuable than seeing how real people interact with your store, learn where they drop out and fix the issues.
Uncover the questions and hesitations users have and make that front and centre in your messaging.
Improving the post-click experience will increase the conversion rate, which lifts sales volume and improves profitability in campaigns. It also makes other marketing channels’ performance increase, so is a real 80/20 if you’re limited on time.
If search and shopping are working for you and you can increase spend, I would recommend doing that first, as capturing in-market users will always be more effective in driving a profitable return.
This is a guest blog written by Byron Marr, Founder of ProfitSpring. To hear more about their PPC and paid social services, get in touch.