Solved: Should I Start a New Facebook Pixel After 2020?

Hi there,

Thanks for reaching out. I saw your query about the Facebook pixel and whether to start fresh. It’s a really interesting question and something a lot of businesses have been grappling with, especially after the weirdness of the last few years totally changed customer behaviour for a while. It's a common problem, so you're not alone in thinking about it.

I'm happy to give you some of my initial thoughts and guidance based on what we've seen work for our clients. Tbh, it's less about the pixel itself and more about the strategy you use to guide it. Below are my thoughts on how I'd approach this.

We'll need to look at whether a new pixel is the answer...

Okay, so your first thought is to just scrap the old pixel and start again. I get the logic. The data feels "muddied" or "contaminated" with low-quality customers, so the impulse is to get a clean slate. But I'd strongly advise against this. It's a bit like throwing the baby out with the bathwater, innit?

Thing is, your pixel has been collecting data for years. It holds every click, every page view, every add to cart, and every purchase from your entire history. Yes, that includes the "bad" customers from the pandemic, but it also includes all the rich data from your ideal pre-2020 customers. Wiping it clean means you lose *all* of that history. You'd be starting from absolute zero, putting the pixel back into its initial "learning phase" which is often expensive and slow. The algorithm would have no idea who your ideal customer is, and you'd likely see your costs go up significantly while it tries to figure it all out again.

What about running two pixels side-by-side? Tbh, this just creates a technical mess. It can lead to all sorts of attribution problems, makes it difficult to track performance properly, and overcomplicates the whole setup for no real gain. In paid ads, simplicity is usually the best approach. The problem isn't a faulty peice of code; it's the instructions you're giving it. So the solution isn't a technical one, it's a strategic one.

I'd say you should refine your audience strategy, not reset your data...

Your current approach of creating segments and lookalikes from pre-2020 customers is absolutely spot on. That tells me your instincts are right. You've correctly identified your 'golden' dataset. The next step is to build your entire Meta ads strategy around this principle, using a structured funnel approach.

The pixel is a tool. It works best when you feed it clear, high-quality signals. Instead of wiping it, we need to be more deliberate about which parts of its data we ask it to use. This is where a proper funnel structure comes in. We split campaigns into different stages: Top of Funnel (ToFu), Middle of Funnel (MoFu), and Bottom of Funnel (BoFu).

This structure forces us to be disciplined with our targeting. At the top, we find new people. In the middle, we warm them up. At the bottom, we convert them. The further down the funnel, the more valuable the audience, because they've shown more intent. Your job is to guide Meta to prioritise these higher-intent audiences.

This is the general prioritisation framework I use for eCommerce clients. It shows how you should think about audiences, from the most valuable (previous customers) to the broadest (interests).

By structuring your campaigns this way, you're telling Meta exactly who to go after. You're not letting it guess based on a mix of good and bad data; you're giving it a treasure map that leads directly to your best potential customers.

You probably should focus on high-intent lookalikes and retargeting...

Let's drill down into that. Your pre-2020 customer list is your most valuable asset. Don't just upload the whole list and create one lookalike. Segment it first. Create seperate lists for:

-> Your VIPs: The customers who spent the most or purchased most frequently before 2020. A lookalike of this small, elite group will be incredibly powerful.
-> Your Regulars: Anyone who purchased more than once.
-> All Customers: The full pre-2020 list.

Now you have three potent seed audiences to create lookalikes from. You can test a 1% Lookalike of your VIPs against a 1% Lookalike of All Customers. I'd wager the VIP lookalike performs much better, because the seed data is so much more specific and qualified. This is how you leverage your good data without being dragged down by the bad.

I recall working with a subscription box client who was struggling to scale. By isolating their highest lifetime value customers and creating lookalikes from just that segment, we were able to focus the ad spend so effectively that we hit a 1000% Return On Ad Spend. It’s not about having massive audiences; it’s about having the *right* ones.

The same goes for your BoFu retargeting. People who add to cart or initiate checkout are sending you the strongest buying signals possible. Your pixel has a record of all these people. You should have a dedicated, always-on campaign just for them, with aggressive offers to get them over the line. These audiences are pure gold and should be your first priority for ad spend.

You'll need a solid testing structure...

So, how do you put this all together? You need a clean campaign structure that reflects the funnel.

-> Campaign 1: BOFU Retargeting. This campaign contains ad sets targeting everyone who has added to cart, initiated checkout, etc., in the last 7-30 days. Your goal here is direct conversions.

-> Campaign 2: MOFU Retargeting. This campaign targets broader website visitors and video viewers who haven't taken a high-intent action. The goal is to get them back to the site to look at products again.

-> Campaign 3: TOFU Prospecting. This is where you test your new, high-quality lookalikes and maybe some very specific interest audiences. You'd have one ad set for your VIP pre-2020 Lookalike, another for your All Customers pre-2020 Lookalike, and so on. You run them against each other and see which one performs.

You need to be ruthless with testing. Let an ad set run until it's spent about 2-3 times what you're willing to pay for a customer (your target CPA). If it hasn't got a sale by then, its probably not going to work. Turn it off and move the budget to the winners. This systematic approach, combined with the right audience strategy, is how you re-train the pixel and tell it what a "good customer" looks like, without losing years of valuable data.

This is the main advice I have for you:

As you can probably tell, getting this right involves more than just flicking a switch. It's about building a robust, strategic framework for your advertising that can weather changes in the market and consistently find your ideal customers. It can be a lot to manage and optimise on an ongoing basis, especially when you also have a business to run.

If you'd like a hand with this or want us to take a proper, detailed look through your account with you, we offer a free initial consultation to do just that. It's a great way to get a clear, actionable plan tailored to your business.

Hope this helps give you a new way to think about the problem.

Regards,
Team @ Lukas Holschuh