Hi there,
Thanks for reaching out. I've had a look at the situation you described, and it's a problem I've seen quite a few businesses grapple with, especially in the last couple of years. The sudden shift in customer behaviour during the pandemic really did throw a spanner in the works for a lot of well-optimised ad accounts. Happy to give you some initial thoughts and guidance on how I'd approach this.
Your idea of starting with a new pixel or running two side-by-side is an interesting one, and I can see the logic behind it – wanting a clean slate. But honestly, my immediate reaction is to advise against it. Your existing pixel, even if it feels 'muddied', holds years of valuable data. Getting rid of it would be like tearing out the foundations of your house to fix a leaky tap. You'd lose all that learning, all your retargeting audiences, and a new pixel in a new ad account would have to go through a long and potentially expensive 'learning phase' from absolute scratch. It's a massive step backwards.
The real issue isn't the pixel itself, but how the data within it is being used. The solution isn't to start over, but to get much more sophisticated and deliberate with how you segment and target your audiences. You're on the right track with creating lookalikes from pre-2020 customers, that's exactly the kind of thinking that will get you back on track. We just need to build on that and structure it properly.
We'll need to look at... why your old pixel data is still a goldmine
First off, lets talk about that 'muddied' data. It's not a total loss. The Facebook algorithm is incredibly powerful. Even with a mix of good and not-so-good customers from the last couple of years, it can still identify patterns and behaviours. The trick is to give it the right signals and tell it which patterns to focus on. Your pixel has tracked everyone who has visited your site, viewed products, added to cart, and purchased for years. This is an irreplacable dataset.
This historical data is the bedrock of effective retargeting. Without it, you can't reach people who've already shown interest. It's also the source material for creating your most powerful prospecting audiences – lookalikes. By deleting the pixel, you're essentially telling Facebook to forget every single person who ever interacted with your brand online. We don't want to do that. Instead, we want to teach the algorithm to be more discerning, to focus on the 'good' data and find more people just like them, while ignoring the 'bad' data.
I'd say you... need a proper audience segmentation strategy
This is the core of it all. Instead of a blanket approach, you need to think about your advertising in terms of a funnel. We typically break this down into three stages: Top of Funnel (ToFu), Middle of Funnel (MoFu), and Bottom of Funnel (BoFu). This just means we're separating people who've never heard of you from those who are aware of you, and those who are on the verge of buying. You need seperate campaigns and audiences for each stage.
Top of Funnel (ToFu) - Finding New Customers
This is your cold audience, people who don't know you exist. Your goal here is to find people who look just like your very best customers. This is where your pre-2020 customer list is so powerful.
-> High-Value Lookalikes: Don't just upload one big list. If you can, segment that pre-2020 list even further. Who were your highest spending customers? Your most frequent buyers? Create a list of just these VIPs and use it to create a 1% Lookalike audience. This is telling Facebook "find me more people who look exactly like my very best customers". This is far more powerful than a lookalike of every single customer you've ever had. I've seen this alone turn campaigns around.
-> Detailed Targeting (Interests): You can also test interest-based audiences, but you have to be specific. If you sell, say, high-end hiking gear, targeting a broad interest like "Outdoors" is useless. It's too big and full of people who just like a walk in the park. You'd be better off targeting followers of specific gear review magazines, professional climbers, or niche hiking forums. Think about what your ideal customer is interested in, not what every vaguely related person might be. Test these interest groups in their own ad sets so you can see what works.
Middle of Funnel (MoFu) - Warming Up Leads
These are people who have engaged with you in some way but haven't made a purchase or even added a product to their cart. They know who you are, but they're not convinced yet. Your goal is to build trust and remind them you're there.
-> Website Visitors: Create a custom audience of everyone who has visited your website in the last 30-60 days (but exclude anyone who purchased). You can show them ads featuring testimonials, user-generated content, or highlight your brand's story.
-> Video Viewers: If you use video ads at the top of the funnel, you can create an audience of people who watched, say, 50% or more of your video. These people are clearly interested. Retarget them with a more direct product ad or a special offer.
Bottom of Funnel (BoFu) - Closing the Sale
This is your hottest audience. These people have shown clear buying intent. They've added items to their cart or even started the checkout process. These are your highest priority for retargeting.
-> Add to Cart (ATC): Target people who added a product to their cart in the last 7-14 days but didn't buy. You can even use dynamic ads to show them the exact product they left behind. A small discount here can often be enough to get them over the line.
-> Initiate Checkout (IC): This is even more powerful. These people were literally one step away from giving you money. Hit them with a retargeting ad as soon as possible, reminding them to complete their purchase.
You probably should... perfect your customer lists before uploading
You mentioned using your pre-2020 customer list, which is great. But the quality of the lookalike audience Facebook creates is directly related to the quality of the source list you provide. A simple list of emails will get you a match rate of maybe 30-40%. But if you provide a properly formated list with as much information as possible, you can get that match rate up to 70-80% or higher, which makes a huge difference to the quality of the lookalike.
You should aim to include: First Name, Last Name, Email, Phone Number, City, County/State, and Country. The more columns you can fill, the better Facebook can find those exact people in its user base and the more accurate the resulting lookalike will be. It's a bit of extra work to pull and format this data from your CRM, but it's one of those things that has a massive impact on performance. I remember one client in the SaaS space who saw their Cost Per Lead drop significantly just by enriching their customer list before creating new lookalikes.
You'll need... a structured testing approach
Once you have these different audiences, you can't just lump them all together. You need to structure your campaigns properly to control your budget and see what's actually working. A simple, effective structure is:
-> Campaign 1: Prospecting (ToFu): This campaign's job is to find new customers. Inside it, you'll have different ad sets. One for your 1% High-Value Lookalike, one for your general pre-2020 Purchaser Lookalike, and maybe a few for your best interest-based audiences. Let them run against each other and see which performs best.
-> Campaign 2: Retargeting (MoFu/BoFu): This campaign is for everyone who already knows you. You can have an ad set for your Website Visitors/Engagers (MoFu) and another, more agressive one for your Add to Carts/Initiate Checkouts (BoFu). The BoFu audience should get more of your retargeting budget as they are closest to converting.
By separating them, you can allocate your budget intelligently. You might spend 80% of your budget on prospecting and 20% on retargeting. You can also quickly identify which specific audiences are driving results and which are wasting money. Turn off the losers, and give more budget to the winners. It takes the guesswork out of it.
This is the main advice I have for you:
| Recommendation | Actionable Steps | Why It's Important |
|---|---|---|
| Keep Your Existing Pixel | Do not create a new pixel. Continue using the one you have to leverage its historical data. | Avoids a long, costly 'learning phase'. Preserves years of irreplaceable data for retargeting and lookalike creation. |
| Create High-Quality Customer Lists | Export your pre-2020 customer list. Segment it into 'All Customers' and 'High-Value Customers'. Format the data with multiple columns (name, email, phone, address etc.). | Improves Facebook's match rate, leading to much higher quality and more accurate Lookalike audiences. This is a critical step for performance. |
| Implement a Funnel-Based Campaign Structure | Create separate campaigns for Prospecting (ToFu) and Retargeting (MoFu/BoFu). Test different audiences in separate ad sets within these campaigns. | Gives you clear control over your budget and allows you to accurately measure the performance of each audience, so you can scale what works and cut what doesn't. |
| Prioritise Your Audiences | ToFu: Start with a 1% Lookalike of your high-value pre-2020 customers. MoFu: Retarget website visitors & video viewers. BoFu: Agressively retarget 'Add to Carts' & 'Initiate Checkouts'. |
Focuses your budget on the audiences most likely to convert at each stage of the journey, maximising your return on ad spend. |
As you can probably see, while the concept is straightforward, the execution involves quite a few moving parts and requires a methodical approach. It's not just about flicking a switch, but about building a robust system that you can manage and optimise over time. Getting this structure right from the start is what separates the campaigns that struggle from the ones that scale profitably.
It can often be helpful to have an experienced pair of eyes to look over the account and help put a concrete strategy like this into action. If you'd like to go through your ad account together and map out a specific plan, we offer a free, no-obligation initial consultation. We could dive into your exact audiences and campaigns and figure out the best way forward.
Hope this helps!
Regards,
Team @ Lukas Holschuh
