Are you creating dozens of beautiful visuals for your campaigns, launching ads and hoping for maximum profits, but the numbers in your e-shop don’t quite match the enthusiastic reports from social networks? In this article we look in detail at why Meta’s modern Andromeda architecture requires a constant supply of new ad assets, and how to tell with absolute certainty which of them are really earning you money. You may very soon find that your performance evaluation so far stood on very shaky foundations.
Summary for Those Who Don’t Have Time to Read the Whole Article
- Meta’s Andromeda today doesn’t look for customers primarily by interests, but by how they react to the specific content of your videos and banners.
- The lifespan of ad assets in Advantage+ campaigns is shorter, and therefore it is essential to keep testing brand-new visual concepts.
- For reliable optimisation you must deploy the Conversions API and send Meta hashed customer data for the maximum match rate.
- Absolute order in ad naming combined with automatically filled-in link parameters is the only way to find out the real contribution of a specific banner.
- Google Analytics 4 works as the only objective source of truth, able to uncover low-quality traffic hiding behind cheap clicks in Meta.
Does the technological background of campaigns seem too complicated? Leave setting up the right analytics to professionals. We’ll be happy to help.
Creative Works as the New Targeting
In the past, as marketers we spent hours creating dozens of ad sets with all sorts of narrow interests, into which we put two or three images. Today’s reality is completely different. Meta’s Andromeda artificial intelligence requires a broad audience and an enormous amount of different creatives.
Wondering why Meta’s Andromeda needs such a volume of varied content? The algorithm works by analysing in depth the pixels in an image, the text in a video and the accompanying ad copy. Depending on who stops at your ad in the endless scrolling of posts, the system automatically and very precisely models the target group. A single visual showing a young family thus reliably reaches a completely different market segment than a plain banner with a detailed technical drawing of the product, even when they run in the same campaign.
Why You Can’t Do Without Constant Testing
Another major problem is so-called creative fatigue. Professional studies from practice show that the overall lifespan of a creative in modern Advantage+ campaigns is significantly shorter than in the past. If you want to keep the audience’s attention and stable performance, it is absolutely necessary to keep sending new concepts into circulation. These can be dynamic videos, authentic UGC (User Generated Content) material, classic static images or interactive carousels.
The Pitfalls of Artificial Intelligence: When Meta Picks a False Winner
Imagine you upload fifty different banners into a running campaign. Meta’s Andromeda immediately starts delivering the content and fairly quickly picks its favourites, which it starts pushing heavily. The catch is that these winners are typically chosen on the basis of a high number of clicks and cheap impressions within the social network.
But this artificially chosen favourite doesn’t have to be a winner from the point of view of your real business at all. It often happens that the most-clicked image brings visitors to your e-shop who leave immediately and don’t buy. It is therefore essential to achieve real conversions with a sustainable cost-of-sales ratio (PNO) (share of costs in revenue) and ROAS. This is exactly the moment when the need for truly precise data analytics steps in.
Correct Tracking in Meta: Without Data the AI Learns Nothing at All
For the advanced Meta Andromeda to pick the banners that sell best, it must receive absolutely immediate and undistorted feedback on what exactly the user did after clicking through to your website. Since the infamous iOS 14.5 update and the introduction of App Tracking Transparency, it has been clear that the basic Meta Pixel measurement code simply isn’t enough.
It is absolutely essential to connect the classic pixel with the Conversions API (CAPI). This solution can send purchase information directly from your website’s server to Meta’s server. In this way you elegantly and safely bypass all sorts of ad blockers and strict web browser restrictions, such as Apple’s ITP technology.
Likewise you can’t avoid using the Advanced Matching function. You must send the advertising system hashed, that is securely encrypted, data about your paying customers, such as an email address or phone number. This practice radically increases the so-called Match Rate. It gives the algorithm an exceptionally clear signal for optimising campaigns towards similar-minded and, above all, similarly creditworthy audiences.
Practical recommendation: In the Meta Events Manager interface, regularly monitor the Event Match Quality metric, that is, the event quality score. If this rating falls below 6 out of 10, it means the artificial intelligence does not have enough quality data to deliver your creatives effectively. You can also mitigate the selection of a false winner by not targeting campaigns at reach or landing page views, but directly at conversions. If there is enough data on purchases, use secondary conversions such as add to cart, viewing the cart detail or proceeding to payment.
UTM Parameters in Meta
While the ad account likes to report rocketing performance growth according to its own attribution models, which are very often inflated thanks to view-through conversions, the analytics on your website may tell an entirely different story. For a real and undistorted finding of the hard impact of advertising on the website, detailed UTM parameters are absolutely essential. Given the astronomical volume of deployed creatives, however, manually tagging every link would be desperately inefficient and full of human error.
TIP: What are UTM parameters and what are they used for?
Fortunately, Meta allows you to use dynamic tags very smartly. These dynamic URL parameters are automatically rewritten according to which specific creative or campaign the user clicks on their phone at that moment. In the settings of every single ad you therefore need to insert a special dynamic formatting code into the URL parameters field.
At the beginning, this code should define the source as meta and the medium as cpc. Then, using dynamic variables in simple braces, it pulls in the exact campaign name, the ad set name and, what is absolutely most important – the ad name itself. It is precisely the parameter working with the name of the specific ad that is the most secret ingredient for perfect analytics.
For this automatic measurement to work reliably, however, you must set absolutely clear rules in your company for how you name all your ads and stick to this system without exception. We call these rules a UTM strategy, or a link-tagging strategy. It can seep deep into your marketing processes, but it is absolutely essential. If you name your banner simply “Image_1”, analytics won’t help you. The name must contain the year, month, format and topic, so the result looks, for example, like “202310_Video_UGC_WinterJacket_V1”. Without this organisational discipline, all that will appear in your systems is a jumble of completely unidentifiable data.
With the arrival of Andromeda, creatives and campaign managers are pushed much harder towards correct naming of the creatives themselves too (i.e. the banner videos and photos), so that it is possible to trace which video “caught on” and to develop it further (different hooks, different angles).
Google Analytics 4 as the Single Source of Truth (SSOT)
Once you have all the parameters set up flawlessly, it is the turn of Google Analytics 4, your one true source of truth. This system relies on a modern event model and precise attribution based solely on hard data (Data-Driven Attribution).
The key difference in the metrics is that the advertising algorithm has a very strong tendency to claim every conceivable conversion after a mere fleeting display of the ad on screen. By contrast, thanks to a data-driven attribution model, GA4 can break down the tangled customer purchase journey across all available marketing channels far more fairly and accurately.
Tracking real performance directly in the analytics interface involves three essential steps. First, you should regularly dive into the Traffic acquisition reports, where you set ad content as the primary dimension. This dimension mirrors exactly your dynamic parameter for the specific visual.
Second, carefully analyse the metrics that determine the overall quality of a visit. Do you notice a high bounce rate or a very short engagement time? A particular catchy banner may show a seemingly wonderfully cheap cost per click right in Meta, but in GA4 it reveals downright clickbait behaviour from users. People do click en masse, but leave the website immediately, which genuinely harms your long-term business.
The third essential step is a detailed comparison of the different attribution models right in the Advertising section. Here it is particularly worthwhile to analyse the share of the most successful graphic concepts in your customers’ complete conversion paths.
Correct Tracking = Effective Marketing Decisions
This whole complicated technological ordeal leads to one single thing: the correct and confident allocation of your precious budget. Let’s imagine an absolutely typical example from practice.
Creative A shows dozens of purchases in the social network’s reports and looks like an absolute champion. In GA4, though, you see that it figures at most as a marginal assist, and the overall quality of its traffic is dismal. By contrast, creative B looks fairly unremarkable in the ads manager, but according to the bulletproof data from GA4 it brings high-value customers with a many times better conversion rate. If you are a sharp marketer, all budget optimisation should immediately head towards massively supporting creative B.
This iterative and highly rational analysis can in no time uncover other strong formats. It may turn out that ordinary dynamic videos with a real person bring on average up to a 30% better conversion rate compared with expensive and polished static product photographs. Similar, indisputable data feedback then enables your in-house production team to make more effective assets and to consistently “feed” the algorithm that Meta’s Andromeda uses exclusively with the content that really generates tangible profit.
We will help you set up correct measurement and connect the data from Meta directly with your website.
An Overview of Creatives Straight from the Meta Insights API
If you implement a creative naming strategy correctly, i.e. label your creatives by angles, products and hooks, you can start downloading summary tables, either in the campaign editor interface in Meta or, above all, in the Meta API, which break everything down by individual creatives, and decide in real time on the priorities of your marketing team accordingly. What we used to analyse at campaign level before Andromeda is now moving to ad set level, and such data can already be overwhelming for many views in GA4. So consider this route too when optimising your marketing processes and your campaigns. The goal, after all, is to get information to the creative team as quickly as possible about which creatives caught on and deserve further development. And we will gladly help you with this goal.
Conclusion
If you want to succeed in the highly competitive environment of modern social networks, you must keep supplying the algorithm with diverse visual nourishment. The system is much better at understanding who your ad actually reaches from the content of the video itself than from manually clicked-in interests. Once you tidy up your ad names, use automatic parameters in links and start deciding based on real data from your website, you will finally stop throwing money out of the window.
You might also be interested in: Looker Studio (Google Data Studio): What It Is and How GA4 Data Visualisation Works
Frequently Asked Questions
What exactly is Meta’s Andromeda?
It is a modern architecture of Meta’s delivery algorithms with huge involvement of artificial intelligence. It abandons classic targeting by demographics or manually set interests and uses advanced machine learning to find the ideal paying customer. The algorithm analyses the visual itself, the video and the text of the creative.
Why does Meta’s Andromeda need so many different banners?
Because the system evaluates the pixels and messages directly inside your content. Depending on what kind of people stop at a specific video or image while scrolling, it automatically models the next target group. A broader palette of quality visuals thus means immediately reaching completely different and new market segments.
Why is it no longer enough to rely on the basic Meta Pixel?
Since the radical iOS 14.5 update, coupled with strict control over the sharing of personal data (App Tracking Transparency), browsers block a not insignificant share of ordinary measurement pixels. Without connecting server-side measurement through the API, the advertising system would lose a lot of information about completed conversions, and its ability to deliver ads well would drop rapidly.
What is the difference in measurement between Meta and GA4?
The advertising network is in the habit of claiming huge credit even when the user merely passively saw your ad and bought only several days later from another source. According to analysts’ findings, the difference between the platforms can be between 20% and 60%. External analytics, on the other hand, judges the whole purchase journey across all traffic sources more fairly.
Why use a combination of GA4 reports and creative evaluation from the Meta API?
In GA4, unfortunately, we all have to wait for the data, especially the attribution data based on UTM parameters, usually 1 to 3 days depending on the size of your project. In the Meta Insights API, campaign data is available at a similar speed as directly in the Ads Manager interface. It lets you make some decisions faster. Even so, you expose yourself to the risk of choosing the wrong winner, because Meta will inevitably over-measure conversions.
Sources
Beshara, J., 2023. Creative is the new targeting: How to succeed with Meta Ads. [online] Madgicx.
Fospha, 2023. The State of Ecommerce Marketing 2023. [online] Fospha.
Loomer, J., 2023. How to Use Meta URL Dynamic Parameters. [online] Jon Loomer Digital.
Meta for Business, 2023. About Advantage+ shopping campaigns. [online] Meta.
PeckaDesign, 2023. Why do data from Facebook and Google Analytics 4 differ? [online] PeckaDesign Blog.
Sava, A., 2023. Conversions API: The Ultimate Guide for Meta Ads. [online] MeasureSchool.
Taste, 2023. Creative is the only thing you can still influence in Meta Ads. [online] Taste.cz.