How to not use third-party data but still make better media buying decisions

As third-party data becomes a thing of the past, what steps can you take to determine advertising budget allocation? It could be time to welcome back Marketing Mix Modeling (MMM).

It’s been a tough few years for marketing teams. Between Apple’s iOS 14 update and Chrome removing cookies, the days of easy digital tracking are over. Without having a stream of data showing what marketing channels are the most effective, it’s become very challenging to know how to allocate a marketing budget.

So how will marketing teams track the success of their various campaigns without access to third-party data? We could find the answer in the return of an old friend called Marketing Mix Modeling (MMM)

What is Marketing Mix Modeling (MMM)?

Marketing mix modeling sounds complicated. Additionally, many of the methods it employs require a working knowledge of statistics. But the concept itself is pretty straightforward.

Essentially, MMM uses statistical modeling to understand the impact of the different channels that make up their marketing mix.

For example, a marketing team might allocate their budget on:

  • social media marketing
  • email marketing
  • TV and radio ads
  • content marketing

MMM offers a way for teams to understand which factors result in sales. That might sound a bit like Attribution Modeling. In some ways, it does a similar job. However, it does this without tracking online behavior.


What attribution models are out there now?

There are several ways that companies track their digital marketing efforts.

You can split them into four categories:

  • Surveys
  • Tracking
  • Modeling
  • Experiments

Let’s explore each one briefly.


Surveys are a great way to get valuable data about marketing. Getting this data can start out simple with a few quick sales calls or more formal interviews.

However, for better data, you can use how you heard about us (HDYHAU) surveys, panel surveys, and eventually, third-party consumer data.



Customer tracking can be done in several different ways. Some of the options here are tracking pixels and URL parameters. However, there are other methods that you can employ, like designated discount codes. For example, if you advertise on a podcast, the host will give a discount code which users can enter if they make a purchase. Simple, yet effective.

There are a couple of other more sophisticated methods, like the dark arts of server-side tracking. Finally, we have the tracking king: customer matching. Essentially, this tracks logged-in users across all their devices, generating excellent first-party data.



Attributing sales to specific ads isn’t easy. Attribution modeling — like last-click modeling — isn’t sophisticated enough to account for the various touchpoints that drive a deal.

If you have various marketing channels, statistical modeling becomes necessary to get at the truth.

Some options are device fingerprinting (shady) or behavior-based modeling, which uses some user’s data but fleshes out the unknowns with machine learning.

Other more advanced options here are competitor activity, causal inference, and, of course, marketing mix modeling.



If you want solid evidence of marketing effects, you need to get scientific. These methods can help you get at the truth of which channels are producing results.

Some tests are pretty straightforward, like A/B testing and geographical testing. However, there are many more complex means, such as deprivation testing (toggling a channel on and off to see its effect) or randomized controlled trials, AKA lift testing.


The attribution model has been broken for years

Many marketing teams are mourning the loss of third-party data. However, it ignores one key fact. The attribution model has been broken for years. The iOS 14 update just removed the input, but the actual model itself was already fairly unreliable.

When you track a user’s journey, it involves several different points. Some of them might be recorded, but often key parts are left out from attribution modeling.

For example, a user watches a Webinar on YouTube or LinkedIn. However, they don’t take immediate action. Sometime later, they remember your service, go to your site, and get counted as a Direct conversion. YouTube gets no attribution, even though it was the main driver.

There are lots of other little inconsistencies that spring up. For example, some Influential channels aren’t tracked, but others are and get the credit. And that’s before we start thinking about Dark Social, like WhatsApp or Slack, which are likely playing a crucial role in marketing too.

What if a friend or workmate sees an ad for a solution and DMs you the name? You buy the product, but the ad doesn’t get credit for the sale. If this happens enough times, you don’t really know which channels are driving sales, so you allocate the budget inaccurately.

Regardless of the iOS 14 update, you need a more advanced, more accurate model if you want to know how to make better media buying decisions.


How to combine attribution models to get more accurate data

There are several different attribution models. However, not all of them produce the same output. Quite often, there is a large degree of disagreement between them.

But who do you trust, Google, your analytics, or your marketing mix?

One solution is to use each source and triangulate them to get more accurate data. Any model you build could have inaccuracies, but by comparing each source, you can gain insights into the areas that aren’t working.


How marketing mix modeling can help

Marketing mix modeling can offer a solution by tying your data sources together. Marketing happens across a wide range of channels, both online and offline. Attribution models often struggle to create an accurate representation of the authentic traffic sources, which is why MMM is having a resurgence.

MMM doesn’t have the granularity of other digital marketing attribution models. However, with the right tests and experiments, it can produce reliable and actionable insights to inform marketing spending.

Privacy concerns and the growth of ad-blockers have damaged the viability of digital tracking. MMM offers a privacy-friendly alternative.

Here are some of the pros and cons of an HMM approach.

Pros of MMM

  • Flexible
  • Privacy-friendly
  • Measures omnichannel marketing
  • More accurate at scale

Cons of MMM

  • Lacks the granularity of other methods
  • Time-consuming to set up
  • Requires a lot of statistical expertise
  • Reliance on human input could lead to a level of human error.
  • Less real-time than other attribution models



Digital tracking attribution models have been hit by iOS 14 updates and Chrome’s intention to remove cookies. Without third-party data, many businesses struggle to understand how they can measure their marketing channels to decide on budget spending.

However, there are several alternatives available, such as marketing mix modeling. While these models lack the granularity of digital tracking, they are more accurate at scale. Additionally, as user privacy grows, MMM could provide the most ethical solution.

Related content

Digital marketing is one of the most exciting and fast-moving sectors in modern business. With great rewards available for teams that outwork their rivals, it’s no surprise the space is stuffed with innovation. Here are the biggest digital marketing trends and insights for 2024 and beyond.
In February 2023, Microsoft announced that its Bing search engine would implement ChatGPT. The announcement caused quite a stir and forced the technology community to rethink how chatbot-powered search would change our relationships and reliance on computers.
AI has opened the doors to incredible marketing opportunities in marketing and beyond. However, like any innovative technology, it also has the potential for misuse. Brands and agencies need to understand the ethical questions that come with using AI for digital marketing.

In this article

Stay updated

Get on the (email) list!

Stay updated

Join our newsletter