Taking on the Media Silos

TikTok has been the latest big tech media platform to release its own attribution measurement platform and like Google and Meta before it it represents yet another siloed approach to marketing measurement. It seems an extraordinary presumption on the part of the media giants that advertisers will only want to measure activity on their platforms, especially when the whole point of attribution analysis is to look at the entire customer journey. No doubt each of these media players prefers to think of consumers as primarily using ONLY their channel to discover new brands and make purchases, and would like brands to think the same way too.

However in fairness to them, the media platforms really have no choice but to retreat into their own walled gardens, with increased focus on privacy and the decline of third party cookies. Even when cross channel tracking was permissible the challenge was super hard – both Meta and Google have launched multi-channel attribution solutions in the past and then had to withdraw them. Servicing these super complex cross channel measurement solutions was never practical for a technology first company.

Right now, Google’s Data-Driven Attribution in GA gets closest to a truly multi-channel solution, but this faces huge limitations such as measuring only clicks not impressions, zero model, transparency and patchy reporting in the new GA version 4, which mean it only really works for brands focussed on a click only performance view, content to trust Google to best evaluate the performance of Google PPC.

How Should Brands Take on these Silos?

Naturally the answer to this question is not exactly the same for all brands. To take on siloed measurement and make the best possible  data-driven marketing investment decisions, we recommend brands consider their needs in the following three key areas –

  1. Marketing Mix
  2. Expertise
  3. Independence

The Need for Marketing Mix

The basic question brands need to ask is what channels their target market is using? The media strategy then has to answer how these combine together to drive the whole process from awareness, through to consideration, research and final purchase. The measurement strategy needs to align with this media strategy in order to get the spend mix right and avoid massive wasted budgets and missed opportunities e.g. to minimise CAC for a given sales target. 

Why not stay in your silo?

One option for brands is to simply go along with the siloed view, especially if they are mainly single channel advertisers anyway. So a small and growing D2C brand may decide that it makes complete sense for them to focus primarily on Tiktok and the latest measurement release would seem like a sensible option. However, to go down this route a brand needs to be fully confident that the attribution model applied is accurate and that they can trust the media channel to report in this way (also see the ‘Independence’ requirements). 

As with Meta, for TikTok the biggest challenge is measuring ‘view-through’ sales, and selecting the right look back window for this. Not every ad scrolled past influences a sale, so a look-back window is used to limit the period of influence allowed. Yet the time window selected is completely arbitrary and does not reflect the reality that brand building influences over longer periods and does not guarantee a post view sale in any time window. At best, you get a relative sense of campaign and ad performance on the platform, but no true sense of incremental sales. In Google Ads, PPC has its own issues e.g. how much brand search is truly incremental is a common concern. 

In any case, the marketing mix that brands consider should not only include channels that the brand is currently using but also the potential channels which the brand should explore. Even for the simplest ‘one click’ impulse purchase brands, investment in one channel will eventually start to hit limits whereby the easiest target customers are all being reached and diminishing returns set in. New channels offer access to more customers and so more headroom for growth. 

So can alternative siloed platforms just be treated independently?

Of course this is what many brands do in practice. They spend on each channel, and use the channel platforms to measure conversions. But this gives rise to new massive challenges. The first is the risk of duplicate sales. Tag management seems to fix this, but in reality applies a tracking fix in order to hide a measurement problem (the duplicates go away but you still have no idea what really drives your sales!). The second challenge is, how do you actually compare the numbers between platforms? Even assuming that every sale follows a single touch journey (never-ever true!), the attribution rules vary across platforms and the channels themselves work in completely different ways. Without a sound attribution method you are forced to fudge the numbers for a combined view, usually in a spreadsheet, leading to sub-standard marketing mix spending choices. 

Conclusion: you need an attribution platform that unifies data from the different siloed platforms. 

The Need for Expertise

With this in mind brands often decide to tackle the silos as primarily a technology issue to solve, working with IT/ data colleagues, and they might choose to sign up to a SaaS attribution platform solution in the hope that this will fix the problem. The platform sales team talks up the impact and then a technical team links up most of the data and shazam, a series of shiny new dashboards appear with non-duplicated attributed sales numbers for each channel which are different to the catastrophic ‘last click wins’. After some training the team is now ready to go. And then they don’t use the platform. The reason? The problem of getting to data-driven marketing is not mainly a technology problem: it’s a people problem.

Who decides what action to take and on what data?

Problem one is that now you have more joined up measurement, what action should you actually take? The CMO is way too busy to look at any platform data in detail. Their question is – tell me what actions can we take, given this data, to hit sales and other KPIs targets at the lowest marketing cost? Someone smart is delegated the job of answering this question.

But who? Data analysts, marketing analysts and data scientists are in very short supply. The people who are smart enough and have the skills are typically overworked and are rarely given the support they need to achieve their full potential. When they do get involved, these people have difficulty trusting an attribution platform, because they know how messy the data really is underneath and how complex the measurement challenge is. Right away there are adjustments to consider and questions to answer. Why are there differences between platform numbers and transaction numbers? How is long term brand awareness building being included? What about margin variations between products? Lifetime value? New vs repeat customers? Offline sales? Marketing experiments? Incrementality?… the list goes on.

And what about the problem that the past is not always the best guide to predicting the future? Does your marketing mix recommendation solution use any predictive technology that addresses this fundamental challenge?

A data issue or a people issue?

What is really needed is an extended process whereby the measurement you take on is adapted to actual business challenges and data, but now the measurement platform is in place, the job of optimisation still remains and it’s still the brands job to try and fix it. An SaaS attribution platform is typically ‘Do-It-Yourself’ because that is how the SaaS business model works – everyone gets a template suite of reports and a generic model and they have to figure it all out themselves. There are 101 questions about what it all means, and about the bugs which inevitably appear, and only a few lowly remote help desk workers to help fix it.

The bigger picture is that businesses today are not short of data, they are short of people who know how to use data to make smart decisions. If the right people are on board, then huge progress can be made. A further challenge is that the very best people tend to get sucked into bigger projects or worse still, leave for better paid jobs elsewhere.

The solution is to focus on a data driven culture and make sure to have a team of people who can work on your unified attribution data, and not just one overworked hero analyst. Most importantly make sure they are fully supported. When you sign up to a SaaS attribution solution that promises to fix everything, ask yourself how much genuine expert support you will get to make sure it is relevant to your business structure and KPIs? Also consider what will happen after onboarding, what quality of support you will get then to help you use that platform to actually take action?

Without the right people who can help you find the key insights and make recommendations, even the most robust analytics solutions have no value.

Conclusion: be wary of magic bullet SaaS tools – think about what expert help you need to actually ensure results are relevant to your brand to help pinpoint actions.

The Need for Independence

With marketing and analytics expertise in such short supply, many brands take the obvious and easy way out. Delegate the entire challenge to their marketing agency. The agency has to hit targets anyway, and has an in-house team who want to make it work, so it seems to be the ideal solution. Someone in the marketing agency can use the attribution tool, or maybe their own tools, to make sure all numbers add up and then make recommendations on how to spend the money. Job done!?

Now agencies vary – there are some truly excellent agencies out there as well as some truly dreadful ones. However, one fact is baked into the agency model, that marketing agencies are incentivised to tell you big success stories about their marketing on your behalf, even if the story is that a decline in sales would have been so much worse without their efforts. While they do want to hit your targets, you can never be entirely certain that the results they achieve are the very best that they could be in the circumstances.

Agencies are also incentivised to minimise the costs associated with servicing each brand. So given a choice between a very robust attribution method applied by an expert data scientist, and a cheaper DIY SaaS tool run by a junior analyst serving several clients, they will tend to incline towards the former.

Marking their own homework

It comes back to the problem with the data silos themselves – you cannot rely on the people spending your precious marketing budget, to always give you the whole truth about how well they are spending it. This is called ‘marking your own homework’. The TikTok measurement solution will make TikTok look great, just as the Google Ads platform makes Google look great, and the Meta platform, and the Affiliate platforms… all look great. And your multi channel agency will tell you their channel mix strategy is just, well great!. It’s how the whole ecosystem works.

The worst part is that the big media platforms now have so much leverage and market presence, they are much more powerful than the agencies, and use their market position and technology dominance to co-opt agencies into their way of thinking. Many marketing agencies get preferential terms and agency specific solutions in return for adopting the media platforms measurement framework and promoting formats and strategies that benefit the platforms more than the brand advertisers they serve.

So, until you achieve fully independent marketing measurement, then you will always run the risk of only hearing what media owners and agencies want you to hear.

Conclusion: when you rely on external experts to help measure marketing spend, make sure they are fully independent of the people who are actually spending it.

Summing up

Today, breaking out of a siloed view of marketing is harder than it has ever been. It is easy to be tempted by simplistic solutions offered by the measurement platforms who want to present their channels as the best, by SaaS platforms who want you believe you only need their one-size-fits all technology to make decisions, and by ad agencies who are adept at spending your money for you and spinning the most optimistic story about their choices and why you should spend even more.  

Mindful of all these pressures, brands should face up to the challenge of breaking out of the media siloes, invest in a robust and independent approach, and make sure to fully leverage the best expertise they can find to make sure it counts. 


Gabriel Hughes 2023

For more on this and related topics, check out our free guidebook on marketing measurement.

If you want to use the only fully-custom + fully supported attribution platform available in the UK, then please just get in touch: hello@metageni.com