Likely your data set is not ready for it. Just not yet.

Did you happen to see the Netflix documentary about Fyre, the once in a lifetime luxury music festival in the Bahamas?

In case you missed it, the overhyped festival, a supposed mix between Coachella and Burning Man, ended up becoming “The Greatest Party That Never Happened”. Everyone was talking about it, it was overhyped, people bought tickets driven by “Fear Of Missing Out” and the organisers had little or no experience with organising such a large festival.

Why is this relevant for a blog on the website of the Digital Analytics Summit you might wonder? Well, there are actually many parallels between the current state of AI and the Fyre Festival. Everyone is talking about AI, it is a bit overhyped, there is a lack of experience and marketers are suffering from “Fear Of Missing Out”.

But there is no need to be afraid you will miss out on AI. Conferences like the Digital Analytics Summit and blogs like this one, give you valuable tips to help avoid making the same mistakes as others made before you.

Recent data collection challenges

With online marketing being a significant and indispensable driver for traffic to any webshop, the importance of data is growing year after year. In the “State of Martech 2024” report by Chiefmartec 71% of the respondents (leading martech and marketing operations professionals), reported that they’ve already integrated a data warehouse within their martech stack.

We all acknowledge the importance of data, and we all have to continuously navigate between collecting valuable data on the one hand while still respecting the customer’s privacy and being legally compliant on the other hand.

In the recent past we have seen businesses migrating en masse to server-side tracking (whereas traditionally behavioral data was captured in the browser of the website’s visitor). The goal was clear, to improve data quality and secure it for the longer term because server-side tracking is not affected by adblockers and tracking prevention settings in browsers.

More recently we have seen almost every website implement Google’s Consent Mode to ensure GDPR compliance while still being able to track valuable data for insights.

Is your data tracking ready for AI?

AI models rely solely on clean and complete data flowing into them. Without correct data, any model lacks the fuel to deliver an intelligent prediction. Therefore your data collection process (which is the foundation of any data driven strategy) should be your top priority. Without validated and rich data, you will never be able to be successful with AI.

So in order to ensure the effective execution of marketing AI you should never forget that garbage in, results in garbage out. This applies to the simplest dashboard in which you monitor basic KPI’s and even more so to AI. Therefore, before starting with AI you need to make sure your data tracking is in order and works permanently. The DataLayer and all tags must do their work well at all times in order to capture- and pass on the right data.

When your tracking works perfectly it will have a significant positive impact on the costs of your first AI project. The correct tracking setup contributes to the data quality and structure and it will eventually save your data scientists lots of valuable time. Instead of wasting time on checking, cleaning and preparing the collected data they can actually spend more time on developing and improving the AI models.

A well working- and well documented tracking setup contributes to transparency and helps explain the AI model’s logic and its predictions to stakeholders or even customers when needed.

Real-time monitoring: Let the AI party begin!

There are some tools available in the market that help you with checking your data collection set-up. What most of these tools don’t do however, is real-time monitoring of the most vulnerable parts in your data collection.

You don’t want your artificial intelligence capabilities to be dependent on the need for human checks. To boost your AI efforts, you will be far better off with a real-monitoring tool which alerts you instantly when something is off and also states in detail what the actual problem is.

Monitoring your dataLayer, tag manager and tags in real-time ensures a constant stream of reliable data at low costs:

AI will be working at full speed and full power without any required human interference;

There is no longer any need for periodical and time consuming manual (human) checks of the tagging setup;

When new code is conflicting with other content or objects it will be instantly detected;

Valuable time for debugging is saved by automatically pinpointing the exact cause of an issue and reverse engineering the setup;

You will have full control over the tracking on your platform and maximise the quality of the data you collect.

Conclusion

Despite the many challenges and potential pitfalls, the successful implementation of marketing AI is surely feasible.

Data quality is key because any AI model needs good quality data. It is the fuel needed to deliver intelligent predictions. High quality data will save time and resources for data preparation and maximise the chances of success with AI. Therefore data quality assurance and correct tracking should be your first priority.

With real-time monitoring of your tracking set-up you have a permanent safeguard in place to protect your data quality. Real-time monitoring of your data collection process will pave the way for being successful with AI.

Why you need to monitor your tracking- and tagging setup

It is probably the biggest nightmare for online professionals, decision makers and data analysts: unreliable and incomplete data.

Within any organisation, data is the foundation for strategic decisions, solving problems and the ability to run and assess marketing campaigns. We are increasingly dependent on the quality of the collected data to stay relevant and competitive.

Moreover, running online marketing campaigns relies for an increasing part on trustworthy data. Algorithms which are used by for instance Google, Microsoft and Meta work more effectively when you feed sufficient and correct data into their systems.

Therefore it is no surprise that data collection is in the spotlight for quite a while already. Already in 2019 the “tracking rat race” between Safari and Mozilla on one hand and marketers on the other hand was in full swing. Since then many episodes have been added to the saga with most recently the launch of iOS17 and the removal of utm-tracking parameters. With the introduction of new technologies like server-side tagging and the introduction of Data Warehousing the landscape is getting more technical along the way.

Bad or missing data gives a wrong image of your return on investment, algorithms can’t do their job well and ultimately you are wasting part of your advertising budget.

Despite all the attention for data collection, the sad reality in most companies is that stakeholders don‘t even know if and when tags are firing or not. Most websites have extensive monitoring in place for their website infrastructure but the functioning of tag containers and tracking tags are frequently overlooked while they have a significant impact. Correct tagging and tracking, a crucial aspect for collecting consistent and accurate data, is almost always overlooked.

Tracking threats

1. Tag managers allow marketers to inject code into your website without the help of a developer. At the same time those tag managers don’t have a safeguard in place to stop new code from conflicting with other content or objects. You will not be alerted when tags are not loading, malfunctioning, causing errors and conflicts or slowing down your pages.
2. Frequent updates on the site as well as third party technologies potentially cause errors which result in a collection gap of several days up to a few weeks. Issues don’t get noticed till it’s too late and the irreparable damage to the data and data quality has already been done.
3. Dependencies and risks are getting bigger over time due to a growing number of technology partners and tags, each forming a potential point of failure.
4. A lack of process, lack of knowledge and lack of communication between teams, departments and external agencies poses a risk. This risk is even bigger for companies with multiple domains and larger teams.
   Manual online checks of the tagging setup are time consuming and are not consistent. The results are influenced by for example the time of the day, the location, the device and browser. It’s impossible to test all edge cases.
5. You can’t influence everything when it comes to tracking and tagging, for example the endpoint could be offline or an API could be malfunctioning without you knowing.




Everyone has the same struggle

Sometimes you find out by coincidence you have been missing conversions for a specific period and you can only guess how many you have missed based on the numbers of the previous period. During this time you have annoyed all those buying customers with irrelevant retargeting ads. This is money wasted on annoying your customers.

You are not alone in this struggle. Most websites are not being protected against the tracking threats described above. An average website regularly encounters issues with error percentages of 5% up to 25%. Basically you and most of your competitors currently have no oversight or insight into how tags are impacting data collection and the user experience of your visitors.

In the land of the blind, one-eyed data collection currently still is king

You never get a second chance to capture the relevant data of your website’s visitors. Therefore it is fundamental that your tagging- and tracking setup is done in a manner which guarantees a steady and uninterrupted process of data collection. You need to put in place safeguards which alert you realtime when shit hits the fan. This will allow you to act accordingly when something is happening with your tagging, whatever the cause may be.

Currently the one eyed man can still be king. But what if you could be the two eyed emperor? You can have full control over the tracking on your platform, maximise the quality of the data you collect and your marketing tagging will no longer negatively affect user experience.

Besides making sure your dataset is up to par, this will save you a lot of time not having to pinpoint the exact cause of the issue and reverse engineer the setup.

It is possible with real-time monitoring of your tracking and tagging. Automated monitoring ensures a constant stream of reliable data at low costs. Reliable data tracking starts with real-time monitoring and it is the foundation on which any data-driven strategy should be built.

=======================

This blog was first published on the website of the Digital Analytics Summit an event organised by the DDMA, the largest association for data-driven marketing, sales and service in The Netherlands.

About the authors: Suze Löbker and Harm Linssen are co-founders of Code Cube