Karoline Grawski, Pexels

Why clean data is essential for your AI-driven marketing success

Maintain high content and marketing data quality for effective analysis and insights

Imagine you can use AI to transform your marketing and content strategy overnight without the help of an expensive analyst, only to be stopped or misled by a hidden obstacle: messy data.

This actually is the least sexy part of content, marketing, and AI: data. Yes, I know, this data stuff is boring, yet it’s quite essential to get in place before you can steal the show with AI.

Imagine you want to…

• Analyze the data of your e-commerce webshop to predict the orders in the next holiday season
• Analyze how your previous years’ marketing and content planning affect ticket sales to predict when to run your campaigns and what messaging you should use

The importance of data governance

Below, I’ll explain the indispensable importance of using clean data to feed AI for these analyses.

All captured in ‘data governance’ with

• Security or your data; privacy, confidentiality, and GDPR
• Quality of your data, for example, standardization in data formats

→ If you have dirty data, it’s too easy to f*ck up your analysis and get to wrong insights to make your management decisions without you noticing it.

From my own experience, doing a content review for one of my clients, I stumbled upon ‘dirty’ and ‘schizogonic’ data that took me hours to clean up and manually adjust before I could make any analysis with AI.

This made me realize that having data is one thing; being able to analyze it with AI is something else that requires clean and compliant data.

We’re not yet aware of how much work it takes to clean it up, let alone that your data needs to be cleaned up before you can use it for valid and representative insights and analysis. At least, I didn’t.

Honestly, we’re all ecstatic about AI. And we get off on data. How did our content perform, what is the attribution to the business objectives, and what is the greater trend behind the superficial vanity metrics? How can you compare this year against the previous years and make a prediction?

Cool.

But note, shit in is shit out.

If you don’t have your data governance in place, you’ll never be able to get valid and representative analysis, no matter how fancy your AI is.

Here’s the boring data stuff that you need to know as a senior content and marketing manager.

1. What is data governance? And 2 tangible examples
2. Why it matters
3. The urgency to start today

Let’s get started.

1. Unlock AI’s potential through data governance

I know it’s boring, and governance feels far-off and remote from marketing, yet if you want to use AI for marketing and content, it suddenly becomes very relevant and essential.

You can’t ignore data governance in AI. It’s just too important, and you’re legally obliged to have this in place.

In short, data governance ensures the management, quality, and security of your data. It enables consistent, accurate insights for decision-making, personalizing campaigns, targeting and retargeting audiences, and compliance with data privacy laws.

For the sake of simplicity and clarity, I will use examples from my own experience in content marketing. Be aware that data governance is about much more than just formatting the data fields. It is also about privacy, security, GDPR, ethnics, biases, and much more.

As I’m not a data and security specialist, so I’ll leave the details to them.

What matters for you is that you realize you need to address this before you can use AI to analyze content and marketing performance, get valid insights, and create campaign briefs.

Example 1. Importance of date formatting

You download a .csv file from your content management tool to analyze your audience’s purchases and see if there’s a correlation with your marketing campaigns.

The date is quite essential here. The way the tool defines the date has to be so that AI recognizes it. It sounds easy, but in reality, it’s not, and small anomalies or variations can make a huge difference and mess up your analysis.

The format: ISO 8601, US, EU, textual, compact, with or without timestamp? For the 6th of July 2024, there are at least 18 ways.

A few date formatting examples:

• 2024–07–06T14:30:00,
• 06–07–2024 14:30:00,
• 07–06–2024,
• 07/06/24,
• 06–07–2024,
• 06–07–24,
• 06 July 2024,
• July 06, 2024,
• 20240706, and so on.

Pick one and stick to it. Make sure that all data you feed into AI has the same format.

Example 2. Financial number formats matter

Another example is the way the purchase amount is written. Unfortunately, a .csv imported data field with ‘4,900.00 EU’ is not recognized as a financial number by Excel, and probably neither by AI.

Simply manipulating the column and assigning the column properties as ‘financial’ doesn’t do the job. Splitting it doesn’t do it, either.

The issue is that my EU Excel didn’t recognize the US way of writing numbers, and it tripped on having numbers and characters in one cell. The more because the date format in that same .csv file, see above, was written in an EU format.

It took me two hours to figure out why I couldn’t use this column for analysis, which, at first glance, looked good. Every query and super simple request I made returned an error.

Solution: standardize your data governance

Make decisions on how you format the output, from date to finance to ways of writing numbers and using separators. I don’t care which one you choose, just choose one, document that, and stick to it.

If you use multiple tools that all have their own quirks, make sure you clean up and standardize the output before you upload these files to your secure AI. Or have a manual process in Excel.

If you don’t have a data analyst or expert team in place, it’s you who has to do this shitty job. And that takes time, a lot of time. And with meticulous precision and an eye for detail.

Solution: think for yourself

You need to keep thinking for yourself and be critical of what you see. Don’t believe and assume everything AI tells you.

• Stay sharp and use your experience to sanity-check AI output. Does it make sense what you see? Is it in line with what you expected or totally opposite? Never waste a chance to double and triple-check your input for the validity of the output.
• My tip is to check what you see with your co-workers as well. The four-eyes principle. Together you know more.

2. Why it matters, the crucial role of data governance

As already stated, data governance seems boring and irrelevant to marketing. However, without it, you risk making management decisions based on incorrect insights or analysis.

Not even to mention the legal obligation of privacy, security, and GDPR compliance of your data management.

So, take care of this and handle this data governance from a marketing perspective if you want to have representative and valid analysis and insights for

• your content marketing strategy,
• targeting strategy,
• personalization strategy,
• how, where, and how much you spend on paid media

Besides the GDPR stuff, data governance also standardizes the formatting of your data. You don’t want AI or Excel to wrongfully understand your input or data; mistakes or misinterpretations can slip in unnoticed.

Especially if you are a marketing manager with one million things on your plate, and you’re not a seasoned analyst, these mistakes happen more often than you realize.

Solution: get your data governance in place

Again, the solution is data governance. The chapter above explains what it is, what to do, and who is responsible.

3. Act now, the urgent need for data governance in AI

There are two main reasons for this urgency: legal and time.

• To start, your legal obligations, such as privacy and GDPR, are important. If you want to use AI and data to, e.g., predict risks of non-paying customers or the volume of returns in a webshop, you have to protect your customer data and have rules in place to prevent any bias based on name, race, gender, or age.

Like everything in life and for big projects, it’s about ‘think slow, act fast.’ With all the GDPR regulations and growing AI and privacy concerns, you will also have a legal obligation to have your data governance in place if it’s not already there.

• You need time to research and define your problem and objective, create a strategy, and a plan to fix it. The more you think about it upfront, the faster you can act. If you start acting immediately without a solid plan, it’s guaranteed that your project will fail, delay, and exceed the budget. Next to the thinking time, you need time to test, iterate, and improve from data output to valid AI-generated analysis you can use for your content marketing strategy and planning.

Solution: stakeholder management

I’m sure you’re convinced that you, as a marketing manager, need to dive into the data governance before you can run AI on it.

As the boring data governance stuff is probably not on top of every CMO or marketing manager’s list. The only way to free up your and your team’s time is to make it a priority in the C-suite. Because, without the data and data governance in place, you can’t use the sexy AI to improve your content and marketing quality and speed up the processes.

Solution: get the experts on board

And realize that data governance is a specialty for a reason. It’s important. Not only for your marketing, maybe even more for the security and confidentiality of your customers’ data. And that last one is a legal obligation as well.

With the fast growing possibilities and features of AI, it’s very easy to breach this confidentiality. Unwanted and unintended. Accidents and mishaps can happen easily.

Key takeaways

Your data enables consistent, accurate insights for decision-making, personalizing campaigns, targeting and retargeting audiences, and compliance with data privacy laws. All while you monitor your data quality and ensure its safety and security.

You need to be aware that the data output of your tools influences the quality and validity of your AI-generated analysis.

Shit in, is shit out. You need to step in, check, and adjust where needed.

The solutions:

• Get your data governance in place to ensure the quality of AI-generated analysis. This is also a legal obligation. Make this a priority in the C-suite. And if in doubt, don’t use AI. You can’t un-upload your data from AI
• Let the experts work on your data governance; this is not a side hustle for the marketing team
• Start today, as you need time to plan, test, iterate, and get things in place, and AI development goes fast as f*ck

Now, it’s up to you to unlock AI’s full potential with clean data to get insights to improve your marketing and content strategy overnight without the help of an expensive analyst.

🤍 If you enjoyed this, follow me for more content on strategy, branding, and innovative marketing insights.

Stay ahead: subscribe to my newsletter

If you want to stay up to date on AI, content, and marketing based on insight and how it matters to you, subscribe to my LinkedIn newsletter.

Two-click subscribe, and one-click to unsubscribe if you don’t like it.

Easy does it.

→ Subscribe

4 Tips to use AI for content based on 1,5 yrs ChatGPT

LinkedIn algorithm secrets, what works best on LinkedIn

3 Tips to use AI and improve your output instantly

Make your content stick, repeat it 5 to 7 times

Writing for busy readers; how to simplify your writing