Reading Time: 9 minutes
In our recent webinar, Top 5 Signs Your MarTech Stack Is Not Future-Ready, we had Meri Williams as our guest speaker. You can watch the event on demand now (#shamelessplug – but I think you’ll find it interesting). As the title suggests, Meri talked about the top red flags that businesses should be aware of in their current martech stack, drawing examples from her time as CTO at Monzo, MOO, Marks &Spencer, and Procter & Gamble. It was incredibly enlightening, not only because of her extensive tech experience but also because it’s not often that you get to hear a CTO’s perspective on marketing tools and their side of the equation with respect to martech investments.
And so right after Meri’s webinar presentation, we asked her a few follow-up questions to really delve deeper into her thoughts on culture, team collaboration, data silos, system integrations, AI, APIs, and her advice for tech and marketing teams when it comes to choosing the right martech stack for faster ROI.
If I were giving the marketing side of the equation a bit of advice, it would be to learn enough about the tech to be dangerous. Learn enough that you can ask good questions with your tech team.”
Meri Williams is an experienced CTO, advocate, and leader for a few technology organizations. She is currently the CTO of Pleo, a leading SaaS Fintech business offering businesses smart payment cards for their employees, after serving as CTO at Monzo, MOO, and Marks & Spencer.
Meri has led teams ranging in size from 30 to 300 people across a variety of industries and organisations, including Procter & Gamble and the Government Digital Service.
She is a published author and international speaker who co-hosts and curates The Lead Developer conferences in London, Austin, and New York. Meri is also a trustee of Stonewall, the UK’s leading LGBTQ+ rights charity.
As a CTO, what do you look for in a tech platform?
So I think the primary thing I look for is how flexible the solution is! Is it able to adapt to business needs? Because if something’s very inflexible, it tends to take a long time to implement, and then it takes a long time to make the changes as you keep going.
I look for how good the APIs are, which I appreciate as a sort of nerdy technical thing to talk about. But think about that as just how well can it connect to other things? And one of the direct questions to ask your technical colleagues is, you know, do the APIs on this look good? Do you think you’ll be able to connect it to our other systems?
I look at whether the integrations are there and whether the software can integrate easily with the other major platforms that we have, if that’s necessary. And then how easy it is to bring data out of that system.
Most of the organizations I’m in—a little bit bigger than, you know, early-stage startups, they tend to be scale ups and larger, and so they tend to have their own data platform of some description, and so making sure that that tool can be useful on its own and has sufficient analytics so that you can make good decisions within the tool, if that’s appropriate, but also so that you can pull the data out and then have it in your broader data infrastructure is really important.
How does the tech team handle marketing requests for shiny new tools?
The mantra that we tend to have in tech is that it’s better to configure than to customize.”
I think there’s a real risk that the tech folks, or the IT department, or whatever we’re called in your company, we can become “chief saying no officers”, and I think that’s not a great outcome for us. It’s not a great image for us to have, but it’s also because we’re very aware of the risks and costs of implementing new services.
And so, I think, when what you want to avoid is having to do an integration project or a transition project every year or two, not least because you then struggle with having consistent data and history and the ability to measure over time, the ideal is to find a solution where you can see that they’re continuing to invest in keeping it current on top of what the trends are in the market in terms of what best practice looks like. And the solution itself is quite flexible.
The mantra that we tend to have in tech is that it’s better to configure than to customize. Configuring is when you just sort of tweak the options that are available in the software itself. And customizing is when you have to write special code to make it do something that it’s not designed to do. And the more of that customizing that you do, the harder it is to keep things up to date, and the harder it is to keep things working in the face of changes to adapt to the future. And although it feels like a great idea because you can make the solution do exactly what you want, it costs you a lot in ongoing maintenance and adaptation.
It’s better to find software and a tooling choice where you genuinely believe that they’re going to keep up with the industry. They’re going to keep offering the latest capabilities as time goes on.
Why does data silo persist despite all the tech that claims to fix it?
It’s about really strong collaboration between the vendors that you use, your own company, and your tech team within your company…”
I think there’s a couple of things.
A lot of vendors are trying to become the single source of truth. If you choose three different systems that are more focused on being the one place that you go than they are on integrating well with other things, then it can be very problematic because you’ve got what is essentially a kind of system fight happening in your company.
Something that’s really positive about the new range of typical software-as-a-service options that have become available is that they’ve accepted much more that they need to integrate well with other solutions and that they need to let data in and out very, very easily in order for you to be able to get that single source of truth in one place.
The mistakes that I see people making are not being clear about where the single source of truth should be and which systems should be like the golden source of information and not caring enough about APIs, integrations, and how things connect to each other. It is solvable, but it’s about really strong collaboration between the vendors that you use, your own company, and your tech team within your company, because most data platforms do have a way of bringing in lots of different data sources, and they can then use those to give you a single view if you have a data platform. But if you don’t have one, then it’s really important to choose a tool that is going to integrate well and that’s going to bring in those data sources from elsewhere so that you can have everything in one place.
It is also partly a cultural thing about collaborating really closely with your technical colleagues, and then partly a decision about the tools that you actually choose and how well equipped they are to reduce those silo effects.
How can marketing and tech teams collaborate better to achieve a common goal?
What I often recommend my technical colleagues do is to make friends and do a skill swap with somebody who’s in a different department. I’m sure that there is a technologist in your company who would love to understand more about marketing and would love to understand that world a bit better.”
I’ll talk more with the tech side of the equation about how they need to adapt. What I tend to say to them is that they need to learn to speak marketing. They need to understand what matters to a marketer. They need to understand how they can meet those needs, what the systems do, how they fit together, and those kinds of things.
And if I were to give the marketing side of the equation a bit of advice, it would be to learn enough about the technology to make it dangerous. Learn enough so that you can ask good questions. You can have a conversation with your tech team about how good the APIs are or aren’t and have good conversations about it. Like how we’re going to make sure this data doesn’t end up trapped in one system and not connected to all of these other places. Which integrations do we need to worry about, and which other systems do you want to be able to cross-reference with? If you really need to be able to know the end impact of your marketing efforts, maybe you want your systems to integrate all the way down into your finance and billing systems, because then you can. Then you can directly see whether sales went up or down, and so on. It’s not always necessary to go to that level in order to integrate, but the main advice I would give is to learn enough of the technology and speak to be able to engage with your colleagues in a really positive way.
What I often recommend my technical colleagues do is to make friends and do a skill swap with somebody who’s in a different department. I’m sure that there is a technologist in your company who would love to understand more about marketing and would love to understand that world a bit better. It’s really, really worthwhile to form that alliance and do a bit of a skill swap, and if you can get one of the people in the tech team to really understand marketing in a really good way, and they can teach you enough tech to ask great questions when you’re evaluating these kinds of tools, then probably you’ll get to the point where you’re bringing a better subset of tools for that real consideration. You’ve already checked that there are API checks that the data is not going to be siloed; checked that there’s enough flexibility in the processes and the actual kind of internal workflows for what you need.
I think it also builds a real bridge if you can have that shared language or learn each other’s language well enough to do really well.
What are your thoughts about AI and its impact on business?
We’re better at emotion. We’re better at speaking to the heart. We’re better at enlisting emotional responses from each other. And there’s a real risk that we end up with just a lot of stuff being written by AI.”
There are a couple that are really interesting. I’m actually from an AI background, so there’s a risk that I will, you know, spend the next half hour just talking about this. Do stop me when I get too excited (as she said with the biggest grin on her face!). I think there are some really interesting things that have happened, and recently, even before Gen AI came along, we had a load of improvements in machine learning.
I think traditional machine learning is super relevant to the more technical conversation because you have so much data. And when you have a lot of data, traditional machine learning is probably the best thing to use. It’s basically a way of taking a data set and learning from it on a really large scale. So computers can process a volume of information that’s just impossible for a human brain to do. And so the one bit of advice I’ve got is: don’t forget about that traditional machine learning. If you have a lot of data, you have more than 5,000 data points. Probably machine learning is the more useful thing for you to invest in and look at, and it’s very worthwhile. And there’s a lot of value there.
On the generative AI side, I think it’s very useful. I think it’s got a lot of opportunity. But a lot of people also misunderstand what it’s good at. Gen AI, the large language models, OpenAI, and similar like ChatGPT sound like a really clever person, a really erudite person, and because of that, we tend to go, “Oh, I can trust this with the sort of decisions that a very smart person would be able to make.” But actually, it’s good at sounding smart but not good at being smart.
I think that what we’re going to see is that people are going to realize that AI is really good at summarizing and synthesizing information. Taking in a load of disparate stuff and then giving you a good summary. And it’s good at giving you a starting point, but it’s not going to be brilliant at emotional content.
I don’t think that AI is suddenly going to write better ad campaigns than marketers can. Because there’s a human element to that. That’s really important. I think that the best attitude towards generative AI is one that I have with a friend who did a presentation recently about how they started out as translators and were very proud of having gotten their English degree (they were from mainland China), and sure, that they could give much higher-quality translations than anybody else but then Google Translate came out. And suddenly, they were slower than everybody else. And that was their real problem. And so their attitude changed. Instead of seeing Google Translate as a real risk and a threat to their business, they thought, What if they could use it? But then also add the fact that they had higher quality, so they used it for speed and then added the human touch on top.
I think that’s what I think is going to happen with AI: that we’re going to start using it to make our jobs more efficient and easier, and we need to use it for what it’s good at. And then we also need to remember the stuff that is much better coming from a human. I don’t think we’re going to suddenly find that there are lots of AI-created advertising campaigns, TV ads, or similar. I think we might use AI in the creation of them. I think that we have to be cognizant of the things that humans are better at. We’re better at emotion. We’re better at speaking to the heart. We’re better at enlisting emotional responses from each other. And there’s a real risk that we end up with just a lot of stuff being written by AI. That’s then summarized by AI before it’s read by a human, which I don’t think is going to be a bright and wonderful future. Try to use it rather than be used by it, and become more efficient using AI. But remember that things still need to be human and have that human touch.
Thanks to Meri for taking part in this conversation with us! If you want to read about Meri Williams’ top 5 signs your martech stack needs an update, here’s the link for it.
