top of page
  • Writer's pictureLucas Bergmans

Know your numbers and invest smartly.

Updated: Jan 4

For any business, it’s crucial to understand how to get the most out of your Marketing investment. Which parts are working best? How can you change your marketing mix to drive better revenue and profit growth? What other factors are at play outside of Marketing spend, like product range or external factors like the weather. For scale-ups this is both more of a challenge (less trading history, less spend) and less of a challenge (more depth of data, generally more numerate).

I chatted with Dr Grace Kite Kite from Magic Numbers about the challenges of getting a deeper understanding of effectiveness as businesses move into scale-up phase and their mix of spend becomes more complex.


Welcome, Grace. I’d love to hear a bit more about you, your career highlights and specifically what you’re doing today at Magic Numbers.

I got into data because I really want to know – big picture - how the world works. And I want to know it because I want to see how things can change. I guess that's why I studied economics, because it has that idea of ‘how does the world work and how does it change?’ And it also enables you to use econometrics - to say ‘well how does that appear in data, then?’ And ‘is it a real rule that when the price goes up, people buy less? Is that a real-world rule or not?’. Econometrics is the toolkit that comes alongside economics that allows you to test theories and I’ve always thought it’s brilliant, the idea of being able to find out how the world works, and test it with data. I've gone back loads of times to university and learnt more and more. I've taught econometrics and I've done a PhD. But where I've had most joy is in understanding how people work and how they respond to communications and price changes and everything else. How do people tick?  That’s what these econometrics projects we do for marketing people are about.

You've been running Magic Numbers for the last few years. How's that been? And what's your offering?

So Magic Numbers exists to help marketing people use their numbers really, really well and feel confident using their numbers. A good part of that is using econometrics to see how the world works and how your particular business works, and what levers work to grow it. Around that, we do a range of other things, we have an algorithm that looks at search data and we do bespoke projects. Our approach is to get to know the people that work for our clients and support them to use numbers to support change. That means spending time, not making people feel stupid, uncovering the real questions. And quite often what we're answering is something that they've been bickering over for ages. Resolving situations where this person wants the budget and someone else wants it too. These are people that are bickering with each other every day and they don't want to be. So, sometimes you just go right, okay, here's the answer as to which would objectively be the better choice. That's quite often the ‘magic number.’

Fantastic. This conversation is focused around Marketing for scale-ups and start-ups. I was at Cazoo over the last few years and we were really keen to get to grips with econometrics as soon as we possibly could, but we found it really, really difficult partly because getting hold of enough data and having enough trading history took time. By definition, as a start-up you’re doing something very new and different and disrupting a category, so that there's not that stable history of data. What do you find are the biggest challenges for start-ups and scale-ups when they're trying to really get to grips with ROI, econometrics and effectiveness?

We’ve worked with quite a few scale-ups actually. We’ve worked with a furniture scale-up and we've worked with a few in financial services and we've worked with a few D2C’s. The challenges are a bit different in terms of evaluation, but actually, they're some of our favourite clients to work with. Because it's fun, right? There's such a lot of potential for growth. And scale-ups are starting to do new and interesting different media buys, which again is really exciting, and that can be massively effective. The other thing that happens with scale-ups which is different to other more established brands is that they quite often have a very good performance marketing team. And that means that as soon as they do something that's demand building or awareness building, it has this huge effect, much better than an established brand that doesn't have that. Because all of that demand that's been created gets converted really, really efficiently via performance marketing.

And is there a point at which a start-up or scale-up should really try and start to build in econometrics, effectiveness and attribution? Might there be different stages where you can introduce different types of measurement, depending on how much data you have or what stage you're at?  How would you advise a start-up to start thinking about how to sequence and plan in different kinds of analysis?

It's such a good question because the right measurement depends on your life stage. And when you're really new, and you're first appearing in the world you don't really even need econometrics. You can see that you switch on this thing and sales go up. So, it's like bringing a very scientific method to something that's just very obvious and you can see with your own eyes, so you don't really need it. And, at that stage, even your basic attribution that comes through in your dashboards can be just hugely helpful. Even something like search is introducing you to new people that didn't know you were there. So quite often, it is incremental, and it is driving growth. And you can see that in the bottom line, you can see your website visits. You can see it all over the place. So it's a simpler evaluation task when you're small and you're introducing yourself to the world. Is that what you saw at Cazoo?

Yes, but it all happened very, very quickly. We had a window of 8 to 10 weeks of using just digital channels, like aggregators or paid search. Then fairly quickly we added in a load of other things. And then within two months, we had the first Covid-19 lockdown. So we had to evolve pretty quickly.

And that's actually when econometrics is so useful because you have all these things happening at once and you can't disentangle what's doing what. And that's what econometrics is for. Not just ‘are we driving sales with this stuff we’re doing?’ It’s also about which bit of it works and which bit didn’t work. That's when you get really good findings from econometrics. And I think when you are a bit bigger, and you're at that scale-up stage rather than start-up stage, that's when attribution starts to be a bit less useful. Because quite often attribution will say ‘this media did something’ but it was a customer that was coming anyway. As you're bigger, more and more people are coming anyway and the attribution error gets bigger.


What you need to really focus on is what's going to give you an incremental sale rather than just rewarding something that would have happened anyway.

That’s right, because if it's something that would have happened anyway, it’s not driving growth. It's just helping people to come that are coming anyway. That can still be an important job. You do need sign-posting to get people to the right landing page and to make sure people can find you and not get distracted by a competitor. All of that is important, but it's not driving growth.

Exactly. When we did our first piece of econometrics at Cazoo, one of the key findings is actually no bit of marketing really was the biggest driver of performance. It was the number of cars we had on the website. We kind of knew that already. We could see it happen all the time, but actually having the full econometrics really proved that out. Obviously, everything else was still important and we still needed to maintain levels of traffic but the biggest driver of growth in terms of sales and profit was just to put more cars on the website.

I've seen that across a couple of ecommerce businesses as well. Range is just massive in ecommerce. We saw it with Asos. We saw it with We saw that thing of getting more and more range because it enables you to get more hits in your SEO or keywords you can target in your performance marketing. And then once people arrive, there's more likely to be something for them and conversion rate goes up.

Very helpful. So, Grace, what would your top three tips be for a scale-up who wants to get better at effectiveness and really get to grips with it?

So, the first tip is if you're not ready for econometrics yet, (for example is you don't have that long period of data because you haven't been going that long) then what you do is you do test and learn. You do regional tests, like try it in one region and compare to a control region. Or you do switch-on/switch-off tests in control and test groups. And that is a really, really wicked way to learn and gets you into incrementality. And quite often, it's quite cheap to do as well. Because you're online, you've got all the data there. If you were a bricks and mortar business, it's harder.


The second tip is to bring stakeholders on your journey. So, in scale-up, what we find is that they're just so numerate. The ‘love language’ is numbers, right? And that’s different to other businesses when most people are intimidated by too many numbers. Our scale-up clients, whenever we're doing a debrief, you can see that they're all thinking, ‘Oh, I've seen some other numbers on this’ and they're working out in their head whether it matches what they’ve seen. There are just more numbers flying about everywhere. Whereas more bricks and mortar clients will not be doing that. So recognise that on your effectiveness journey your stakeholders are numerate and bring them with you. Let them see the nuts and bolts. Sometimes its founders and CMOs, as well; very senior people who are very numerate and they're in touch with the numbers in their business and they need to see the nuts and bolts of evaluation. Would you say that is your experience?

Definitely. You’ve also got investors who are very numerate as well, so you're having to explain the numbers back to them on a regular basis. So really being all over the detail is a key part of being at a scale-up.

So that leads on to my third tip which is to choose a partner that is used to working with scale-ups. Many scale-ups can get a lot from econometrics because even if you haven't got a huge time period, you've got much more data and much more granular data. There are things you can do with that in econometrics. And interestingly, your data quite often comes in standard formats. The back of your website is the same. The Facebook ads or Google ads are all the same. So good providers like Magic Numbers, for example, have ways to handle all of that in code so that makes it a bit quicker and easier. So you should choose a provider that is used to working with scale ups, that has all of that code in hand and then also will let your top bods look under the bonnet.


We spent a long time trying to pick the right partner. We went with Scanmar in the end, partly because they could do all of that, but also because we wanted someone who could teach us how to do it ourselves over time. To give us access to the models and let us pick them up and run with them longer term ourselves. Having something that can turn into a long-term partnership was quite useful because you won't crack it first time. You'll learn stuff every time and you'll learn to improve and refine the model as you go. Over a number of years, potentially.

You want a set up and gentle handover service. We do that with a few clients where we build the first models. They've been done professionally. We're using all our skills and expertise in collaboration with you. And then the in-house team can be updating that and learning as they go and using less and less support from us as the partner but still having someone there that will come and fix it if you’re in trouble.

Great. And what one mistake would you highlight that people should avoid?

Something you need to think about when you get to that point of being a scale-up instead of start-up is whether you are spending too much on ‘signposts’ in the digital space. Many performance marketing teams are into incrementality testing now and they're not making this mistake. But there are still some out there that are spending a lot on online ads that worked really well to drive growth when they were a start-up but aren’t driving growth any more. In the early days a very simple ad, for example on search, might say to a potential customer ‘hey, I'm here!’ which was enough when you were new, cool, maybe you didn’t have many competitors. But later on saying ‘hey, I’m here’ is not enough to convert people. You’ll see that people still click on those ads and think they’re working like they used to, but actually people are using that ad as a route to the checkout.

I see. As in paying for customers to be reminded that you're there when they would be going through that journey anyway.

Yes. At some point, for example, you can do a lot less brand search. Middle-sized businesses don't need to buy their brands terms as much as they did when they were smaller.

We spent a lot of time at Cazoo still bidding on our brand terms even when you had a lot of competitors bidding on your terms as well. So it bumped up the cost every single month and just went up and up to the point where we didn’t need to keep doing it. Almost all of that traffic will come through organically anyway.

If you’re spending a lot of money on telly advertising Cazoo it becomes a well-known brand name and it’s unique. And people have your brand name in their brain and are putting it into Google, so they're on their way to you. Some brands suppress their existing customers, they load up the first party data and suppress existing customers in brand search, and even in some generic search. That can save you quite a lot of money as well. Because your existing customers know you so they'll find you if they want to.

Great, thanks so much, Grace. I also understand you have an exciting launch of your own coming up?

That’s right. We are putting together a Magic Numbers Academy. That’s the working title as we’re still thinking about the proper brand name for it! Our first ever course is going to be all about scale-ups. It's only available as a pilot to an exclusive group at the moment, but we will be launching more of that in future. The pilot is going to run from the 18th of May.

Very exciting. Thanks so much, Grace.

50 views0 comments


bottom of page