Lessons from top companies on building a better data-first strategy, WebValue, BlogPost

Data empowers marketers to make better decisions and take smarter risks, but sometimes the best intentions lead to the wrong solutions. Interpreting data isn’t always easy, and I’ve seen marketers come up short by not allowing themselves the space to learn, grow, fail, and improve from their collective experiences.

A campaign that falls short of its goal can teach just as much as one that succeeds. And marketers that wish to do the right thing, well, can learn from how they do the right thing, poorly.

Through leading Customer Analytics at Google, I’ve spent years not only studying the technical aspects of measurement and the relationships that users form through digital media, but also understanding how to strengthen the analytical cultures of world-leading organizations themselves. Time and again, I see companies make crushingly common mistakes with data, and refusing to give themselves the room to experiment and to fail.

What I’ve noticed is that marketers have become experts at doing the wrong thing, because they’re grounded by the past and the “way we’ve always done it”. Their organizations expect them to succeed, even if that success is dependent on the wrong technique or marketing channel, or in pursuing customers who are detrimental to the company's long-term growth.

One of the biggest mistakes that a marketer can make is to look at their data in isolation.

But I’ve also seen what companies can do when they allow themselves to take steps in the right direction, even if they fall flat at first.

Here are a few examples of how successful organizations think, and how you can apply them to your business.

1. They look at their metrics as part of a story, not the whole picture

One of the biggest mistakes that a marketer can make is to look at their data in isolation. If you oversimplify your data, you’ll lose out on the magic that’s happening around you.

One thing that I’ve observed from successful companies is that they don’t capture metrics for the sake of it. For every metric they set and optimize toward, they go a level deeper by asking themselves some key questions:

  • Do I know what this metric truly means? Let’s look at conversions, for example. They are central to everyone’s business, but not all conversions are created equal. When considering for cross-device users, online-to-offline users, and view-through conversions, we start to see real differences emerge between platforms. And they could determine how you interpret their performance and future spend. It’s important to understand the specifics and the distinction when calculating for each metric.
  • What could influence that metric, and how? The late Andy Grove, former CEO of Intel, said “For every metric, there should be another ‘paired’ metric that addresses the adverse consequences of the first.” I suggest taking his advice to heart. If you are optimizing toward using deals or coupons that are driving a ton of customers to your store, are you doing so to the detriment of profit margins and customer retention?
  • Am I limiting what I can learn from my metrics? Don’t focus your attention solely on what’s underperforming. Think about what you can learn from what you’re doing well, but could do even better. If you’re a shoe company and you’ve successfully incentivized customers to buy multiple pairs of your shoes online, that’s a great win! But why did that happen? Do you even know? Can you replicate the findings to other marketing activities? 

Lessons from top companies on building a better data-first strategyI’ve seen that top companies don’t just look at their metrics as numbers. They look at their metrics as opportunities to ask more questions: where is the market headed? What should we be aware of? That way, a single metric becomes part of a larger story, not the whole picture.

2. They expect human behaviour

Machine learning is growing fast and teaching us a lot. But people are not machines, and as such, they’re not always rational, efficient bidding and buying engines. They don’t necessarily respond in the way you’d think they might. As a marketer you have to plan for that by gaining a better understanding of the human story behind your data—because it’s those behaviors that may drive your business forward.

Here are some examples of what I mean from recent research on behavioural economics:

  • Slow doesn’t always mean no: It’s hard to deny that when it comes to site speed, faster is far better—the longer people wait for your site to load, the more you’ll lose. But a study by Harvard Business School found that, due to what it called the ‘operational transparency’, people can tolerate—and in some cases, prefer—websites with longer waits as long as there is an understanding of the work being performed. If you demonstrate that your site is exerting effort on a customer’s behalf (consider theDomino’s Pizza tracker, which keeps you updated every step of the way in your journey to getting your pizza), it can contribute to have a stronger sense of loyalty and reciprocity towards your company.
  • The perils of proactivity: Facing the challenge of increased customer attrition, many service companies will start to recommend lower-cost pricing plans to their customers in an effort to demonstrate care for their customers, and the greater benefits they can provide. Researchers from Columbia, Wharton and IAE Business Schools found this tactic had the opposite effect: encouraging customers to switch to cost-minimizing plans can actually increase churn. In some cases, it inspires customers to be less inert about making a change, and they start to look at other service providers. 

Lessons from top companies on building a better data-first strategyI’m not suggesting you throw your conventional wisdom out the window with these varied examples of counterintuitive consumer behaviour. But I think it’s important to know that successful companies know that you can’t predict every single element of the customer journey—no matter how much you measure, you’re not going to capture everything. If there are no perfect humans, there’s no perfect data.

3. They’ve fallen in love with failing

When we work with smaller businesses or startups, we tend to see some incredibly miserable attempts at marketing. That’s all part of growing, right? But we can learn so much from how these companies tend to respond to those failures: they look inward. They consider that perhaps their brand isn’t strong enough yet, or that they haven’t properly optimized their campaigns in these early stages. What they don’t do is look for something or somewhere else to place blame.

There are multiple components to performance, and failure is one of them.

Here’s what I see over and over in larger organizations: They’ll test something, and if it fails, they’ll pivot immediately to a strategy with which they can win, arguing the customers simply aren’t there or that the channel doesn’t work for their business.

This is where doing the right thing, poorly, needs to become your new manifesto, no matter what size your organization is. There are multiple components to performance, and failure is one of them. Give yourself and your teams the ability to fail, with the understanding that it’s the first step to growth.

With these best practices in mind, ask yourself: What is the right thing for me to be doing? Even if you can’t immediately take action on the answer, acknowledging it to yourself is the first step. I hope you’ll consider these best-in-class approaches to data and measurement as a step toward doing the right thing, well.

Key takeaways for doing more with your data
  • Look at your metrics as a whole picture: Question every metric you set, and don’t oversimplify; ensure you’re learning from the wins as well as the losses.


  • Anticipate human behaviour: Humans aren’t perfect, rational beings. Your data won’t be perfect, either. Accept that you can’t measure everything.


Fall in love with failing: Embracing failure is the only way to grow, and find the right thing for your company to be doing.

Source: Think with Google


Share This Post
This block is broken or missing. You may be missing content or you might need to enable the original module.

Subscribe to Our Newsletter and Stay in Touch