How to use data to compete with Amazon post-Covid
My predictions for Black Friday weren’t good. I warned that the usual draws for Black Friday had disappeared. Deep discounting had already been the primary strategy that companies had been using to try to recover from Covid-19, which meant the massive sales had become normal. Plus, 70% of American adults report struggling to pay bills, leaving less disposable income for Black Friday and Cyber Monday. It looked like price-sensitive customers had temporarily disappeared, making cutting prices less effective than in the past.
I wrote an article forecasting that Black Friday and Cyber Monday were poised to disappoint. After all, over 60% of consumers had made no plans to shop on Thanksgiving weekend in the days before the holiday. Even if a few more people ended up browsing the sales, Black Friday disappointments seemed assured.
It wasn’t a popular opinion. No publication wanted to highlight my article. I don’t entirely blame them. Who wants to hear more doom and gloom about the retail sector in 2020? As a data scientist, however, I believe that we have to face the numbers head and deal with their reality— and statistically, we were heading for a reverse to the decade-long sales trend upwards for Black Friday weekend.
From Black Friday to Bleak Friday
And yet, somehow, the reality was even worse. Despite over $9 billion in online sales over Thanksgiving — a 21.6% increase over last year — overall Black Friday sales dropped by 20%. This kind of drop hasn’t happened since the 2008 financial crisis. Record online sales may have made the flashiest headlines, but those sales are overshadowed by the overall decline in revenues over the so-called biggest shopping weekend of the year. Unless you’re Amazon, reporting record-high sales, the holiday shopping season isn’t looking good.
Analysts have christened the days after Thanksgiving 2020 Bleak Friday. It’s an inauspicious data point for retailers.
Bleak Friday: a phrase I have not been able to get out of my head. 2020 felt like a bleak year, so it seems apropos. This was not an outcome I wanted for any retailer, but it is the one I saw coming. We sounded the alarm, but no one wanted to listen.
Yet despite this disappointing performance, it is possible that this is just an outlier. Perhaps, the trend upwards will return once the pandemic is over?
I don’t think so. Why? This is just a particularly stark example of the pull-retail trend I have been talking about for years.
So is there any way to recover? Can retailers compete against behemoths like Amazon? It’s possible when you use data effectively.
The pull model of retail matters even more post-Covid
For several years, we’ve been trending away from the traditional push model of retail where retailers themselves could dictate what consumers bought. Whatever was available in our local store was all that we could choose. The rise of online shopping transformed the retail industry. Consumers were no longer limited to the few stores in their area. Items can be shipped from around the world, allowing consumers to pull the markets towards their preferences and needs.
In other words, people today do not choose from what they can find, but rather search for what they want.
Covid-19 has magnified this trend. After all, we are all buying online now, so why not search for precisely what we want? We can dictate the color, size, style, and even the tiniest details of our purchases and get them shipped directly to our homes. The dream product is just a click away.
While we may venture out of our homes again once a vaccine becomes widely available, our shopping habits have likely changed forever. A study by the UN Conference on Trade and Development indicates that people are unlikely to stop primarily shopping online even once it is safe again. Studies by Deloitte and McKinsey have had similar findings. Online shopping is here to stay, killing the push retail model forever.
What does the triumph of the pull model of retail mean for companies? Success requires offering your customers the exact product they want at the exact right time.
Why Amazon is perfect for the pull-retail world
It’s no wonder that Amazon has grown so much during the pandemic. Practically anything you could ever want is available every day on one website. People have even bought prefab homes on Amazon! Amazon is the ultimate vehicle for consumer-driven retail. Ask, and you shall receive.
With the Amazon model, success breeds more success. As they sell more, they collect more data on consumers, allowing them to tailor their offerings to customer demand further. The more we buy, the more they know. As time goes on, the competitive advantage grows exponentially.
How to beat a behemoth: Data
Most retailers can’t be Amazon: they cannot offer such a massive selection or such low prices that they trounce the competition.
So how do you compete with the behemoths? Data.
Instead of endlessly discounting items that may not even interest consumers, you need to anticipate what your customers want and make it available at the right moment. Data makes that possible. When you use data to make decisions about what inventory to stock where you get much higher returns. Revenues rise, even when overall consumer activity is sluggish.
This Black Friday might not have been so bleak if retailers had made the right offers instead of assuming that any discount big enough would move merchandise. In a pull-retail world, sales come down to the right offer, not the cheapest one.
Data, not discounts, convert.
A better model for the holidays and beyond
But Amazon has tons more data, right? The average retailer collects only a fraction of the internal and transaction data that Amazon collects. How can any retailer hope to keep up?
That’s where data science expertise comes in. With the right model, you can still get an incredibly accurate demand forecast, down to the exact SKU needed in each store. What does that take? 2 things:
- More, better data
- Well-designed algorithms
The best offense against Amazon’s massive cache of data is access to Big Data of your own. At my company Evo, we tap into a database that contains data on the transactions of over 1.2 billion people, as well as data on everything from weather to economic trends. The more information, the better.
Of course, you have to select your data carefully to ensure that you aren’t adding noise to your model. Data quality is just as important as quantity. Carefully choose traditional and alternative data that will guide your understanding of what the customer wants.
Next, you need to process your data well. That requires a well-designed algorithm. You have to train your model to leverage the data efficiently to make increasingly accurate forecasts of what customers want and how to get products to them at the right time.
The result? 94% accuracy in predicting what your customer wants, where and when they want it with much less inventory. It may not be possible to compete directly with Amazon’s business model, but a data-driven approach means that any retailer can sell more with less inventory. Change your course now to reap the benefits of what’s left of the holiday season — and beyond.
A data-driven path past Covid
The struggles of this holiday season aren’t going to disappear overnight. It will take time to transition into a true post-Covid world. Even once we have returned to some semblance of normal, things will never be quite the same. For retailers struggling to stay relevant and thrive in the Amazon Age, effective use of data is critical.
It is not enough to use your own data in the same ways as in the past. To compete with the biggest players like Amazon, you have to augment your data and deploy it effectively. A data-driven retailer today needs to commit to a pull retail model and then create the best possible algorithm to support that strategy.
I tried to sound the alarm about Bleak Friday too late, but it’s not too late to make 2021 a brighter year for retail. Anyone can find a path out of these difficult times, so long as they follow the data. You will find that upending traditional “best practices” with data-driven decision-making leads to even more optimal practices — and higher holiday returns.
About the author
Fabrizio Fantini is the brain behind Evo. His 2009 PhD in Applied Mathematics, proving how simple algorithms can outperform even the most expensive commercial airline pricing software, is the basis for the core scientific research behind our solutions. He holds an MBA from Harvard Business School and has previously worked for 10 years at McKinsey & Company.
He is thrilled to help clients create value and loves creating powerful but simple to use solutions. His ideal software has no user manual but enables users to stand on the shoulders of giants.