Marshall Haas is the CEO and Owner of Need/Want, a portfolio of profitable small businesses that his team starts, owns, and operators. One of those businesses is a multi-million dollar e-commerce brand: Peel. When Marshall was on the hunt for ways to improve post-purchase customer experience and increase repurchase rates, he turned to Windsor for support.

Below, he shares firsthand how he deployed personalized thank you video campaigns at Peel – as well as the quantitative results that he’s seen thus far. This case study was initially published here. Enter Marshall:

Every business owner wishes they could clone themselves. In this article, I’ll show you how I actually cloned myself to send 10,000 personalized thank you videos to each customer over the course of 2 months.

To be clear, when I say personalized video, I mean a video of me, on camera, thanking each customer by name. I sent out 10,000 of these videos to our customers after they ordered from my mobile accessory brand called Peel. We're an e-commerce brand, so each video was sent after someone purchased from the website.

Here’s a video of what I mean.

How I did it

You’re probably wondering, how the heck did he do that 10,000 times in 60 days. If you do the math, that’s an average of 166+ videos per day working 7 days a week. 

There isn’t enough time in the day to make that many videos. 

Well, It’s time to come clean.

I did this with a new startup I’ve partnered with called Windsor. I was so impressed by them I also invested in their company. 

I recorded that video once, and then using artificial intelligence, we trained a computer to say each person’s name using my face and voice. The result is the ability for me to effectively record once, and then clone myself for each video.

“By recording less than an hour’s worth of video from Marshall, we were able to create a model that accurately recreates his face and voice as he speaks. Once we had the model, Marshall recorded the perfect video he wanted to send to a customer - but, instead of saying any real names, he said something similar to “Hey {first name}”.Using an integration with his Shopify store, we were able to automatically create a video minutes after a new purchase - just from the name someone used on the order. A few hours later, the personalized video is delivered to their inbox without Marshall so much as lifting a finger.If this reminds you of how businesses send personalized emails at scale, you’ve got the right idea. With the amount of emails people receive, it’s hard to stand out as a business. A genuine email gets lost amongst endless promotions and marketing. We’ve seen businesses before personally record and send out videos to every customer. People raved about them. However, as a business grows this becomes impossible to scale and that’s where we step in. We help businesses maintain this personal connection at scale.”

Ethics

I want to address the ethics of this head-on.

My excitement for this technology and Windsor as a startup can be summed up in 2 points:

• Personalization is a good thing – Windsor allows you to personalize and scale experiences for your customers. 
• This is the business use case of the Deep Fake technology. 

The technology of Deep Fake videos has been around for a little while, but most people have only seen it used in nefarious ways like faking a President saying absurd things that they didn’t really say, or creating obscene images of people as blackmail or profit. I think Windsor can change that and create the first business use-case of Deep Fake technology to be used for good. 

“We’re at a point in history where deep-fake technology has most people scared. This is with good reason - the possibilities with deep-fakes are quite scary. That said, we’ve been here before. Whenever the world was introduced to new technology, people with malicious intent were quick at figuring out how they can use it to serve themselves. By bringing this tech to businesses like Peel, who have a track record of doing what’s good for their customers, we’re opening up new possibilities. Along the way, we hope to educate people not only that deep-fakes exist, but that we can actually use them more productively.”

But, is this genuine? 

I want my customers to know who’s behind our brand, see my face, know I’m proud of what we are selling, have a great experience, and ultimately invite them to reach out to our team for help if they have questions. 

That is all truly genuine and exactly what I wanted expressed in the videos we sent out. The only difference is I was able to scale that experience for customers by using this technology. I think that’s a win-win. 

If I was a retail store owner greeting customers at the entrance and someone came to me and offered to place a clone to do the same job, so I could be productive in the back, would I take that offer? Yes.

Windsor’s team has an interesting stance on this technology. The technology exists already - they didn’t create it, they adapted what’s already possible to solve a real business problem: personalization. Since it exists, there’s no putting the genie back in the bottle. 

The best thing we can do now is to use this for good and not evil. They want to normalize the fact that anyone can create Deep Fakes, and raise awareness on this technology. You should be aware of what’s possible and not believe everything you read and see on the internet. They also plan to develop technology which helps detect deep fakes.

Stats

June 1 to July 31

• 68% open rate on videos sent
• 86% views watched to completion
• > 1 view per recipient on average

While preliminary, the most interesting stat from this experiment is:

During our 60 day experiment, inside that window of time we had a split of customers that received a video, and those that didn't.

• Of those that didn't receive a personalized video from me, only 3.75% purchased from us again in those 60 days.
• Of those that received a personalized video from me, 7.02% purchased from us again in those 60 days.

That means customers were 87% more likely to buy from us again when they received the personalized video from me.

I will be very interested to see how this stat changes over time. This blew my mind 🤯