This June, BuyerGenomics was invited to speak at the Elite SEM Summit held at Google’s New York City headquarters. We were accompanied by one of our clients Dylan’s Candy Bar to present their current campaign to acquire new better customers.
SEM Presentor: [00:00:00] We’re going to keep on moving with the elite partner showcase your top up of so 2017 your next speaker may come as a little bit of a surprise. But we are going to bring on BuyerGenomics and Dylan’s Candy Bar a big surprise for you guys. They’ll be next up so please give a big round of applause for BuyerGenomics and Dylan’s Candy bar.
Frank: [00:00:29] First off we just wanted to thank the elite team for putting on such a great summit. So thank you guys very much. I would like take one second to congratulate Ben, Zach, and the entire company on just building an incredible company and culture. I’ve known Ben the CEO for about 17 years and was with him when he started a lead out of his apartment. So to see what it’s grown to is really really special so.
So just congratulations really.
So what are we doing here today. We’re going to walk through a brief case study of how BuyerGenomics which is our predictive marketing database identified and cloned Dylan’s Candy Bar best customers. We did this in a new customer acquisition campaign that generated customers that spend more and spend more often. So instead of me just telling you what we did for Dylan’s I thought it would be cool to bring up Tushar. Who is the president of Dylan’s Candy Bar to talk a little bit about the brand and what we did for him.
Tushar: [00:01:42] Thanks Frank. So for those of you who don’t know us we’re a small specialty retail company expanding aggressively across America now and also playing in e-commerce wholesale and licensing with about 20 stores in the U.S. and soon to be in some stores outside the country. Prior to my days of Dylan’s I was at McKinsey for a while and working with the BuyerGenomics team over the last couple of months reminds me of sort of the caliber of people I worked with for a large portion of my life a very analytical problem solving driven and very results oriented. So that’s sort of been a great experience. But given the context of the issue prior to using BuyerGenomics you know we had customers coming through two channels. One is of course retail walk ins. We have retail stores across America. And second is search. Having said that while we are destination driven and tourist driven so we get a lot of people coming into our stores from all over the country and the world. We’re also trying to focus on more new customer acquisition to other channels and search. Having been one of them in the past had sort of maxed out and in fact we’ve actually lost some momentum in branded search.
[00:03:05] Actually branded search is down.
But they do not work with Elite yet.
[00:03:09] So just the transparency we are not an Elite client. So we haven’t had that advantage yet. So we were looking for new avenues of you know finding new avenues for customer acquisition. Having only in addition to that you know we had customers coming in where we were basically the new customer growth was declining and also a lot of our customers given that they are tourists ended up becoming one time buyers which is also an issue of course because frequency of purchase is also not wasn’t really positively trending as well. So that’s when we sort of turn to BuyerGenomics.
[00:03:51] So what is BuyerGenomics. BuyerGenomics is a predictive marketing database. They can mathematically identify your most valuable buyer to find the most predictive attributes of those buyers and find new customers that will spend like your most valuable buyers do today. BuyerGenomics can also do a lot more than that but for right now we’re going to talk about cloning. So how did we clone Dylan’s best customers.
So we start with their best customer and really wanting to know everything about them. So there’s this rarefied customer that spends the most they spend the most often it’s usually the top 10 to 15 percent of your transactional database. So we really wanted to study and analyze those customers. So what we’ve done was we enhanced the data with 850 demographic, psychographic, and lifestyle data points. And then we don’t just build a model and try to outperform that with another model slowly over time. Instead we actually automate the construction of every permutation of those 850 demographics, psychographic, and lifestyle variables which creates roughly about 1.9 million models. The system then runs what’s called the Monte Carlo simulation to identify the most predictive model of the ones that we just built and then having access to over 180 million households in the US. BuyerGenomics can then find new customers out in the wild that have never purchased from Dylan’s before that matched the most predictive model. So the winning most valuable buyer targeting model is tested and generates new customers. Once a critical number of net new customers is achieved which is usually around a few hundred We then perform a similar process on those customers.
[00:05:50] We just brought in so we remodel them off of those customers from the wild. And this is called our lift off model. And usually when we do this the lift off model has a ROIS over four to five hundred percent I like to shop talk about the results.
[00:06:06] So we’re still in the back end of back end of the pilot but so far we’ve already acquired over about 300 most valuable buyers obviously increase in prospect prospecting over 225 percent and 30 percent of the investment costs so very different from what we were doing before. In addition what’s really important is as Frank mentioned that it’s not just about acquiring new customers we’re acquiring a much higher quality customer. So already in a short period of time we’ve seen that that average spend is over 50 percent more than our average spend of general customers online and in stores. And then in addition in that we have a higher repeat rate but just rate them up by 60 percent. So that’s just been within the last. I think we started this process now six to seven months ago. And these are the results so far. The next step from here is to you know buy by fall this year. We’re planning on launching the lift up model. And based on the projections which will hopefully increase you know continue to give us the results but we’re expecting a couple of million dollars in incremental customer value from from this from this product program and by the end of the year.
[00:07:26] So I just want to quickly touch on a few other of the powerful capabilities of BuyerGenomics. So this is a marketing owned database that’s really geared for marketers. So in this dashboard there’s like a million screens on the way to look in the dashboard. You guys are looking at you can see very easily your number of customers or a number of new customers your full price purchasers your you know a percentage or one time buyers in your database but also right under dashboards that select an event.
[00:07:58] So you can actually sort by moments in time whether it’s Black Friday Mother’s Day whatever it might be not just a date range
this is an example of our life stage groups. So what we do is we overlay the transactional database of the of. In this case Dylan’s over the national model of 21 life stage groups. You know every brand has their story. Our customer is early thirties wealthy lives in the city. So what we do here is we actually allow you to statistically validate those assumptions and then also help your brand identify new segments and personas that you might not have thought are very valuable but actually spend like your most valuable buyers do in our database today. Here’s an example of one of those life stage groups so fortunes and families. Here you can see how many of them how many customers there are in the database what they’re worth.
[00:09:00] But we also show you some market sizing so you can see here there’s seven point six million fortune and families in the U.S. That’s about 6 percent of them. So as you find these valuable segments you can know exactly what level of effort to put into them based on market sizing and not to mention you get really important information about this group so you can do more personalized messaging. Aside from the cloning of customers this is probably my my favorite part of BuyerGenomics where we take the transactional database and break it up into six different segments prospects we just consider customers with zero transactions for now. And a lot of people when you think about a transactional database think of it as something that’s static but it’s not it’s actually dynamic and all of these numbers that you are looking at will change every single day and they’re adaptive to the brand and adaptive to the individual. So for example Tushar and I are both in market I might become a Fader before Tushar does it based on my past purchase history and my cohort analysis. So it’s very very specific. And you know in a roomful of marketers I’m sure your like oh great every day like there’s no way we’re going to look at this every day and how do we how do we act on this. So we’ve actually solved that with something that is called auto pilot. So what autopilot does is it looks for changes in buyer life cycles.
[00:10:27] So if I go from in markets of fading the system will automatically send me an email and they’ll send me an email based on a few pieces of dynamic variable content. So for example from full price first discount purchaser my most valuable buyer What’s my life stage group what buyer lifecycle I’m in. So it’ll automatically send me a relevant and targeted email at the right time which is something that’s really important and then we’ll send the suppression list to your email provider to make sure that person doesn’t also get the bach blast spam emails that everyone sends.
[00:11:01] But this is one of the things that our clients really love and see value in so I want to thank Tushar for his time and we all should thank him because in your little goody bags there’s Dylan’s Candy Bar chocolates. That’s for them. And any questions please let us know.
[00:11:17] Thank you.
