This week we interviewed Liz Connors, our Director of Analytics, to see why Headset's data is the most advanced and reputable in the cannabis industry. Continue reading for a deep dive into what makes Headset data and analytics so valuable, and how our expertise can help you gain a comprehensive understanding of the cannabis industry so you can make better-informed business decisions.
The cannabis industry is still quite young and, like anything in the beginning stages of development, is going through various changes as it matures. Using data to track and monitor how the market is rapidly changing is critical right now. Trends are born and disappear quickly so cannabis operators are at risk of spending valuable time headed in the wrong direction if they don’t have access to this data. The ability to track these changes helps cannabis companies better understand the entire landscape to see where they fit today and where they can see themselves in the future.
Industries that operate without information literally do not exist. Legacy retail operations have a wealth of fantastic data. Companies like Target and Walmart employ entire departments of analysts dedicated to diving into the data and finding meaning in the numbers. Meanwhile, many cannabis companies only have one employee tasked with analyzing data, and they’re usually split between other duties.
At Headset, we’re trying to turn analytics into a service. Rather than hiring analysts who gather and clean data, we build dashboards so users can self-serve, which are especially useful to those without enough resources to dig into the data every single day.
As a cannabis data intelligence company, the core of our data comes from cannabis retailers. We are currently connected to over 1,200 cannabis retailers across the U.S. and Canada. This means for every one dollar spent on legal cannabis in the U.S. and Canada, 25 cents of that is tracked through our databases. Our data is then transformed for all different uses.
For retailers, we offer a suite of analytics products to help them better use their own data to make more informed business decisions. We make the data actionable and easy to use for someone busy running a business.
We also have a team of analysts dedicated to our Market Intelligence service, Headset Insights, who create forecasts for adult-use consumer trends in the U.S. and Canada. This information is available to all types of cannabis companies, from operators and retailers to service companies. They can discover what brands are winning or losing in the marketplace, which new product trends are coming in, pricing analytics, demographics and more. Instead of answering simply what is bought, we can answer who is buying it, providing deeper insights into these purchases and the actual consumer behind them. It helps transition from this popular perspective of B2B and a brand’s relationship with a dispensary to their relationship with the consumer.
Our vendor-managed inventory tool, Bridge, helps brands connect directly with dispensaries to see how well their products are selling and whether or not they need to restock. It’s difficult for brands to manage distribution to all of their retail stores and partners, so we provide them with a platform with all of their sales and inventory data in one place.
Headset sources data directly from cannabis retailers’ point-of-sale (POS) systems in real-time. Today, we are connected with over 20 of the largest POS providers in the industry. Due to our ability to connect with almost any type of retailer, we are able to source a statistically representative sample of data.
When making market projections it’s imperative that the sample data you use is representative of the whole population. Imagine, for example, that you are trying to project the height of the average American using sample measurements from NBA players. This would lead you to postulate that the average American is very tall; the same can be true with sales data. If you use only the data from large high-end stores, only data from MSOs, or only data from small, vertically integrated mom-and-pop stores you will not have very accurate forecasts. At Headset, our integrations allow us to connect to all types of retailers and so we can balance out our sample to ensure our forecasts reflect the market as a whole.
This approach is rigorous, time-consuming and requires a team of people to target the proper retailers, plus another team to develop forecasts. Despite the necessary resources involved, we believe this approach is the only one that can be trusted to provide accurate measurements of the market. Simply publishing the data from a small sample of similar retailers will not give accurate enough results that our clients need in order to make their decisions.
Transaction level data is critical to our ability to offer insights. Specifically, with transaction level data we can provide our clients with more granular data and include cutting edge analyses like market basket analysis, attachment rate analysis and even customer measures like repeat purchase rate and customer lifetime value. None of this would be possible without transaction level data.
We do not rely on surveys to create our projections of the market. We believe that, while beneficial in some cases, surveys are more error prone and result in less actionable and accurate insights. For example, if I asked you to recall how many times in the last 6 months you bought cannabis, what products you bought, and how much you paid, could you recall precisely? Would you forget or have to give wide ranging estimates? With transaction level data we do not rely on the memory of consumers but rather recorded data about the transaction at the time it happened.
Our Insights product uses the most comprehensive methodologies since our other products collect automatically. We analyze data at the market level using a series of statistical models and machine algorithms to project total sales for the market.
Our biggest differentiator is that we’re POS-agnostic, meaning we can gather all types of retailers connected to our platform. That means I am working with a very balanced sample rather than data from a handful of stores, only large stores or only those that are well-funded. Our data collection approach gives us a representative picture of the market.
Transaction-level data lets us get deeper into these measures. We are able to see cross-product purchase rates, consumer’s whole basket purchase and their lifetime purchase cycles instead of just total purchases throughout the year. Aggregated total sales only answers what is happening, whereas we are trying to figure out why it’s happening, how it’s happening and what we can do next with that information. Transaction-level data provides granularity for deeper questions to drive actual decision-making rather than providing just facts and numbers. Our data shows where to go and how to better perform.
The way we build our projections and conduct our sampling is the best way to get a truly representative sample of what is happening in the marketplace. Aside from that, we have a very large team of analysts focused on making these data easy to use and actionable. It’s relatively easy to find raw data and numbers, but actually knowing what to do with that data and where to take it is very tough.
We are experts at cannabis retail, building dashboards and doing analysis for 1,200 retailers every single day. We know what you need before you need it. Brands only make money when consumers hand over cash for that product and we are experts at knowing what, when, and how they are purchased. Since we do so thoroughly understand retail, our data is the most powerful for brands compared to surveys or financial forecasts based on financial results. That type of data doesn’t equate to being an expert in retail.
First, the size of our data. For every consumer dollar spent on cannabis, 25 cents comes into our database. This provides us with a massive set of real-time, true data. We are not asking people in a survey to recall what they bought months ago, or what people think they did. We have compiled years of information on what actually happened.
Plus, since we are POS-agnostic, we have a very representative sample. We specifically pinpoint the proper retailers we need to get a truly accurate market forecast. Since we’re connected to so many retailers, my team of analysts can statistically balance our sample in a way that competitors are unable to do.
We’ve also been at this for almost six years, so we’ve been in this space for quite a long time. A lot of thought and planning went into building Headset and it’s data collection methods to ensure it is the most accurate information available.
First, we are so rigorous with creating a proper sample so we can better rely on our projections. We work to overcome statistical hurdles that our forecasting needs to pass before we’re allowed to share with the market.
Second is our product normalization. There aren’t any UPC’s in the U.S. cannabis market, so it’s difficult to know if products sold at store A are the same as those sold at store B. We use machine learning algorithms and our analysts normalize products so users can analyze purchases across stores and better understand the marketplace as a whole rather than across just a few stores.
The most significant reason is in the way we build our data products. They are meant to provide the end user with information in a way that’s super actionable. Other data providers will hand over massive data files that the customer has to sift through to discover what’s important. We have that step built directly into our tools. For example, during COVID-19, others put out news articles while we were building dashboards for users to go in and see what is happening in real-time. The dashboard updated itself every single night and didn’t require users to do any analysis themselves.
To me, the most interesting data we gather is for loyalty programs. We receive anonymized consumer records which allows me to follow customer A, B, C and so on throughout time. Rather than knowing what was purchased on any given day, I’m able to actually follow their journey. I know what their introductory products are, which products are considered good substitutes and how much money a brand should spend on a customer given how much revenue they will receive from that customer over time. We know how often customer A visits a dispensary, what they buy this time versus last time and seasonal trends. Using our understanding of individual consumers, we can aggregate that data to unravel why people are buying cannabis and what they are buying.
More recently, we were doing an analysis for how consumer trends shifted in the last year. We’re about a year out from Vapegate and the onslaught negative media, which we saw caused a substantial shift in consumer purchasing patterns. Since then, flower has had some resurgence yet due to COVID-19, edibles have also gained popularity. We can see consumer trends shift very quickly, within a few weeks. To me, that was interesting to see how in grocery stores, for example, market shares of certain products increased by 15-20% during COVID, while that’s just a typical day in cannabis.
Working at Headset as an analyst is different from most other data companies. As an analyst at a CPG company, you’re really just answering one question for one vertical of the business. At Headset, everything you build and learn needs to be reproducible for thousands of users. The goals of a retailer franchise in Canada are totally different from a vertically-integrated company in Florida or a consumption lounge in California. One of the most challenging and most rewarding tasks is trying to find ways we can make data products that resonate with all of these types of businesses. How do we get down to what’s important for a business to grow and succeed? Then, how do we replicate that for thousands of different people? It’s truly exciting and innovative work.
Before joining Headset I worked in CPG and Financial Services as a Data Scientist. Retail analytics has always been a passion of mine and I believe that with data we can make informed decisions and avoid costly mistakes. I left the private sector to do research on managerial decision making as it relates to data literacy and truly fell in love with helping managers use their own data to make better decisions. Shortly thereafter I met Cy, Scott, and Brian who were working to bring data analytics into the cannabis space. After working with them it became clear that cannabis operators lacked the tools and data they needed to grow with scale and build the critical efficiencies to operate in the razor-thin margins of retail and CPG. Data science and data analytics were so common in CPG and Financial Services but there was very little of it in the cannabis space and it seemed like an amazing opportunity to help an industry that was growing and changing every day. Getting data into the hands of decision makers has always been my passion and with Headset I now help to create tools that over 1,000 cannabis retailers use daily to support their decision making.
Being a start-up in an ever changing environment means there really isn’t a “typical day.” I manage a team of analysts and data entry specialists that do everything from data engineering (imagine how messy data from 20 different POS sources is!) to writing industry reports and webinars and creating our daily product level forecasts (with over 300k products and 8 markets this is a lot of daily forecasting!). As part of another service offering, my team also helps business decision-makers better use and understand data. This is my favorite part of the job - sitting down with someone that has questions and knowing that I can find the answers they need to succeed!