You are using an unsupported browser. Please update your browser to the latest version on or before July 31, 2020.
Home > Insights Premium > Insights Sampling Process & Analytical Methods
Insights Sampling Process & Analytical Methods
print icon

Analytical Methods

How We Make Our Forecasts

At Headset we create our Insights daily, product-level market forecasts by combining data from government agencies, publicly available financial information/press releases, and our sample of retail partners. Before we launch Insights in any market we first run all of our data through a rigorous statistical process to ensure we are confident in our projections. This process includes ensuring that our sample of data is reflective of the entire population, that no one store or chain of stores unduly influences our projections, and that our projections would not significantly change if additional stores connected to Headset.


What Is Insights Data?

The data in Insights reflects market forecasts for the total {legal} Adult Use + Medical cannabis sales (in AZ, MI, and OR) and Adult Use sales in all other markets. We get market sales totals from each market’s regulatory agency (usually via cannabis sales tax collection disclosures). We then use data from our connected retail partners (over 1500 as of Jan 2021) to feed to our statistical processes to come up with daily, product level market share forecasts.


Sales Totals in Insights

Sales totals in Insights are the total sell-thru sales (in local currency). This is presented after any discounts and before consumer paid taxes. In addition, we present cost information (wholesale price paid by retailers to acquire products) and discounts (these are explicit discounts and exclude TPRs). Thus the average item prices are not "out-the-door" prices but rather total store collected revenue before taxes.


A Note on Taxes

Insights presents data before consumer paid taxes. This amount varies considerably state by state and even city by city. For example, there are some cities in CA where there is a 7.25% sales tax, however, sales tax may be at or above 10% in other counties. Each state has different laws. For example in CA excise are paid by retailers at the time of purchase and implicitly built into prices (rather than explicitly charged to consumers); thus, our CA sales totals include excise tax (15%) and exclude sales tax (7.25%-10.25%).


In contrast, in WA excise taxes are paid by the consumer at time of purchase; thus, our WA sales totals exclude excise tax (35%) and sales tax (10%). If you are unsure which taxes are consumer paid in each state you can find that information in the cannabis legislature or reach out to your customer success rep at Headset.


Vertically Integrated Companies

In all current Headset Insights markets vertically integrated companies are grouped under a single brand called "House Brand".


  • For companies that are 100% vertically integrated you will not find their brand listed in our Headset data.


  • For companies that are partially vertically integrated you will find their brand listed in our data, however, our forecast reflect sales outside of their own retail establishments only. For example, if we ran a company called "Headset's Flower and Retail" we sold $10M in Flower of which $6M was sold in our own stores and $4M of which was sold in other retailers you would see $4M for Headset's Flower in our Insights forecasts. The remaining $6M would be in house brand.


Frequently Asked Questions (FAQs)

How does your sampling process work?


  • Headset Insights always utilizes government reported cannabis sales totals where possible. In all states, excluding AZ, MI, and OR where we report Adult Use + Medical sales, we use government reported Adult Use sales as the topline for all our category and product level forecasts. Using these market sales totals, we then take a sample of Headset-connected retailers in the state/province to create category and product level forecasts.
  • Using a combination of human knowledge and machine learning, Headset taxonomizes a wide array of products from its connected retailers to clean and standardize Point of Sale data, ensuring that the ever-growing catalog of cannabis products in the market is captured by Insights.
  • Next, the Headset Data Science team uses rigorous statistical analyses to select (by state/province) subsets of stores from the 2,000+ Headset store sample that provide the most accurate and wholistic view of each individual marketplace. These statistical analyses are conducted frequently to ensure Insights is always utilizing the most representative stores in any given state/province.
  • The combination of taxonomized Point of Sale data and Headset's large sample of retailers allows their Data Science team to create daily product level forecasts, which then aggregate up to the government reported sales total, providing a complete picture of the marketplace.


Does your Insights sampling process vary by state/province?


  • The Headset Insights sampling process does vary by state. In states where a census of data (e.g., a complete enumeration of a population) is not achievable, Headset uses the sampling process outlined in the preceding FAQ. This varies from a state, such as PA, where Headset has a census of data, meaning Headset captures 100% of the transactional data. In this case, Headset does not need to run statistical analyses or forecasting as the raw data already reflects 100% of the market.


How do you avoid aggregation bias?


  • Headset Insights avoids aggregation bias inherently through it's sampling process. This is accomplished by using product level forecasts as the basis of our product. By first using product level data to create forecasts, Headset Insights can avoid incorrectly assigning an overall market trend to an individual product/category.
0 out of 0 found this helpful

scroll to top icon