Consider the last trend forecast you read: did you get something tangible and trustworthy — a verifiable forecast with real metrics from a structured system with a documented methodology? Or was it just marketing hype, distributed for free? In today's data-driven world, forecasting has become a staple of fascination in many industries.
However, amid the buzz and promises, there's an unsettling problem: most forecasting clearly lacks methodology and accountability.
We're here to change that. And here's how it works: the genesis of every Trend resides in a group of early adopters, known as the Trendsetters, who champion its acceptance ahead of the masses. Their enthusiasm gradually propels a wider acceptance among the Mainstreamers, which drives corresponding revenue for companies who offer related products or services.
To do accurate forecasting, our platform identifies the Trendsetters, and models how their economic activity transfers into Mainstream sales.
To validate accuracy, we offer a Data Quality SLA with auditing.
Trendsetters are defined as accounts which, in the past five (5) years, have first activated at least 20% of our pre-defined Key Trends a minimum of three (3) years before each Trend’s Peak Adoption.
Trendsetters are identified by analyzing an array of Key Trends that have peaked in recent years across all Accounts in the segment. For each Key Trend, we measure the historic Adoption for each Account posting about the Trend to determine which Accounts activated earliest. When aggregated, these accounts have consistently been “early adopters”.
Mainstreamers are the accounts which are not Trendsetters. By default, Mainstreamer adoption is “uncalibrated”, or not directly correlated with another signal, such as retail sales. A calibration process can be taken to ensure Mainstreamers are weighted such that the adoption aligns to retail sales.
Our method of forecasting is based on accountability and has both short-term and long-term modes. We model using an energy-based model of Trend propagation. Trendsetters serve as the initial catalyst for any trend. As adoption grows within this group, it mirrors the accumulation of potential energy within a system. The more Trendsetters adopt, the greater the potential for expansion. As Mainstreamers adopt a trend, it propels the trend forward, akin to the conversion of potential energy into kinetic energy, or sales.
Growth Potential measures the difference of user adoption between Trendsetters and Mainstreamers:
Trend Growth Potential = Trendsetter Adoption / Mainstreamer Adoption
A Trend with higher adoption among Trendsetters harbors more potential energy to catalyze sales among Mainstreamers. The Potential Growth score, while simple, is a great starting point for forecasting as can be seen below -- it's a key starting indicator to quickly identify potential trends.
The above chart shows a collection of ingredients and their Potential Growth as of 2019, compared to CAGR between 2019-2023. As you can see, this metric is a good starting point. From this, every forecast is done including two, more specialized, metrics called the Transfer Rate and Lead-Lag Scores.
Another key metric for forecasting is gauging how Mainstreamer activity will impact actual Sales / Revenue. This is called the Revenue Differential and is only available when relevant revenue metrics can be made available. The Revenue Differential is the ratio of Mainstreamer Adoption to Revenue; it is required for Long Term Forecasting and requires a Calibrated Mainstreamer segment
The Lead-Lag Score is another important metric is that’s assigned to every account based on how many years in advance or delay they tend to align with trends. Scores are given as integers, such as -1 year, 0 year, +1 year, +2 years, etc., which indicates how far ahead of a peak the given account is known to be.
We combine these metrics into a finely tuned regression model that’s able to precisely predict when trends will peak and the extent of their growth.
Social signals are generally out of synch with sales data.
To fix this issue, we use a calibration process that takes in regional sales data and aligns the social Share of Posts for the Mainstreamer Segment to the sales data such that the two signals are sufficiently correlated. It's a detailed process, but afterwards we measure success with correlation coefficients and a calibrated segment looks similar to below.
A 94% Pearson Correlation Coefficient is high, and shows everyone that the sales data can be used as an independent basis for forecast accuracy. This also helps us isolate Calibrated Trendsetters for a given Region + Category, which are the most valuable group for forecasting.
Uncalibrated Mainstreamers support a 1 year forecast window. Calibrated Mainstreamers can support forecasting up to 3-4 years in advance, depending on the region and category.
Our commitment to delivering accurate and reliable forecasts is built on stringent data quality standards. This Data Quality Service Level Agreement (SLA) outlines our promise to provide top-tier data integrity, ensuring that your forecasting processes are underpinned by dependable and precise information.
With our SLA, we empower forecasting with high-quality data, fostering informed decision-making and strategic planning. Trust in our commitment to data excellence and drive your business forward with confidence.