Why Crowdsourced Data Matters In Retail Forecasting

CrowdSourcedData

With the recent unveiling of the new Google News app, we, like many other users, had to re-implement many of our prior preferences, favorites and saved searches.  Re-establishing these preferences at first seems like a mildly annoying chore but it forced us to think about the preferences and screens that were really important to our business.  Along with the obvious searches, such as retail, sales forecasting, Sales Temperature, weather impact calculator, restaurants, ML/AI, predictive scheduling and minimum wage laws, we also added a search for Crowdsourced Data.

You’re probably familiar with crowdsourcing.  As defined by a quick Google search, crowdsourcing is “the practice of obtaining information or input into a task or project by enlisting the services of a large number of people, either paid or unpaid, typically via the Internet.”  Crowdsourcing became well-known with the emergence of popular services such as Wikipedia and Yelp but according to Wikipedia, the first known example of crowdsourcing was the Longitude Prize in 1714, when the British government offered a monetary award for the best solution to measure a ship’s longitudinal position.

But what about crowdsourced data?  This may be a less familiar term, but the idea is reasonably simple.  Crowdsourced data refers to the creation of data sets utilizing a large number of data providers or sources.  Similar to crowdsourcing, when executed correctly, crowdsourced data produces a type of network effect; as the number of contributors and the data set itself expands, that specific data set becomes more valuable for all of the users.

For an illustration of the power of crowdsourced data, think about the Waze traffic and navigation app, now owned by Google.  By using the app while driving, individual users contribute to the traffic information collected by Waze, allowing other drivers to benefit from the aggregated, real-time traffic conditions.  The application wouldn’t be nearly as powerful if the data set was limited by some meaningless constraint like drivers of a certain brand of car or users of a specific model phone.

Crowdsourced data is becoming increasingly important with the emergence of artificial intelligence.  Most of the applications that utilize AI rely on and greatly benefit from large amounts of data that allow the AI platforms to continuously “learn” and improve the accuracy of their answers.  Crowdsourcing data is a very logical method to develop and maintain the types of large, growing data sets utilized by AI.

What is crowdsourced data utilized for today?  Our search for crowdsourced data on Google News returns a range of topics that span from the mundane to the fascinating.  Crowdsourced data is being used to help counties locate and fix potholes, allow cities to address dog mess problems and enable scientists to study the human microbiome.

What do all of these examples have in common?  First, the data contributors benefit from the resulting data set that is crowdsourced by all of the contributors.  Second, the universe of contributors is as inclusive as possible to enable the creation of a large useful data set.  Similar to the prior Waze example, would the dog mess data be as useful if it only included poops from pekingese dogs or if the American Gut Project to study the human microbiome only included samples from vegetarians?

At Sales Temperature, we are committed to the belief that retail sales forecasting benefits tremendously from utilizing crowdsourced data.  That’s one of the reasons that we recently released our new data co-ops that allow our users to join groups that share data based on brand, category or location (while of course keeping each individual user’s data confidential).  By sharing data, we can produce a more accurate forecast for all of our users and provide greater detail about the historic and future impact of weather on their business.  All without erecting artificial constraints that hinder other forecasting tools, such as POS systems or scheduling apps.  After all, shouldn’t the information you use to run your business have the same advantages and benefits as the apps you use to avoid traffic jams?

 

 

Advertisements

“You Can Observe a Lot Just by Watching” – Yogi Berra

Well, in our case more like listening.  While our blog updates have been just a teeny bit well…lacking, we have been hard at work improving our services.  Many thanks to all of our users and potential users who have offered up their input and guidance as we continue to strive to make Sales Temperature as useful and valuable as it can be.  We have been listening, and we are getting ready to roll out a number of new features.  We’ll follow up this post with a series of social media and blog updates to announce specific enhancements, but here are the top three highlights:

1) Launch of a weather impact calculator to determine the historic and forecast impact of weather on a retail location (sample below).

cropped-sales-temperature-weather-impact-calculator.png

2) Creation of data co-ops to allow users to categorize their locations.  By utilizing co-ops, we can improve the accuracy of the forecasts for all of the co-op users.  We can also allow users to compare their historic and expected future performance versus other members of the co-op.  And don’t worry, we will always maintain strict adherence to data confidentiality among users.

3) Exposure of the Sales Temperature services through an API.  This will allow our users to integrate our services into their existing dashboards and applications.  Want to use our real-time forecast as an input to your scheduling app or quickly see the quantified impact of weather on your sales last week in your BI tool?  The API allows for these kinds of data exchanges and once they’re set up, updates occur automatically.

Check back soon for additional information on our exciting new features, and thank you again for all of the user input and guidance!  Please keep your thoughts and comments coming, we can always be reached at info@salestemperature.com or www.salestemperature.com.

 

 

How Does Machine Learning Work?

Here’s a nice quick read about how Machine Learning works.  A few points we’d emphasize: 1) Usefulness of current machine learning tools is a function of data and pattern recognition.  The more relevant data that the machine learning tool can utilize, the more likely it is to recognize patterns to improve accuracy in the assigned task.  2) In order to facilitate the learning, “researchers continually monitor the input data streams and make adjustments if necessary.”

Here at Sales Temperature, we are rapidly growing our user base, allowing us to feed our machine learning tool with additional retail sales data.  We are also constantly testing and adjusting external variables.  As a result, our machine learning tool is getting smarter every day and producing increasingly accurate sales forecasts.  Visit us at http://www.salestemperature.com to sign up for a free trial, and “Know Tomorrow’s Sales TODAY.”

Wall Street Journal Article Re: Employee Scheduling

Welcome to the Sales Temperature Blog!  Here we will try to highlight and discuss articles that we feel may be helpful and informative.  We recently read this article in the Wall Street Journal regarding some of the regulations that are being adopted by state and local governments to address concerns of part-time workers:

http://www.wsj.com/articles/local-governments-arrive-on-schedule-to-buttress-part-time-workers-1480161606

The article highlights three primary goals shared by many of these regulations.  First, provide employees with more notice of their schedules.  Second, give employees more access to extra hours.  Third, provide compensation to employees for last-minute scheduling changes.

We believe that these kinds of regulations highlight the need for retail managers to more proactively manage employee schedules.  This may seem like a daunting challenge, but we are here to help!  By utilizing advanced machine learning technology, Sales Temperature provides an accurate, continuously improving seven-day revenue forecast that can be used to predict labor needs and provide the insight and lead times to anticipate schedule changes.  All of which will help retail managers to effectively react to changes in regulations.

Accurate retail sales forecasting at your fingertips.