AI, forecasting and real-time analytics for venues with Microsoft and Cisco Meraki

Leveraging Microsoft’s Azure SQL Data Warehouse and Cisco Meraki’s cloud-based networking infrastructure, Presence Orb are taking brick and mortar venues into the future with real-time ‘presence aware’ analytics and accurate forecasting powered by artificial intelligence (AI).

People using a space are tracked as their smartphones issue a “probe” request looking for a Wi-Fi signal. As the person moves through the site, the probe request is picked up by Wi-Fi access points (APs) on the client network, such as those manufactured by Presence Orb Technology partner Cisco Meraki. The APs release the data, which is streamed direct to Presence Orb in the cloud. Data is then processed and presented to space owners on a customised, web-based analytics dashboard.

Data collected includes footfall, street to venue conversion, dwell times, new vs. returning visitors, shared custom between venues and more. In addition, the Presence Orb dashboard pulls in open source data such as weather conditions to overlay against venue data. Third party data such as sales figures and staffing can be imported to both provide simple KPIs and to enrich actionable insights for operations management to drive sales, reduce costs and evaluate initiatives.

Exceeding 1.5 billion data points per month

This process produces a large and growing volume of data. Processing previously took around 24 hours, which prevented Presence Orb from providing clients with faster results for real-time marketing and operations management.

With data collection exceeding 1.5 billion data points per month across thousands of locations in over 10 countries, Presence Orb needed a solution that scales faster and would reduce processing time – a data warehouse.

We considered quite a few data warehouses, including Microsoft Azure SQL Data Warehouse, Amazon Redshift, and Google Cloud Bigtable. We assessed the ease of transferring over to those tools, and their capacity to work for us in the long term. After about three months of testing, we decided to go with Azure SQL Data Warehouse, primarily because it decoupled data storage from compute.

— Tom Sheppard, CEO and Founder of Presence Orb

Processing reduced from 24 hours to 90 minutes

Increased scalability of the data warehouse reduced data processing time to just 90 minutes in SQL Azure Data Warehouse, while the previous solution was taking closer to 24 hours to process. This saves both compute time and costs, and provides near-real-time benefits for Presence Orb clients.

Previously, a single deployment of data was provided once a day, meaning a client would log in today to see yesterday’s data. Now, using Microsoft Azure SQL Data Warehouse, near-real-time compute means clients can view and manipulate data from the last hour. In some cases, the default system can be bypassed and go into a true live system that shows how people are using the space at that exact time.

This allows venue or event managers to view real-time path mapping to see the flow of people through a venue so they can rearrange staff to handle changing traffic on the spot. It also means customers of those venues who have signed up to receive text or email offers can find out about in-store promotions there and then.

Presence Orb’s data journey with Microsoft

AI forecasting up to 94% accurate

With service improvements, Presence Orb has added Microsoft Azure Machine learning to provide predictive modelling for venues. A range of historical information about a venue is collected – such as footfall, dwell time, local transportation information, weather, local event information and proximity to competitors – and supplied to Azure Machine Learning’s artificial intelligence to predict what’s going to happen in a week’s time. For example, weather forecast will be taken for the next seven days and, based on past history, the likely footfall based on those weather conditions will be presented in the client dashboard.

The results help venue owners to staff, stock and manage operations more appropriately – resulting in reduced costs and increased revenue. If the system predicts a peak footfall due to a local event nearby, or if transportation data indicates a break in service due to maintenance, venues may need more or less staff that particular week. Presence Orb has been using Microsoft Azure Machine Learning since July 2016, and footfall prediction accuracy is already proven at between 92% and 94%.

Example location leader board on the Presence Orb client dashboard

Find out more

In Microsoft’s words, Presence Orb “offers a solution that uses the Wi-Fi technology already in place to provide brick-and-mortar retailers and venues with analytical tools similar to those available to online retailers. By analyzing this data, retailers and venue operators can make better decisions on staffing levels, hours of operation, crowd flow, and even promotions to improve their customers’ experience.”

Email for more information, or to arrange a no obligation discovery workshop.