What We've Found Analysing Footfall Data in Retail

5 Trends We've Noticed...

"At HoxtonAi, we work with customers to ensure their data is high quality, presented in the right way, and integrated into their workflows.  After speaking to hundreds of customers about how they want to use street traffic and entrance count data, we've noticed five emerging themes which we're delighted to share with you."

1. Get the data right at the start.

Unlike other providers, our system has been built for ease of installation, so our customers set up and configure their own sensors and system, but that doesn’t mean they’re on their own.  We walk them through the deployment step-by-step to ensure they are collecting the data they need.  We then provide an accuracy audit, and whilst there can be issues at the outset, such as strange angles or video screens distorting the data, getting this right at the start means we can rely on high-quality data for many years to come.

2. Customers may want the same data point, but they want it for different things.

We’ve helped customers that need the same data for many different reasons.  Some retailers wanted the system to improve staff rotas, another for monthly reporting, and another for A/B testing window display.  It’s important to have a wide-ranging discussion with the client to make sure we understand their business priorities and that our solution is ideal for what they are looking to achieve.

3. The unknown unknowns.

Part of what we do that we love most is helping clients use their footfall data to optimise store performance in ways they didn't realise were possible.  For example, one client wanted passing street traffic data for one single store to help with their rent review.  However, with the platform up and running, we discussed how they could use our API to embed in their own CEO reporting dashboard.  Now it’s become an essential part of their weekly reporting, and they rolled out the sensors to all of their stores in the UK.

4. Strange things can be seen in the data.

 Occasionally, some surprising results arise from the visitor counting data, and it’s our job to play detective and get to the bottom of it.  One such occasion was when a store on a quiet street in central London had a low-footfall day inside the store, but had huge numbers of passing traffic – over 200,000.  The client noticed at the month end, and questioned the data. The game was afoot.  Could it be a glitch in the system? A crashed server, or a database error?  On conducting our research, the answer was far more mundane: on this day, there was a major demonstration in central London, and the route changed at the last minute – to pass the store due to roadworks!  We let them know and were then able to exclude this data from their reporting.

5. Cover all bases.

Our technology can provide real-time occupancy – not just counts- by deploying sensors at each entrance and merging the data together in real-time.  One client dataset had clear errors – it appeared that at the end of the day, there were always 50 people stuck in the store at closing time!  Thankfully, this was resolved just a few days after the error was noticed by speaking to the store manager.  After opening, many staff entered through the front entrance and exited through the (unmonitored) side entrance.  Once they exited and entered out of the back, the problem went away.

What it's all about.

I hope this gives a little insight into some themes we see at HoxtonAi.  Applying data to a customer's need can be a messy business, but we love to dig deep, understand what the customer is looking to achieve and make sure that’s the very least they end up getting.