Verifying In-store Footfall Accuracy Using Sales Performance Results
This article shows how mobility data can be used to accurately predict net sales results for consumer- driven businesses like Walmart, the leading US big box retailer.
The Objective
The goal is to extract results from Huq Industries’ Footfall and Dwell-time modules that are highly correlated with the net sales figures published by Walmart in its quarterly trading updates. This output can then be incorporated into downstream models and systems to inform research for analysts and investors.
Step by Step Guide
Step 1: Get Walmart net sales data
Walmart publishes its net sales data within its quarterly trading updates, which can be found on its Investor Relations webpage. Note that for the purposes of this analysis, use is made only of the ‘Walmart US’ net sales segment.
Step 2: Find Walmart store visits
Huq’s event-level mobility data serves as a proxy for consumer demand across Walmart’s US stores. This event-level data is pre-enriched by Huq Industries in a process that extends raw geo-spatial data to include point of interest (POI) attributes such as business name, type and location.
Extracting Walmart visit data is therefore made easy by filtering on the raw place name (place_name), the standardised name (brand_name), or indeed by ignoring the POI attributes and using the WiFi SSID observed by the mobile device (impression_ssid) to match Walmart-specific patterns.
Step 3: Extract Walmart store visits
One way to quantify demand across the Walmart estate using Huq’s enriched geo-spatial data would be to count the number of distinct mobile devices (ie. panelists) present at Walmart each day. This approach can be useful in many analyses but after much experimentation there is a second strategy that produces results that more closely reflects behavioural nuance – and this is related to dwell.
During the same enrichment process that supplies the point-of-interest attributes, an estimate of dwell is also calculated and added to the resource. These properties can be accessed and manipulated via the columns impression_dwell_lower_bounds and impression_dwell_upper_bounds, which represent the upper and lower estimate for visit duration. These are calculated using the cumulative elapsed time calculated by continuous observations of a mobile device in the same place.
For the purposes of this study however, it is beneficial to develop a measurement of dwell that is less strict. The strategy used in this exercise groups observations into visits where they occur within 65 minutes of each other – without requiring them to be continuous per se.
Step 4: Test and training datasets
Walmart provides seventeen quarters of historical results via its investor portal. These are split into two sets; one to use for our training set and the other to test our output against. The training set helps us to select the optimal combination of parameters from the signal candidates described in Step 6. How this is evaluated is explored in Step 7.
The first twelve rows are chosen for training, and the remaining five are retained for testing. Separating them chronologically avoids test information leaking into the training process and causing lookahead bias.
Step 5: Define rough signal form
At this point take the training set from Step 4 then prepare it by eliminating behavioural outliers and concretely defining our measure of ‘dwell’.
i. Preparing the data
It is very common to find elements of noise within a dataset, and Huq’s enriched ‘Events’ feed is no exception. Some characteristics may be derived from interference at the sensor level; some artefacts may be behavioural and completely natural. As our demand metric relates to dwell, it is necessary to eliminate data points that show excessively little or large dwell before applying it. Filtering the data in this way excludes facets such as Walmart employees or other false positives derived from the enrichment process. Our chosen strategy filters the detected dwell value by an upper threshold Du , a lower threshold Dl and also determines whether to filter on a daily or per-visit basis, Db.
ii. Transform dwell into ‘demand signal’
Let’s assume that dwell-time and spending money in- store is a non-linear relationship. Specifically, let’s suggest that there is a ‘normal’ level of dwell Bm and a ‘normal’ propensity for Walmart store visitors to spend, Bbase, both of which are constants.
We can then raise Bbase by the difference between the observed dwell value and the ‘normal’ value, Bm. To keep this value from exploding or vanishing, it is expedient to truncate the difference in the range of Ol to Ou before raising the power.
This can be summarised as follows:
iii. Normalise the data for panel growth
The size of Huq’s mobility panel changes over time as the number of apps supplying data increases, and apps’ own audience sizes fluctuate. As this study results in a time-series output, it is imperative to account for these changes in the normalised result so as to accurately represent the real trend. Similarly, it is also important to account for growth in the number of daily measurements observed per device using Huq’s measurement software in order to maintain a consistent view of ‘dwell’.
It may also help to consider how these characteristics vary geographically. The normalisation strategy employed in this exercise works by dividing the ‘demand signal’ observed across Walmart locations by the equivalent metric for the full US panel on equivalent day. This approach may be further improved by normalising on a localised basis to account for regional variations in data coverage, and by pre-filtering the data to remove individual app or panelist outliers.
iv. Respect seasonality in signal generation
Different week parts – weekdays, weekends and public holidays – have significance for in-store retail behaviour, and it is beneficial to recognise this in signal preparation. Accordingly, the normalised output is grouped using this classification, and is supplied to the model independently.
Step 6: Test and training datasets
The many possible combinations of parameters in Step 5 produce a huge number of candidates for signal representation, numbering 750K+. So, which candidate set offers the closest match to Walmart’s net sales values? A simple regression model (see Step 7) allows us to identify the best candidate set.
Step 7: Regression and validation
On the basis that we can expect longer dwell-times to lead to higher net sales results, a suitable model to use in this instance is non-negative least-squares regression as the inductive bias is well suited to this problem.
How do we know which signal output is best suited to net sales prediction? We’ll look for the result with the smallest mean absolute percentage error (MAPE), and use ‘leave-one-out cross validation’ to make best use of the limited supply of training data available in Walmart’s quarterly net sales figures archive.
The end result

The Pearson correlation between the output of the regression model and Walmart’s actual net sales figures on the (completely unseen) test set is ρ=0.85, with a MAPE of just 3.8%.
Conclusions
Using this parameter selection strategy it emerged that the optimal preparation steps and parameters are:
(i) remove dwell-time outliers, keeping daily device dwell values in the range of 0 to 160,
(ii) construct ‘demand signal’ by subtracting 30 from the daily dwell figure, and bound to the range of -12 to 60, then raise to the power of 1.03
(iii) normalise by counting 2hr-truncated timestamps across the full US dataset, where devices must have visited a non- residential location
Lastly, aggregate results by day and divide the ‘demand signal’ by this value.
Insight reliability is key
Senior Data Scientist, Large US Asset Manager said –
We have tested Huq’s footfall and dwell-time data in our forecasting models and found that it added significant benefits to the accuracy of our signal.”
How Aviva Investors Used Huq's High-Frequency Insights to Invest During Covid-19
The Pandemic has had systemic consequences for places and how people use them. Covid-19 has accelerated change in existing trends. It has also created new ones.
Aviva Investors, the global asset management business of Aviva plc (‘Aviva’) has sought to build these thematic changes into its investment process. The speed of change made decision-making complex. Trends that unfolded over years changed shape in days.
The Challenge
The challenge for Aviva Investors was how to stay informed. Conventional indicators report monthly at best. Other data lacked the detail to be useful. Aviva Investors needed access to high-frequency data with granular capabilities.
The Solution
As one of the largest European real assets investment managers, Aviva Investors looked for data that reflected how much activity is taking place in different parts of the economy. Key metrics centre on how people use places, and how that changes with time. Topics include the measurement of pedestrian footfall levels, store-visits and workplace presence.
Huq Industries offers effective measurement of these KPIs through its macro indicators outputs. This mobility-derived product allows for comparisons by sub-market, or industry, and geographic region. Huq’s mobility data has full timestamp coverage. This enables its macro indicators to be published daily with a 48-hour lag. These attributes make it ideal for measuring fast-paced change in great detail.
Over the last 15 months Huq has supported Aviva Investors in its analysis of UK and European Real Asset spatial trends. Aviva Investors draws on 30+ verified indices from Huq Industries’ Macro Indicators catalogue. Sectors measured are as diverse as waste plants, hotels, data centres and airports. In addition, regional outputs contrast office usage across 16 major cities on 3 continents. Huq’s measurement platform for cities and retail, provides Footfall and Catchment Area data and how that changes with time.
The Results
Aviva Investors used Huq’s macro indicators and insight solutions to track changes in behaviour during the pandemic. The frequent and granular data helps to inform their decision-making. Insights derived from Huq’s mobility solutions form the basis of research and articles written by Aviva Investors’ research team for their investment partners.
Huq’s mobility data allows us to measure footfall, almost in real time, by sub- market and by region.” said Jonathan Bayfield, Head of UK Real Estate Research at Aviva Investors. “The data helps us to make better investment decisions and allows us to appropriately manage risk on behalf of our investor clients”
Transport
Transport
Build transport networks that work
Are you still using 2010 census data to drive your transport models? Are costly, infrequent surveys holding you back? Get up-to-date data now!
Having granular up to date origin-destination insight is a huge help in transport planning.
- Senior Transport Manager,
City Council



What is Transport?
The Transport module is an origin-destination matrix that describes the strength of the relationship between every area of the country.
Why use it?
Origin-destination matrices are used to plan more effective routes between centres, and to manage collateral risk during road and transport closures. It's critical to the smooth running of operations and keeping traffic flowing.
- Incl. year-to-date 2022 data
- Percentage flow between areas
- Standard MSOA as area unit
- Full nationwide coverage
- Offers 3 built-in journey types
- Forward and reverse flows
- Includes journey flow data
- Income demographic filtering
- 4+ years' history available
- Multiple centres supported
- Export data as CSV file
- Download live reports as PDF
- Hardware-free solution
- Instant monitoring setup
- Data accuracy validated
- Training & support included
- Year-on-year comparisons
- Used by 50+ UK councils
"How can I detect and assess routes that need change?
Transport Planning Manager, County Council
"How can we divert traffic without impacting on the local network?
Transit Infrastructure Manager, Borough Council
"How do we know whether the interventions we made worked?
Senior Transport Insights Manager, County Council
No hardware. Instant setup. History included out of the box.
Weekly Updates
Monitor performance across the places and centres you manage in near real-time. Use high-frequency insights to plan and react at pace.
4yrs History
Huq provides up to 4yrs of monitoring history for every new location out of the box, making annual comparisons fast and easy.
Instant Setup
Get access to Huq's monitoring platform today! Instant setup. No hardware, cameras or any other infrastructure needed.
UK Coverage
All Huq's place monitoring products are available for any location in the UK and beyond. Any place, any size, anywhere - country wide.
Related products
Often bought together
Explore companion products from our insights platform!
Browse modules ➜



One-to-one customer success support built in
Huq's unique Customer Success offering provides hands-on training and support in reports creation for each and every one of its customers. Learn to interpret, visualise and talk about your data!
- Hands-on user training
- Custom report building
- Expert advice & support
Granular Catchment
Granular Catchment
Where do most visitors come from?
Discover where visitors to places travel in from. This precision module offers the ability to quantify how many visitors come from each postcode district. Use Granular Catchment insights to enrich your footfall data!
This granular insight allows us to quantify how many visitors come which postcodes
- Visitor Insights Manager,
National Park Authority



What is Granular Catchment?
Granular Catchment is the first ever product that quantifies number of visitors to specific destinations according to where they travel from.
Why use it?
Use Granular Catchment to understand the impact of accessibility projects - ie. public transport improvements - on local and wider mobility. Explore the impact of events and interventions in terms of town centre appeal. Measure how seasonality affects how tourists visit an area.
- Export data as CSV
- Uses postcode district units
- Footfall insight enrichment
- Understand seasonal trends
- Measure accessibility projects
- Monthly reporting cycle
- Up to 4+ years' history
- Quantify tourism volumes
- Available for any town or centre
- Download live reports as PDF
- Hardware free solution
- Instant monitoring insight
- Full nationwide coverage
- Data accuracy validated
- Training & support included
- Year-on-year comparisons
- Used by 50+ UK councils
"Has increased marketing spend attracted visitors from further away this summer?
Economic Development Officer, County Council
"Since our transport improvements have we seen positive in accessibility?
Senior Transport Planner, County Council
"What's our conversion rate for visitors travelling from this location?
Store Planning Manager, National Multiple Retailer
No hardware. Instant setup. History included out of the box.
Weekly Updates
Monitor performance across the places and centres you manage in near real-time. Use high-frequency insights to plan and react at pace.
4yrs History
Huq provides up to 4yrs of monitoring history for every new location out of the box, making annual comparisons fast and easy.
Instant Setup
Get access to Huq's monitoring platform today! Instant setup. No hardware, cameras or any other infrastructure needed.
UK Coverage
All Huq's place monitoring products are available for any location in the UK and beyond. Any place, any size, anywhere - country wide.
Related products
Often bought together
Explore companion products from our insights platform!



One-to-one customer success support built in
Huq's unique Customer Success offering provides hands-on training and support in reports creation for each and every one of its customers. Learn to interpret, visualise and talk about your data!
- Hands-on user training
- Custom report building
- Expert advice & support
Visit Frequency
Use visit frequency to know how often people return to your towns|
Get reliable visit frequency insights made for decision makers in Local Government, BIDs, Retail and Real-estate teams.
Frequency of visit is a core indicator of success in our centres.
- Economic Development Officer,
County Council



Visit Frequency Counting
How often do visitors come back?
Visit Frequency is a monthly measure of how frequently unique visitors return to the place, street or centre you manage.
- Monthly reporting
- Income demographics
- Residents excluded
- Export as PDF/CSV
- Over 4 years' history
- Available nationwide!
Getting started is easy
Simply trace your places, and go!
Get instant frequency of visit data for any place of any shape or size - UK wide. Use our specialist place tracing tools to define the area you want to cover and get accurate results today!



Get the free report
In 2021-22 UK Govt. DHLC made £56 million in funding available to UK councils.
Huq analysed footfall performance for centres UK-wide to rank the greatest winners during that period.
Start measuring visit frequency today!
Step 1: Get a personal demo
Get a personalised demo with our team of experts
Step 2: Trace your places
Define the areas that you would like visit frequency data for
Step 3: Get monitoring!
No equipment, no installation. Start monitoring your places today!
Module pricing explained
Platform pricing is simple. Access costs £1,000 + VAT per insight module per place (ie. street, park, mall) per year. We offer volume-based discounts too!
Basic
We're just getting started- 1 Insight module
- 1 Place monitored
- Footfall, Dwell or Catchment
Grow
Get deep place insights- 3 Insight modules
- 1 Place monitored
- Basic + Density & Frequency
Scale
For PRO measurement teams- 3 Insight modules
- 3 Places measured
- 360-degree insights
Hourly Footfall
Hourly Footfall
When are we busiest?
Find out when the peak times are for your green spaces, town centres, shopping centres and retail parks. Use Hourly Footfall to track performance.
Knowing when people choose to visit tells us a lot about why they come
- Night Time Economy Manager,
District Council



What is Hourly Footfall?
Hourly Footfall is a measure of the number of unique visitors present at an area split out by hour of day and day of week.
Why use it?
Understanding when people visit informs how best to manage and maintain local spaces. Compare weekdays and weekparts to track trends in commuting patterns, leisure activity and the effect of your interventions. Retailers use Hourly Footfall to help optimise their retail estate portfolios.
- Daily hourly footfall reports
- View trend and daily results
- Get actual and indexed values
- Make year-on-year comparisons
- Any place, any size, anywhere
- Up to 4+ years' data history
- Benchmarking data available
- Full nationwide coverage
- Compare with multiple places
- Income demographic filters
- A zero-hardware solution
- No installation or maintenance
- Export results data as CSV
- Download live reports as PDF
- Fine-grained date filters
- Data accuracy validated
- Training & support included
- Used by 50+ UK councils
"What's the busiest day for footfall locally?
Economic Development Officer, County Council
"What's the optimal time for maintenance?
Parks & Open Spaces Manager, City Council
"When can we expect peak traffic flows?
Transport Planning Manager, District Council
No hardware. Instant setup. History included out of the box.
Frequent Updates
Monitor performance across the places and centres you manage in near real-time. Use high-frequency insights to plan and react at pace.
4yrs History
Huq provides up to 4yrs of monitoring history for every new location out of the box, making annual comparisons fast and easy.
Instant Setup
Get access to Huq's monitoring platform today! Instant setup. No hardware, cameras or any other infrastructure needed.
UK Coverage
All Huq's place monitoring products are available for any location in the UK and beyond. Any place, any size, anywhere - country wide.
Related products
Often bought together
Explore companion products from our insights platform!
Browse modules ➜



One-to-one customer success support built in
Huq's unique Customer Success offering provides hands-on training and support in reports creation for each and every one of its customers. Learn to interpret, visualise and talk about your data!
- Hands-on user training
- Custom report building
- Expert advice & support
Footfall
Use footfall data to learn how many people visit your towns|
Get the leading footfall counting system built for decision makers in Local Government, BIDs, Retail and Real-estate teams.
Footfall monitoring is the single most important insight we use to manage places
- Head of Economic Development,
County Council



Accurate Footfall Monitoring
How many people are in this place?
Learn how many unique visitors are present in the places you manage, and how that changes with time. Verified people counting methodology. No double counting!
- Daily footfall counts
- No hardware needed
- Available UK-wide
- Any place of any size
- Up to 4+ years' history
- Demographics included
Getting started is easy
Simply trace your places, and go!
Get instant footfall counting data for any place of any shape or size - UK wide. Use our specialist place tracing tools to define the area you want to cover and get accurate results today!



Get the free report
In 2021-22 UK Govt. DHLC made £56 million in funding available to UK councils.
Huq analysed footfall performance for centres UK-wide to rank the greatest winners during that period.
Start measuring footfall traffic today!
Step 1: Get a personal demo
Get a personalised demo with our team of experts
Step 2: Trace your places
Define the areas that you would like footfall data for
Step 3: Get monitoring!
No equipment, no installation. Start monitoring your places today!
Module pricing explained ?♂️
Platform pricing is simple. Access costs £1,000 + VAT per insight module per place (ie. street, park, mall) per year. We offer volume-based discounts too!
Basic
We're just getting started- 1 Insight module
- 1 Place monitored
- Footfall, Dwell or Catchment
Grow
Get deep place insights- 3 Insight modules
- 1 Place monitored
- Basic + Density & Frequency
Scale
For PRO measurement teams- 3 Insight modules
- 3 Places measured
- 360-degree insights
Dwell
Use dwell-time to find how long people spend in your towns|
Get accurate visitor dwell-time insights made for decision makers in Local Government, BIDs, Retail and Real-estate teams.
Dwell-time provides leading indicators on the performance of the local economy
- Head of Economic Development,
County Council



Dwell-time Monitoring
How long do visitors spend in this place?
Dwell-time is a measure of the average time that visitors spend within an area per trip. Get the output in minutes, updated everywhere on a daily basis!
- Daily dwell-time reports
- Results given in minutes
- Organise by day of week
- Group by hour of day
- Over 4 years' history
- Available nationwide!
Getting started is easy
Simply trace your places, and go!
Get instant visitor dwell-time data for any place of any shape or size - UK wide. Use our specialist place tracing tools to define the area you want to cover and get accurate results today!



Get the free report
In 2021-22 UK Govt. DHLC made £56 million in funding available to UK councils.
Huq analysed footfall performance for centres UK-wide to rank the greatest winners during that period.
Start measuring visitor dwell-time today!
Step 1: Get a personal demo
Get a personalised demo with our team of experts
Step 2: Trace your places
Define the areas that you would like dwell-time data for
Step 3: Get monitoring!
No equipment, no installation. Start monitoring your places today!
Module pricing explained
Platform pricing is simple. Access costs £1,000 + VAT per insight module per place (ie. street, park, mall) per year. We offer volume-based discounts too!
Basic
We're just getting started- 1 Insight module
- 1 Place monitored
- Footfall, Dwell or Catchment
Grow
Get deep place insights- 3 Insight modules
- 1 Place monitored
- Basic + Density & Frequency
Scale
For PRO measurement teams- 3 Insight modules
- 3 Places measured
- 360-degree insights
Density
Density Monitor
Discover local visitor hotspots
Find the most popular parts of towns, streets and green spaces with visitor density monitoring.
Density monitoring helps us find the most used parts of our towns
- Town Centre Manager,
Borough Council



What is Density Monitoring?
Density monitoring is a visualisation of pedestrian flows and hotspots across whole towns, streets and centres. Density maps are updated daily.
Why use it?
Determine the places that attract most visitors to the areas you manage. Find the most valuable retail opportunities for within a centre or street. Learn how best to manage and maintain parks and spaces based on usage.
- Daily density heatmap outputs
- Hotspot granularity to 10m
- Density score value provided
- Measurement of the full area
- Any place, any size, anywhere
- Up to 4+ years' data history
- Fine-grained date filters
- Compare with multiple places
- Download live reports as PDF
- A zero-hardware solution
- No installation or maintenance
- Full nationwide coverage
- Data accuracy validated
- Training & support included
- Year-on-year comparisons
"Have our changes brought people to new areas?
Town Centre Manager, Local District Council
"Where do different income groups spend time in our town?
Parks & Green Spaces Manager, County Councity
"Shall we open a store at this end of the road?
Estate Planning Manager, Multiple Retailer
No hardware. Instant setup. History included out of the box.
Weekly Updates
Monitor performance across the places and centres you manage in near real-time. Use high-frequency insights to plan and react at pace.
4yrs History
Huq provides up to 4yrs of monitoring history for every new location out of the box, making annual comparisons fast and easy.
Instant Setup
Get access to Huq's monitoring platform today! Instant setup. No hardware, cameras or any other infrastructure needed.
UK Coverage
All Huq's place monitoring products are available for any location in the UK and beyond. Any place, any size, anywhere - country wide.
Related products
Often bought together
Explore companion products from our insights platform!



One-to-one customer success support built in
Huq's unique Customer Success offering provides hands-on training and support in reports creation for each and every one of its customers. Learn to interpret, visualise and talk about your data!
- Hands-on user training
- Custom report building
- Expert advice & support
Catchment
Use catchments to find the origins of visitors to your towns|
Get accurate catchment area maps made for decision makers in Local Government, BIDs, Retail and Real-estate teams.
Catchment area analysis is the single most important way we characterise visitors to our town centres
- Senior Town Centre Manager, District Council



Catchment Area Modelling
Where to our visitors come from?
Catchment areas, or models, are the standard way to represent the dominant locations that visitors to a town, place or centre travel from.
- Monthly reporting
- Map visualisations
- Residents excluded
- Fine date controls
- Over 4 years' history
- Available nationwide!
Getting started is easy
Simply trace your places, and go!
Get instant visitor catchment data for any place of any shape or size - UK wide. Use our specialist place tracing tools to define the area you want to cover and get accurate results today!



Get the free report
In 2021-22 UK Govt. DHLC made £56 million in funding available to UK councils.
Huq analysed footfall performance for centres UK-wide to rank the greatest winners during that period.
Start measuring visitor catchment areas today!
Step 1: Get a personal demo
Get a personalised demo with our team of experts
Step 2: Trace your places
Define the areas that you would like catchment data for
Step 3: Get monitoring!
No equipment, no installation. Start monitoring your places today!
Module pricing explained
Platform pricing is simple. Access costs £1,000 + VAT per insight module per place (ie. street, park, mall) per year. We offer volume-based discounts too!
Basic
We're just getting started- 1 Insight module
- 1 Place monitored
- Footfall, Dwell or Catchment
Grow
Get deep place insights- 3 Insight modules
- 1 Place monitored
- Basic + Density & Frequency
Scale
For PRO measurement teams- 3 Insight modules
- 3 Places measured
- 360-degree insights