When it comes to measuring store shopper trends, it can seem like there are no simple answers. The tech landscape for retail analytics solutions is crowded, making it difficult to clarify what’s noise and what will actually deliver value to your organization and therefore your customers.
You deserve the truth about your options and what you need to succeed. In this presentation, learn about your options for measuring foot traffic in your stores and how different tech can help you measure your success differently.
Full recorded webinar available here: http://bit.ly/2DGZf9F
8. We believe you need accurate customer counts in
order to deliver a consistent, meaningful experience.
9.
10. 1. Working definition of accuracy: Closest to 100% of visitor traffic measured consistently
2. Our criteria for people counting tech
○ Not just “What is our % accuracy” -- too small a picture
○ How often is our accuracy that %?
○ In which environments do we perform well or poorly?
Our goal: Consistently accurate counts across different environments,
which delivers the highest accuracy over time.
What do we mean when we talk about accuracy?
11. ● Given all your options, and what accuracy each provider can offer you, the way you
measure customer traffic matters.
● Keep your goal in mind.
○ Marketing success?
○ Saving on staffing?
○ Increasing conversion rates?
● The right solution for you is the one that gets you closest to your goal.
Your goals
13. ● Method: Bidirectional lasers on either side of a doorway
point at each other to form a beam.
● Accuracy: Counts anyone (or thing) that breaks the
beam. Prone to overcount. Lasers sometimes need to be
pointed back at each other.
● Installation: Usually done in-house using wired power.
Generally not network connected.
○ Sometimes built into loss prevention devices.
● What to keep in mind: Your team will need to log the
count manually in a spreadsheet via a view panel.
Break-beams
14. ● Method: Wifi captures unique sign-ons. Bluetooth
beacons emit a short-range signal to nearby phones.
● Accuracy: Prone to undercount, because it only counts
visitors with a device who log into Wifi or are in the
beacon’s range.
● Installation: Same as setting up Wifi in your store, or
sticking/installing beacons at key touchpoints.
● What to keep in mind: Beacons only count visitors who
have Bluetooth turned on and also have the necessary
app to be counted (yours or third-party).
Wifi/Bluetooth
15. ● Method: A recorded video feed is sent to the cloud,
where counts are then tagged via human labor,
machine learning or a combination.
● Accuracy: Usually high when calibrated properly.
Counts people in the cameras’ field. Lighting conditions
may affect accuracy
● Installation: Difficult; requires network bandwidth.
● What to keep in mind: This method captures
everything in the feed and then requires a heavy lift to
sift through that data.
Cameras
16. ● Method: Data from sensors that detect human heat
signatures is processed by machine learning.
● Accuracy: Usually high. Counts anyone that emits this
wavelength. Weather conditions can affect accuracy.
Machine learning algorithms learn from them.
● Installation: Depends on the sensor.
● What to keep in mind: These counts are anonymous,
which can be a privacy pro and an attribution con.
Thermal
17. ● Method: Visitors are counted by staff by hand, tallies are
compiled into a spreadsheet and sent to HQ.
● Accuracy: Depends who counts and what they think is
worth counting, how busy you are, how often counts get
reported in.
● What to keep in mind: When you ask staff to count
customers for you, you’re asking them to prioritize data
collection over the customer experience…and revenue.
Honorable mention: Manual
18. ● Accuracy
○ How do they define accuracy? Can they tell you how accurate their tech will be in
different environments?
○ Will you need to budget around regular calibration appointments from the solution
provider to maintain accuracy?
● Deployment
○ How long does installing a single sensor usually take? Could a store associate easily
do it or will you need additional resources?
○ Will you need to run power over ethernet or connect the sensor to an existing
network?
What to ask these tech providers
Thank you for joining us for The Retail Analytics Landscape: Choosing the best tech for counting customers.
Before we begin, we’d like to share some information on how to engage with today’s webinar. On your gotowebinar panel towards the bottom, please submit any and all questions you have for our speakers and we will do our best to get to them during and/or at the end of the webinar during Q&A. You can also engage with us on Twitter using the hashtag #dorwebinar.
Today’s webinar will also be recorded and sent to all of you and anyone else who has registered within 48 hours. Now on to our speakers!
Mike Lyons is a Senior Engineer who’s been at Dor since the beginning. He is driven by solving interesting problems and has a relentless curiosity that powers everything he does. Having built the brains of Dor from scratch, Mike is now responsible for making sure our machine learning algorithm keeps getting more accurate and delivers higher value to our customers. Before Dor, he was an engineer at Apple.
Joanna is the content marketing manager here at Dor:
She’s here because she loves empowering retailers audience with resources that inspire them to create meaningful in-store experiences
Verizon, Vend, Advancing Retail, Storefront, ASD
Forbes, Shopify, Fit Small Business, RetailWire
Today we’ll be discussing ...
Digital transformation = “what it takes to move from a buy low/sell high model that is product-centric, to an insight-driven model that is customer-centric.” https://www.forbes.com/sites/nikkibaird/2018/03/13/what-digital-transformation-actually-means-for-retail/#61d43e177038
According to a recent study by Oxford Economics, only 3 percent of retailers have completed company-wide digital transformation projects.
You’re probably here today because either:
Those meaningful experiences in your stores are disappearing, uncaptured, instead of being tracked well, or
Those meaningful experiences are being measured incompletely or inaccurately, and you know you need better data.
The purpose of retail is to create meaningful experiences that keep customers happy and coming back. The challenge of retail is measuring, replicating and optimizing the variables that make those experiences happen.
In order to build those meaningful experiences for your customer and understand how they are performing.
We believe you need an accurate customer foot traffic counter
But it’s a little crowded out there.
Let’s start at the core of what all of these technologies promise to deliver to you.
In order to have a conversation about the different types of tech that help you count your customers, we have to agree on what “accuracy” means.
Upkeep: Pointed at each other, Can’t distinguish between entrances and exits, tend to overcount
Counts must be divided in half.
Keeps counting someone standing in the middle. Prone to overcount.
Buy online, but not high quality.
We lumped these together because the methodology and value is similar.
Second easiest to install.
We did an experiment; Bluetooth beacon was counting less than 1% of people entering the store.
Low accuracy because it doesn’t count everyone. Undercounts.
Apple and Google have started obfuscating ad ID when you search for bluetooth and wifi networks because each time it pings, you have a new ID. No idea how many different phones in that store. 5 yrs ago, become less accurate in the past few years. Why new retailers are pushing their apps so hard.
Adopters are your least concern. Foot traffic isn’t what you already know.
Lighting conditions
Costly process, esp when involving humans. Sent to outsourcing dept to be tagged.
Regular calibration (3-6 months) sometimes incurs an extra cost
Way more data than they need to count people. Build sophisticated process to sort through what they don’t need. All of what’s happening all the time.
Security vulnerability.
Requires constant communication with it’s home server. If it loses connection then you will lose counts for that time
We’re really counting the blob centered on the top of their head entering and exiting a frame.
Passive sensor, counting light that travels to sensor, it’s easier to pick up light than to send out and receive. Heat data, not all visual data.
Build smart algos that count anonymously and send to our servers. We improve the ML running on the device.
HIPAA compliant
Who it counts: Anyone that emits this wavelength in 80-120F degree spectrum.
The oldest tech in the history books!
One point person? They miss people.
More? Overcounting.
Also inconsistent counting. Different people will count differently
Those counts aren’t timestamped so it can be difficult to impossible to drill in to your best and slowest hours of the day.
And though it’s unsavory, it is possible for staff to manipulate this data to their favor.
We’ll follow up with our Buyer’s Guide, which includes a full list of questions to guide your next call with a provider.
Thank you to Mike for joining me today, and a very special thank you to all of you who attended. The recording of this webinar will be emailed to you within the next 48 hours.
Please stay tuned for webinar and event invites coming soon. To learn more about Dor, visit www.getdor.com. We’ll see you next time!