For years, the retail industry has relied on Annual Average Daily Traffic (AADT) to understand traffic around existing and planned locations. AADT originates from the transportation planning industry and is used by public departments of transportation to secure federal funding on a yearly basis. The practice relies on automated traffic counters or personnel performing on-site surveys. This data has served as a source for retailers and real estate companies for years as an indicator of potential foot traffic to their locations. However, AADT is no longer an accurate measure as a differentiating indicator because there are now more reliable sources of traffic data available, and traffic patterns are highly volatile in the post-COVID era.
”We did not leave the Stone Age because we ran out of stones, but because we found something better.”
– Don Huberts, Royal Dutch Shell, as quoted in The Economist, 1999
The new standard for today and the future is using Detailed Historical Traffic (DHT). It leverages a more timely, more accurate, and higher frequency source: probe data. This data provides greater coverage and reflects actual variations throughout the year. It offers many advantages to AADT, which is the reason why more organizations are migrating to this source to improve operational analytics.
What Is AADT?
Annual Average Daily Traffic has been around for a long time. This data generally provides average traffic in one-hour intervals each day as an average for an entire year at specific intersection points. As an example, users can determine the average traffic on Monday between 11 AM and noon during the past year. It can be used for site selection purposes or to understand traffic patterns at a high level. They are modeled from samples of actual traffic counts at certain intersections, which are then extrapolated to other intersections in the same vicinity using these actual counts with other data sources.
As stated, while AADT served its purpose and is likely sufficient for basic site selection, this data is collected at time intervals that are irrelevant given the plethora of information available today. As one customer put it:
“It is like someone saying that ice cream sales are better when the weather is hot than when it is cold; there is not much value to be derived from it.”
To be actionable from an operational standpoint, more accurate and more current data is required. Thankfully, there are now better options available to organizations.
What Is DHT?
Detailed Historical Daily Traffic has also been around for some time and was mostly used by public sector organizations. Only recently have retail organizations started to find value in it.
The data is gathered from probes on the road and is provided for every road segment in a geographical area. Specifically, a probe represents an actual vehicle on the road gathered from in-car navigation systems, telematics applications, or many different data sources. Therefore, it depicts a factual subset of all vehicles in an area and allows for accurate comparisons of different locations according to geography. Depending on the provider, probes account for anywhere between 15% and 40% of the actual vehicles on the road. DHT data is available within a 48 to 72-hour delay, in five-minute intervals, although most are aggregated at the hourly level every day for each road segment for up to five years.
Considerations When Choosing the Right Traffic Data
There are various factors to consider when evaluating a traffic dataset depending on the use case. The most important one, and the primary business driver, is the specific expectation and use of this data. If your only requirement is for site selection, then perhaps Annual Average Daily Traffic will be sufficient for your needs. Even in this context, the lack of spatial coverage and the assumption of a steady trend throughout the entire year make it less desirable. Additionally, if you have operational analytics requirements, want to use the data for competitive analysis, or need a more recent, hyperlocal view, Detailed Historical Traffic is a better alternative.
From a technical standpoint, here are other factors that should be considered when choosing the appropriate dataset:
This is obvious, but the data must address all or most countries where your organization operates.
The data is either sourced from raw probe data or modeled by data scientists using a combination of actual field collected data with other variables, like the population. Modeled data often creates problems for customers trying to use it to create their own models (in essence, you don’t model on modeled data).
The ability to readily use the data or the fact that it requires extensive data preprocessing and aggregation to be leveraged by the business. For Detailed Historical Traffic data, consider that you will be managing billions of records daily, which requires the right infrastructure and knowledge to leverage it properly.
This is one of the key evaluation factors from a technical standpoint. Field collected data can’t be refreshed daily in most cases, and sometimes the data was acquired five years ago for some locations. From a probe standpoint, data is rarely older than 48 to 72 hours. This freshness requirement is even truer now after the pandemic when traffic patterns have changed significantly.
They account for whether you will use the data for operational purposes, such as understanding road closures or near real-time changes in the traffic patterns of an area.
Update Frequency or Velocity
This addresses how often the data is updated for the end users. It could be yearly, quarterly, monthly, or even daily in the case of probe data.
Breadth of Data
This refers to the reliability of the source. As an example, if one provider has only a single source that reports on 10 million probes reporting 8 billion feeds per month and another has five different sources for 100 million probes reporting 150 billion feeds per month, the latter will obviously provide more reliable data. It is also about the sources of the collection points, such as field surveys or traffic camera surveys. The more sources you have, the greater their freshness and accuracy, and the easier it will be to identify the data that is right for you. As stated, depending on the provider, probes represent anywhere between 15% and 40% of the actual vehicles on the road.
Data Source Reliability
This involves whether the data collected is influenced by human behavior or is device-driven, and whether or not someone opt-in or opt-out of reporting information, such as checking in a location or accepting to be tracked by a mobile application, or whether he or she is automatically reporting information through a device like in-car navigation systems or telematics applications.
Certain providers offer data at the intersection or point level, while others offer data at the street segment or line level.
Level of Detail
This has already been addressed in some form, but if you need data for every single day in five-minute intervals versus monthly or daily intervals, then, you may choose one provider over another.
As you can see, choosing between AADT and DHT will depend on your business drivers, but organizations today tend to choose Detailed Historical Traffic because its capabilities address more business needs and cover almost all AADT use cases. Once that decision is made, the next step is selecting the right data provider, as there are multiple ones and, as described previously, there are many aspects to consider in choosing the right one.
Korem has extensive experience with traffic data and its business value for its customers. This expertise allows us to help them select the right provider and seamlessly integrate the data into their business processes or analytics.