Bottlenecks Are Not Always Where Traffic Looks Heaviest

The planned opening of a 128,000-square-foot Target store at Firestone City Centre in Colorado, part of a 140-acre mixed-use development, is the kind of project that can reshape local travel patterns before the first customer enters the parking lot. New anchor retail, restaurants, housing, and regional access points can turn ordinary arterials and intersections into recurring friction points during weekday peaks, weekend shopping periods, school dismissal windows, and event-driven surges.

That is the transportation question behind many land-use and infrastructure decisions in 2026: not simply whether traffic will increase, but where the network will begin to fail first, at what time of day, and whether the right intervention is signal retiming, turn-lane treatment, access management, ITS deployment, or capital construction. This is where bottleneck identification becomes an engineering discipline rather than a complaint map.

Ticon’s Traffic Analysis Approach

Ticon’s work in traffic analytics starts from a core finding that is easy to overlook: congestion is not explained by volume alone. In Ticon’s pandemic-period research, later published as Brodski, Stepanyan, Kozakevich, Vyazinko, and Naskidashvili, “Impact of Traffic Volume Variations on Travel Delays as Illustrated by Pandemic Period Data,” traffic flows were examined at 126 road intersections across nine U.S. states, with about 200 million datapoints analyzed. During COVID-related restrictions, traffic demand fell by up to 30 percent or more in many locations, yet delay reduction on signalized roads was much smaller than the drop in volume. In some cases, delay changed little even when demand was nearly cut in half.

For urban road networks, that result has direct planning implications. If a corridor remains slow after traffic volume falls, the source of congestion may lie in signal coordination, phasing, turning movement imbalance, saturation at a specific approach, or queue spillback from an upstream control point. Conversely, a road carrying high AADT may perform acceptably if its control strategy, access spacing, and available capacity match the demand profile. Bottleneck identification has to distinguish high-volume infrastructure from capacity-constrained infrastructure.

Highly Detailed Traffic Data Analysis

Ticon’s methodology is designed for that distinction. The platform combines permanent and portable detector data, traffic counters, GPS and connected-vehicle data, GIS information, demographics, traffic organization inputs, events, and other sources. These inputs are cross-verified, filtered, and processed through proprietary algorithms to estimate speeds, volumes, and derived traffic performance measures for approximately 95 percent of roadways. Coverage extends to more than 97 percent of roads at FRC 6 and above, with 100 percent temporal coverage. Spatial resolution can reach short road segments of up to 35 feet, with an average of about 225 feet, and temporal resolution can reach 5-minute intervals, with many cases processed at intervals as small as 15 seconds.

That granularity matters because bottlenecks are often local and time-specific. A corridor-level average can hide a failing left-turn pocket, a short merge section, an access driveway near a retail entrance, or a signal phase that works at 8:00 a.m. but collapses at 5:15 p.m. For a mixed-use project like Firestone City Centre, the same intersection may experience different stress patterns during commuter peaks, lunch traffic, weekend retail peaks, and residential return trips. AADT alone cannot identify those patterns. Engineers need intraday volumes, speeds, saturation conditions, and turning movements by time period.

Precision in Traffic Intervention Planning

Ticon’s intraday volume estimation work supports this level of analysis. In its “Ticon intraday traffic volumes estimation” white paper, the platform estimates traffic volumes using road geometry, intersection geometry, road segment connections, speed distributions, expected vehicle composition, and driver behavior under different conditions. At the AADT level, Ticon reports a median average percentage error of 4.78 percent and relative root mean square error of 11.97 percent, keeping expected volume estimation error within 20 percent boundaries with 90 percent confidence. For more granular field comparisons, median average percentage discrepancy between hourly Ticon estimates and detector measurements ranges from 5 percent to 20 percent, depending on infrastructure data availability. In cases with ample traffic infrastructure data, 15-minute traffic flow volumes were estimated with a median absolute percentage error of 11.24 percent, compared with 6.5 percent from a pre-calibrated video detector under the same circumstances.

These figures are important because intervention planning often depends on relatively small timing and capacity differences. If an intersection approach is saturated for only two 15-minute windows each weekday, the correct response may be targeted signal timing or a time-of-day operating plan. If saturation persists through multiple hours and queues interfere with adjacent intersections, the solution may require access control, lane reassignment, geometric change, or broader corridor coordination. Without reliable temporal detail, agencies risk spending capital money on a problem that could have been solved operationally, or applying signal changes where geometry is the real constraint.