From Reactive Enforcement to Predictive Safety Investment 

How Agencies Use Data-Driven Traffic Safety Enforcement 

Data-driven traffic safety enforcement uses crash history, speed data, and roadway characteristics to decide where enforcement resources are deployed before a corridor’s crash pattern worsens, not after. Whether the decision sits with a state DOT scoping a work zone program or a city council and police department weighing school zone and red-light enforcement, that distinction shapes RFP language, first-year budget justification, and how the program is evaluated three years in. Agencies that build predictive data use into the procurement stage, rather than treating it as an add-on, tend to reach stakeholder buy-in faster and have an easier time defending the program at renewal. 

From Crash Response to Crash Prediction 

Most legacy safety programs are reactive by design: a location gets attention after a cluster of crashes, a fatality, or a public complaint. The Federal Highway Administration has spent the past decade pushing agencies toward the opposite model — using data to find the highway’s problem sites before the crash history proves it. 

FHWA’s roadway safety data guidance describes this directly: crash data alone leaves practitioners with a purely reactive approach, identifying locations only after crashes have already happened. Adding traffic volume and roadway characteristics allows agencies to estimate expected crash frequency and compare crash risk across corridors with very different levels of service.

The models behind that estimate — safety performance functions — are what make the comparison defensible. Because an SPF is calibrated using data from an entire state rather than a single location’s crash history, its predicted value isn’t distorted by regression to the mean, the statistical artifact that makes a bad crash year look like a permanent problem. That lets agencies rank many sites by predicted crash frequency instead of reacting to whichever corridor had the worst recent stretch. 

That same logic extends naturally into enforcement siting and program design. A corridor with rising speed variance and increasing violation rates is a different investment case than one with an established multi-year crash history — but only if the data is being tracked and reviewed on a rolling basis, not just at renewal. 

What This Looks Like in Practice 

Two long-running state work zone speed programs show what data-anchored programs produce — and, in Delaware’s case, what enforcement adds beyond signage alone. Both sets of figures below come from the agencies’ own published evaluations, not vendor reporting, which is exactly the division of labor described later in this post. 

Maryland’s SafeZones program, authorized under Transportation Article §21-810 and operating since 2010, has published some of the most detailed before/after data of any state program: state figures show a more than 90 percent reduction in vehicles traveling 12+ mph over the work zone limit at monitored sites, alongside multi-year declines in work zone crashes and injuries. 

Delaware’s Work Zone Safety Program makes a sharper point about what enforcement itself contributes. In the I-95 and SR 896 interchange work zone, DelDOT’s published reporting shows construction drove monthly crash rates up substantially, and reducing the posted work zone speed limit alone did not bring them down. Automated enforcement did: total crashes fell 22 percent, and injury crashes 23 percent by rate, measured against the immediately preceding period with the most similar operating conditions. Average speeds dropped more than 7 mph northbound. For an agency weighing a camera program against the cheaper option of simply posting a lower limit, that sequence settles the question. 

Procurement Considerations for a Predictive Program 

For procurement officers, public works directors, and command staff scoping a new RFP, the data-first framing above becomes concrete evaluation criteria. Several of these are not preferences. They are already law in a number of states, and a bid that gets them wrong is a bid that cannot be awarded. 

Check your compensation statute before writing the RFP 

Vendor compensation tied to citation volume or fine revenue is prohibited outright in several authorizing states. Colorado requires that compensation be based on the value of the equipment and services provided, and bars any portion of a collected fine from being paid to the vendor. Washington’s automated traffic safety camera statute requires compensation be based only on the value of equipment and services rendered. Connecticut states plainly that a vendor’s fee may not be contingent on the number of citations issued or fines paid. So, for most agencies the question is not whether to structure compensation this way — it is whether the proposed structure complies, and whether it is disclosed publicly in plain language. Doing that disclosure early heads off a “cash grab” narrative before it starts.

Budget for declining revenue 

The automated enforcement checklist published jointly by AAA, Advocates for Highway and Auto Safety, GHSA, IIHS and the National Safety Council is direct on this point: programs should be data-driven and prioritize safety over revenue, and communities should expect revenue to decline over time as fewer drivers speed or run red lights. Build the declining curve into the first-year forecast, and the program’s safety results and its budget projections move in the same direction rather than against each other.

Separate operational metrics from outcome metrics 

Mean speed reduction, time into red, and compliance rates are the operational signals a program can act on mid-year. Crash outcomes come from agency and state records, on a longer lag. Both belong in a program review, but they should be labeled for what they are: a vendor can cite agency-published crash results, with the source named. What it should not do is present them as its own measurement. 

Agree how the program will be judged before it starts. 

Most programs are evaluated by comparing crashes before the cameras to crashes after. It sounds obvious, and it is a weak argument. Crash counts move for reasons that have nothing to do with enforcement — traffic volumes, weather, the economy — so a skeptic can always offer another explanation. The safety coalition’s checklist recommends comparing your corridor against similar control sites instead. It also warns that cameras tend to reduce crashes across a whole community, not only where they are installed, which means a poorly chosen control site makes your program look less effective than it is. Settle the method in the RFP. At renewal you want to be defending your results, not arguing about how they were measured 

Confirm how you reach your own data 

Ownership is usually straightforward; access is where programs get stuck. Ask what staff can see without filing a request, how current the data is, and what exports are available for the agency’s own analysts and auditors. 

Make calibration independently verifiable 

Are calibration and accuracy records available to the public on request, and can a third party check them? 

Treat public information as program infrastructure 

The safety coalition checklist frames transparency and public communication as core program design rather than marketing, and recommends convening a stakeholder advisory committee — law enforcement, transportation staff, victim advocates, civil rights advocates, school officials, residents — during the design phase. It also recommends taking racial and economic equity into account in camera placement and fine decisions, language that increasingly appears in RFPs and in council votes. 

Programs built around compliant compensation structures, transparent calibration reporting, agency-controlled data, and a pre-agreed evaluation method tend to hold up better across multiple contract cycles and changes in political leadership than programs where the financial structure creates even the appearance of incentivized enforcement. 

The Legislative Landscape Is Shifting Toward Data-First Programs 

North Carolina authorized automated speed enforcement for cities and counties under G.S. 160A-300.4 and G.S. 153A-246.1, via S.L. 2025-47, effective October 1, 2025 — giving agencies a fresh legislative framework to build data governance and reporting expectations into their first RFPs rather than retrofitting them later.

Massachusetts is positioned to move next: S.2344 and Governor Healey’s FY2027 budget proposal both signal 2026 as the most likely year for statewide authorization. Because the bill vests authority with municipalities, that puts city managers, councils, and police departments in the position of designing evaluation criteria now, ahead of a competitive rush once authorization passes. 

Michigan has authorized work zone speed camera enforcement under HB 4132 and HB 4133 — but authorization alone doesn’t start a program. Funding remains the gating factor, which puts Michigan agencies in the position of doing site analysis, stakeholder alignment, and procurement groundwork now, ahead of appropriations rather than after them. Agencies that use that interval well tend to reach a defensible RFP faster once funding clears.

Turning Program Data Into a Business Case 

Site selection and evaluation methods are engineering decisions, but the program gets approved or cancelled in a budget meeting. Council members and finance staff evaluating a new program want to see safety impact expressed in terms they can defend publicly, not just crash counts. 

USDOT’s current guidance puts the value of preventing a single traffic fatality at $14.2 million for analyses using a 2025 base year — the Value of a Statistical Life figure federal agencies use in benefit-cost analysis. It reframes crash reduction as avoided cost rather than an abstract safety benefit, and gives public works directors and council members a common unit for weighing a corridor’s enforcement investment against its avoided-cost potential.

Agencies moving through the procurement stage can start that conversion early by working from projected corridor risk data — before a system is deployed — rather than waiting for a first-year crash report to make the case retroactively. Elovate’s Automated Speed Enforcement Impact Calculator does this using IIHS speed-reduction findings, Nilsson’s Power Model, and NHTSA comprehensive crash cost values. 

Why This Matters 

Programs designed around predictive, rolling data review — rather than annual look-backs — are easier to defend to a council, easier to evaluate at renewal, and easier to adjust as a corridor’s risk profile changes. As more states authorize automated work zone, school zone, and corridor speed enforcement, the agencies that build data governance and reporting expectations into the RFP stage will be the ones with the cleanest procurement record and the strongest case for renewal three years in. 

Frequently Asked Questions