Resources/Collaborative Trip Planning
Collaborative Trip Planning

Neighbors who pool budgets cut twice the emissions.

The Commutrics team published a peer-reviewed system that lets nearby businesses pool incentive budgets, expand carpool matches, and personalize the dollar offered to each commuter, validated on 253 Denver employees and shown to roughly double emissions reductions versus going alone.

Case study
253 employees
Businesses
3 neighboring
Logit model
6,616 trips
Published
IJST · 2026
Headline findings

Six numbers that change the math on commute programs.

The case study showed that what's feasible alone is very different from what's feasible together. These are the most actionable numbers in the article.

Emissions
9%vs 5%

Collaborative cuts at the same budget

At $2 per employee per day, pooling budgets across three businesses cut emissions by 9%, versus 5% when each business optimized on its own.

Match expansion
+15×

Carpool match pool, expanded

Collaboration grew the feasible carpool match space from roughly 2M options across the three businesses to over 32M, more than fifteen times bigger.

Incentive design
$8.51

The minimum dollar that flips a commuter

For one example commuter, just $8.51 per day was enough to make carpooling more attractive than driving alone, a fraction of what walking or biking would have cost.

Rider savings
25%

Cheaper commute for carpool riders

Riders pay only 75% of their old commuting cost to the driver, pocketing a 25% saving while skipping the parking hunt. Drivers earn that plus an incentive.

Mode shift
64%→ 48%

Drive-alone share, reduced

In the case study, drive-alone fell from 64% of trips to 48% as incentives scaled up, almost entirely through carpool adoption, not transit or active modes.

Mode shift
+2.5×

Carpool's share, multiplied

As incentives scaled, carpooling climbed from 2% of trips to roughly 5%, the workhorse behind nearly all of the drive-alone reduction the system delivered.

How it works

Five models, working as one.

The system runs a streamlined pipeline from raw commuter input to personalized, emissions-minimizing recommendations, each stage feeds the next.

01

Input data

Web and mobile surveys capture home, work, schedule, vehicle, parking, and mode preferences.

02

Travel attributes

Time, cost, distance, fuel, calories, and CO₂/NOₓ/VOC computed per mode via Google Maps and Here360.

03

Logit discrete choice

A Random Utility Model calibrated on 6,616 Denver trips estimates the probability each commuter picks each mode.

04

Monetary incentives

Computes the minimum dollar offer that makes a sustainable mode 1.5× more likely than driving alone.

05

Multi-objective optimization

Allocates incentives across all commuters to minimize total emissions within the pooled budget.

Each stage informs the next, end to end, from raw survey to personalized weekly plan.
The case study

253 employees. Three Denver businesses. One pooled budget.

Three neighboring downtown employers, staffed at 102, 92, and 59 employees, opted into the pilot in partnership with Downtown Denver Partnership. Each provided home and work addresses, schedules, and existing commute behavior for every participant.

The starting point was overwhelmingly car-dependent: 64% of trips were drive-alone, and 81% of commuters had transit accessibility scores below 50. Bike scores told a different story: roughly a third of commuters lived along bikeable routes, suggesting real headroom for shift.

253
Total employees across the three businesses
64%
Drove alone before incentives were introduced
51
Pareto-optimal incentive plans generated
~1 min
Average system runtime per employee
Existing mode share · case study
Where they started.
Drive alone
64.0%
Transit
26.1%
Walk
7.9%
Bike
2.0%
Collaboration vs going alone

The same $2 a day, twice the impact.

At a fixed daily incentive of $2 per employee, every business cut more emissions when they pooled with their neighbors than when they optimized on their own.

01
Business one
102 employees
Solo planning6%
Collaborative13%
02
Business two
92 employees
Solo planning2.1%
Collaborative4.3%
03
Business three
59 employees
Solo planning6.2%
Collaborative13.5%
Solo · all three combined
5%
emissions reduction · $2 / employee / day
Collaborative · pooled
9%
emissions reduction · same daily budget
Personalized incentives

One commuter. Five modes. Five different price tags.

For each commuter, the model computes the minimum incentive that would make each mode 1.5× more likely than driving alone. The cheapest sustainable option is usually carpool, by a lot.

Promoted mode
Min. incentive
Probability
Relative cost
Drive alone Reference
baseline
Carpool Cheapest shift
$8.51
56.8%
2%
Transit + bike
$36.30
44.7%
9%
Transit + walk
$67.60
43.4%
16%
Bike
$78.60
43.0%
19%
Walk
$420
43.2%
100%
Read it this way: for this particular commuter, $8.51 per day buys a real shift to carpool, but spending $420 a day to push a walk that takes 243 minutes one way is not a serious option. The optimizer never recommends it; it routes those dollars to commuters where the same budget moves more emissions. That's the difference between blanket subsidies and a targeted incentive plan.
Built for thousands

Scales to 3,000 commuters in about two hours.

The system's most expensive step is preprocessing, calling external mapping APIs to compute travel attributes for every commuter-mode combination. Without optimization, that step explodes past 750 minutes at 3,000 commuters.

Two engineering choices fix this: (1) prune infeasible options before they ever reach the solver, and (2) parallelize API calls across cloud functions. The optimizer itself stays fast and flat across all population sizes, the heavy lifting is the data prep.

No preprocessing, every option computed~770 min
With preprocessing, infeasible pruned first~380 min
Parallelized preprocessing, cloud functions~120 min
Solver alone, the optimization step itselfflat
Computational time vs commuters
End-to-end runtime, in minutes, across four configurations
8006004002000Minutes05001,0001,5002,0002,5003,000Number of employees770 min380 min120 min
What this means for your program

Five things to do this quarter.

The evidence underneath the Commutrics platform. Use it to design any TDM program, yours or ours.

01

Find your neighbors.

The single biggest finding in this article is that nearby employers should be planning together. A 0.5-mile radius is enough to make pooled carpool matching dominate solo planning. Map every business within walking distance of your office or campus and ask which of them already buys some form of commute benefit, that's your starter coalition.

02

Pool the budgets before spending them.

Three businesses each spending $2 per employee on isolated programs got 5% emissions reductions. The same total dollars run as a single pooled program got 9%. The dollars didn't change; the matching pool did. Where local law and accounting allow, a shared TDM budget across co-located employers is the highest-leverage move on this list.

03

Personalize the dollar amount, not just the perk.

Most TDM programs offer the same benefit to everyone, $X for any carpool trip. The optimization model shows that flipping different commuters costs wildly different amounts: $8.51 here, $420 there. Spending those dollars equally is wasteful. Use a model to find the commuters whose mode shift costs least per kg of CO₂ avoided, and route the budget there first.

04

Lead with carpool, not heroic mode shifts.

Across 51 Pareto-optimal solutions, transit, walk, and bike shares stayed nearly flat. Drive-alone fell almost entirely because carpool rose. For most office commuters in car-shaped cities, carpool is the only realistic alternative, it preserves door-to-door speed, doesn't require fitness or weather, and the model can find matches cheaper than $10 a day per commuter.

05

Run incentives on a weekly cycle, not a forever cycle.

The system is built around asynchronous, weekly batch planning, not real-time recomputation. That's a deliberate design choice: it scales, it gives commuters predictable plans, and it lets the optimizer recalibrate as people update their home, schedule, or vehicle. Programs that hard-code one benefit menu for a year lose the ability to reallocate when their data tells them to.

Peer-reviewed · IJST 2026

Read the full article, or cite it.

The Commutrics team published the complete study, system architecture, all five model formulations, the logit calibration results, the full case study, and the scalability analysis, in the International Journal of Sustainable Transportation, Taylor & Francis.

Suggested citation
Monghasemi, S., and Abdallah, M. (2026)."Collaborative and incentivized trip planning system to minimize transportation emissions for business commuting."International Journal of Sustainable Transportation, 20(2), 184–208. DOI: 10.1080/15568318.2025.2578219.
University of Colorado Denver, Department of Civil Engineering. Funded by the Mountain-Plains Consortium. Case study data collected in partnership with Downtown Denver Partnership.

The model behind the platform, working for your program.

Every component in this article, the travel attributes engine, the logit choice model, the personalized incentives, and the multi-employer optimizer, is built into Commutrics. Bring your neighbors. We'll handle the math.