Role note: This work was led by Baofeng Dong while at TriMet (pre-LMAi Labs), and was presented to the TRB AP050 committee in January 2018. It is included here because it illustrates the kind of innovation-tech program LMAi Labs now helps small and mid-sized transit agencies ship.
The Challenge
Public transit agencies are required to run on-board surveys for Title VI, FTA reporting, and service planning. TriMet's traditional paper-based on-board Origin-Destination (O-D) survey had a number of compounding problems:
- Slow. It typically took more than six months between data collection and usable, geocoded results — long after the data was useful for planning.
- Lossy. Roughly 20% of records had to be thrown out because handwritten addresses couldn't be geocoded.
- Expensive. A typical survey cycle of ~30,000 questionnaires meant ~$12,500 in printing and mailing, ~$63,000 in data entry, ~$30,000 in geocoding (for 51,337 addresses in one cycle), and ~$60/hour for vendor surveyors.
- Low response. Paper response rates were typically under 20%.
The agency needed a better way to capture rider O-D data — without the cost, the throw-out, or the six-month wait.
Our Approach
We pioneered a tablet-based on-board survey program built on Open Data Kit (ODK) for form design and data aggregation, integrated with an in-house Android mapping app for capturing on/off stop information. The whole program was managed centrally in IT-GIS by a dedicated Electronic Survey Developer & Analyst, with field supervisors and 25–30 college-student surveyors.
The first tablet-based on-board survey ran in 2014–15 as a pilot O-D study. The savings were significant enough that it became a dedicated program — and all of TriMet's on-board surveys are now conducted with tablets.
Two survey methods
- Short survey: Surveyors scan QR codes at front and back doors with tablets to identify boarding/alighting ridership patterns and set sampling goals for the long survey.
- Long survey: Tablets with pre-populated answers capture detailed ridership behavior — origin and destination geography, transfer rates, mode of access, and demographic information — with built-in skip logic and validation.
Key Features
Data quality, by design
- Built-in skip logic so respondents only see questions that apply to them.
- Automatic validation of values against constraints and integrity rules at entry time.
- Multi-language support by changing language settings on the device.
- Geocoded origin and destination coordinates captured directly, eliminating the address-to-geocode loss.
Real-time program management
- Real-time web dashboard streaming with map visualization.
- Real-time QA/QC and analytics on incoming responses.
- Real-time sampling-quota management by route, time of day, and segment.
- Real-time progress reporting and shift management for the surveyor team.
Hiring approach
College students at $17/hour replaced consultant surveyors at $60/hour — over two-thirds less in labor cost — while providing bilingual coverage, local-area knowledge of the transit system, and a recurring talent pipeline for adjacent customer-service projects.
Results
- ~30% cost savings vs. the paper-based survey overall.
- 92% reduction in lead time from data collection to usable results.
- 60–65% response rate with tablets, compared to under 20% with paper.
- Near-zero throw-out rate — paper surveys typically lost ~20% of records due to un-geocodable addresses.
- Eliminated ~$105K in printing, data entry, and geocoding costs for a ~30,000-survey cycle.
- The program scaled: tablet surveys are now the default for every on-board survey at TriMet, and tablets are reused for facilities inventory, on-street customer service, and other field work.
Why It Matters For Small/Mid Transit Agencies
Most small and mid-sized agencies are still stuck with the paper-survey model — or paying a vendor a premium for a black-box tablet program. The TriMet program shows that the right combination of open-source tooling, in-house GIS, and tight program management can give an agency a survey instrument it owns, runs, and can adapt to the next study without starting from scratch. That same playbook — break the vendor lock-in, own the data, ship the dashboards — is what LMAi Labs brings to AI-powered service planning and Title VI equity analysis for transit agencies today.