Before any research, I started with the frustration itself. The jeepney system is not broken - it moves millions of people every day at a fraction of the cost of ride-hailing. But for anyone who did not grow up with it, the system is completely opaque.

Routes are painted on windshields in shorthand locals read instantly. Stops are not marked. Fares are not posted. The entire system runs on insider knowledge that no app had ever tried to translate.

My initial hypothesis was simple: the problem is not the transport system - it is the legibility gap. Tourists and new residents default to Grab not because jeepneys are worse, but because the system feels unreadable and unpredictable. Fixing that did not require new buses or better roads. It required better information flow.

What are the daily challenges participants face during Manila commutes?

Watching commuters in real time revealed the workaround systems they had built to survive the information gap. One local office worker checked Google Maps, a Facebook commuter group, and asked a station guard before boarding a single train. A tourist screenshotted his route before leaving the hotel because he expected to lose signal mid-journey.

A student knew every stop on his route from memory but had no reliable way to share that knowledge beyond talking to someone in person. These were not minor inconveniences. They were full systems people had built to manage a trip that, with better information flow, should have taken only a few taps.

Existing tools each solved one part of the problem while ignoring the rest. Sakay.ph handled jeepney routing better than anyone else but lacked live conditions and driver visibility. Google Maps offered global scale but treated jeepneys as data gaps. Grab won through predictability, but separated riders from public transport entirely and at a much higher cost.

The gap was not a missing feature. It was a missing layer: no app consolidated routing, real-time conditions, safety context, and fare clarity across all modes in one place.

I developed two personas that represented the most distinct user groups and the most different relationships with Manila's transport system.

Mapping Maria's journey from Quezon City to Makati revealed the exact moments where confidence collapsed. She could not see all available transport options in one place, could not confirm whether she was at the right jeepney stop, and eventually defaulted to Grab because uncertainty felt more expensive than money.

For Stefan, the critical failure was a single missed stop. No alert told him when to get off. He overshot his destination, had to walk back, lost signal, and switched to Grab. That one failed trip ended his willingness to try jeepneys for the rest of his stay.

The journey maps made two things clear: confidence does not require perfect data, and a single failure can permanently destroy trust.

The most striking contradiction was behavioral. In interviews, nearly everyone said they preferred jeepneys and buses and wanted to stop overspending on Grab. In reality, the moment any uncertainty appeared, they opened Grab.

Every commuter had already built a mental Plan B: if this gets confusing, I will just take a taxi. That fallback is always one tap away, and the moment any uncertainty appears, the fallback wins.

1. Uncertainty and unpredictability. ETAs shifted mid-journey without explanation, and commuters added 30 to 45 minute buffers just to feel safe.

2. Information fragmentation. Users juggled 2 to 4 apps for a single trip, with no tool covering multimodal travel end to end.

3. Jeepney and bus opacity. Routes, stops, and fares were either unavailable digitally or impossible for outsiders to interpret.

4. No real-time conditions data. There was no structured crowd density, flood routing, or live tracking, so people relied on Facebook groups and social feeds.

5. Safety anxiety. Safety shaped route choices more than price or time, yet no existing app structured or surfaced that information clearly.

6. Weather and flooding. Rain could completely transform the city's transport landscape without proactive warnings or rerouting built in.

The sitemap went through a significant revision after early testing. An initial version included separate 'My Tickets' sections in account navigation. In testing, participants could not locate their purchased ticket after buying it, and several assumed digital payment was available on jeepney legs because no distinction was made.

Both sections were removed entirely. Payment now surfaces only at the relevant step in the route, and a purchased QR ticket appears immediately after payment with no additional navigation required.

For MVP, I prioritized the parts of the journey that reduce uncertainty fastest: one place to compare routes, clearer jeepney guidance, approximate fares, and real-time condition signals that help people decide whether a route is still worth taking.

I did not prioritize features that depended on infrastructure the city did not realistically have, such as precise live jeepney tracking or a unified payment flow across every transport mode. Those ideas sound valuable, but they would have made the product look more capable than it could actually be.

The goal of the first version was not to solve every transport problem in Manila. It was to help people make better decisions with the information that could be delivered credibly from day one.

Adding a driver-side system would make real-time jeepney tracking possible for commuters while helping drivers simplify daily decisions by knowing which route to choose based on demand.

Introducing the system that supports commuters from the beginning to the end of their journey, and provides full reliable real-time information: from planning to arrival will allow them to travel with more confidence and make more informed decision on every step of the journey.

Before turning the research into features, I looked at adjacent products to understand whether drivers already had access to the signals they needed. The goal was to see which tools supported real demand decisions, which ones only solved part of the problem, and whether any platform actually served traditional transport drivers in Manila.

Three field observations made the research tangible. In the morning, Romeo, a jeepney owner-operator on the Espana-Lawton route, arrived early but still had no reliable way to judge demand. He watched other drivers leave, made small route decisions based on conversation and instinct, left half-empty, manually tracked cash in a notebook, and ended the shift unsure whether he had made the right calls or simply wasted fuel.

In the afternoon, Manny and Berto, a bus driver and conductor pair, were dealing with a different version of the same problem. Traffic stretched a normal trip far beyond expectation, passenger counts were tracked manually, and decisions about stopping or passing riders were made reactively. The conductor could feel crowding building but had no structured way to communicate it forward, which meant the route kept operating on word-of-mouth instead of usable data.

By evening, Carlos, a UV driver, was moving between popular pickup spots almost entirely on gut feeling. He compared notes with other drivers, relied on observation rather than any platform, and accepted suboptimal rides because there was no visibility into which areas were actually worth waiting in. Across all three sessions, the pattern was consistent: demand decisions were made in an information vacuum, and that uncertainty translated directly into financial risk.

Drivers have almost no real-time visibility into passenger demand. They operate somewhere between pure guesswork and delayed tribal knowledge, while Grab drivers see live demand heatmaps. Traditional transport drivers compete blind.

Experienced drivers have built sophisticated mental models and informal networks to compensate, but that knowledge takes years to acquire, does not scale to new drivers, is not accessible to passengers, and breaks down during disruptions.

Fixed daily costs such as boundary payments and fuel turn every quiet hour into a financial risk. Without knowing whether demand will improve, decisions about waiting, leaving, or repositioning become binary and anxiety-driven.

Jeepneys, buses, and UVs are systematically invisible in the digital tools passengers use to navigate Manila. Drivers lose passengers not because the vehicle is worse, but because the system is illegible and hard to discover.

There is no mechanism to match supply and demand between stops. Drivers, conductors, stops, and passengers all operate in isolation, which leads to empty return trips in one direction while people wait in the other.

Disruptions such as events, weather, protests, and road closures are learned too late, usually through passengers or direct observation. By the time drivers know, the opportunity to prepare or reposition has already passed.

Drivers make both tactical decisions, like whether to continue a shift, and strategic ones, like which route or vehicle to operate, without comparative data. High-stakes choices with direct earnings impact are made blind.

Key moments from research translated directly into product decisions.

Extremely low tech literacy. The driver interface cannot require more than 2-3 taps to reach any critical information.

One-handed use while parked. Drivers often check phones with the engine running while eating or drinking, so every interaction has to be fast and readable at a glance.

Fixed route franchise structure. Most jeepney drivers cannot freely reroute, so the system can support micro-adjustments and stop prioritization, but not full route changes outside the legal corridor.

Drivers already improvise sophisticated workarounds despite zero system support. Miguel, for example, combines four different data sources, which showed me how much latent capability is being wasted.

Drivers know exactly why they lose to Grab. They understand that invisibility in digital systems is the issue, but they feel powerless to change it.

The PHP 1,800 daily boundary forces blind decision-making under financial pressure. It is not just a cost structure; it shapes behavior.

Drivers want supply visibility almost as much as demand visibility. Knowing where other vehicles are clustered matters because it changes whether waiting is smart or wasteful.

Event disruptions are learned reactively from passengers, not proactively through alerts, which means opportunities are lost before drivers can act.

No visibility into where passengers are waiting, so route and stop decisions are forced into guesswork.

No way to track what makes money, which means drivers cannot learn from past shifts or optimize behavior.

Passengers cannot find jeepneys digitally and default to Grab before the first interaction even happens.

Drivers learn about flooding, events, and road closures from passengers instead of alerts.

Conductors cannot tell stops ahead that a bus is already full.

Drivers do not know where other drivers are, which leads to inefficient clustering.

Vehicles return empty while passengers wait in the opposite direction.

Drivers operating traditional transport in Manila compete blind while ride-hailing platforms dominate through information asymmetry.

Without visibility into passenger demand, real-time conditions, or supply dynamics, drivers make navigational and strategic decisions based on guessing, tribal knowledge, and incomplete information. The result is wasted fuel, missed earnings, and an inability to compete with platforms like Grab that already provide demand signals.

This is not a missing technology problem. It is a missing decision-support layer for the drivers who need it most.

Mapping Romeo's full day showed that the hardest decisions were not operational in the usual sense. They were moments where financial pressure replaced information and forced him to act on instinct.

Designing the driver-side experience meant accepting that the system could not promise perfect live demand. Only a minority of riders generate trackable real-time data, so the forecasts rely heavily on time-of-day patterns and become directionally useful rather than fully live. That tradeoff felt acceptable because even imperfect guidance is better than the total guesswork drivers operate with now.

Another tradeoff was choosing financial honesty over emotional comfort. Showing a driver exactly how far they are from boundary can feel uncomfortable in the moment, but hiding that number makes the cost of continuing invisible and leads to worse decisions. I also chose simplicity over depth. Richer analytics and comparison tools were intentionally cut because tech literacy constraints meant any critical action had to stay within two or three taps.

The biggest open question is how much confidence the product could build over time, not just in a first-use test. I validated that clearer routing, stop alerts, and more honest transport information made the experience easier to understand, but I did not test whether that would be strong enough to shift real commuting habits away from Grab over the long term.

The driver-side concept also remains only partially validated. The logic of demand guidance and earnings visibility tested well directionally, but it still depends on incomplete data, uneven driver adoption, and operating conditions that can change faster than the system can reliably reflect.

This product does not fix Manila's transport infrastructure. It does not create reliable GPS where none exists, standardize jeepney stops, remove traffic unpredictability, or solve the deeper operational issues that make the system hard to coordinate at city scale.

It also does not eliminate cash-based friction or guarantee perfect real-time accuracy. The goal of Tara is not to make the network flawless. It is to make an already functioning but opaque system more readable, less intimidating, and easier to trust.

The first requirement would be a phased rollout with a narrow, realistic scope: a limited set of routes, clear data confidence labels, and commuter features that remain useful even when live information is incomplete. The product would need to be honest from day one about what is estimated, what is community-reported, and what is truly live.

For the driver-side system to work at all, adoption would need to come from both individual drivers and transport operators, supported by onboarding that respects low-tech workflows. Shipping this credibly would also require partnerships for route data, a lightweight reporting system for disruptions, and repeated field testing to understand whether the product improves decision-making in real conditions rather than only in prototype use.

I would research technical limitations and constraints more deeply before creating the prototype. The absence of fixed pricing was discovered by me only during user testing and was not stress-tested before.

The use of the app was tested by users in the short term, but there is still a need to observe how the app affects habits and decision-making in the long term.

User control and openness to try something new can reinforce each other: giving users options can motivate them to make choices they usually would not make.

Giving useful information is more important than giving precise information.

1. Confidence is a design problem, not an infrastructure problem. The jeepney system is not broken. It is opaque. Before redesigning systems, the better question is whether existing systems can be made more readable.

2. User control and openness reinforce each other. Giving users the ability to filter transport modes upfront did not reduce their willingness to try jeepneys - it increased it. When users felt they had agency, they became more curious, not less.

3. One bad experience can end trust permanently. Stefan missed one stop and gave up on jeepneys for the rest of his trip. The GPS stop alert mattered not because of frequency of use, but because its absence caused irreversible damage to confidence.

4. Useful information beats precise information. A historical wait estimate labeled approximately four minutes is more useful than silence. A plain-language safety label is more useful than a numeric score nobody can interpret.

5. The assumptions that feel most obvious are often the ones that need testing first. Showing all routes upfront, using a numeric safety score, and assuming digital payment could be unified across modes all felt reasonable before testing. The research that surprised me most shaped the design most.

6. Next steps would include exploring long-term behavioral tracking to understand whether the app actually shifts commuters away from Grab over time. A rewards system for reporting delays and disruptions could also build better community-sourced data and reduce reliance on historical patterns alone.

Manila's public transport problem is local, but the underlying problem is universal: systems designed by insiders are often illegible to outsiders. Immigrants struggle with bureaucracy. Tourists get lost. New employees cannot navigate company culture. Patients do not understand healthcare navigation.

The app demonstrates that before we redesign systems from scratch, we should ask whether we can make existing systems more readable.

Often, the system is not broken. It is just undocumented, built on local knowledge and unwritten conventions. Design's job is to externalize that knowledge so newcomers can participate without having to be insiders first.

This is why the project's real value is not the app itself. It is the proof that confidence is a design problem, not an infrastructure problem. When users understand how something works, they stop avoiding it. When they feel they have agency, they explore it. When systems are honest about constraints, they become trustworthy.