The Last MileAll the Way

1out of 100 make it
and come backSimulation

Exploration 02 · The Last Mile

What the report says

Governments have invested in digital transformation. Yet many users fail to engage, drop out along the customer journey, or don’t return after initial engagement. (based on p. 17)

SimulationIn the game, the crowd, the pass rates, and the “out of 100” counter are invented to make a mechanism visible: they can’t be compared with the report’s figures. Everything that carries a page number comes from the report, summarized by Awlers.

Infrastructure without Impact

Governments have invested heavily in digital transformation, yet many services experience low uptake and high drop-out rates, and repeat usage is elusive. These services don’t fully account for how people actually behave. (based on p. 17)

What government services often assume (p. 18)

If we provide a digital service, people will use it.
If we make it mandatory, compliance will follow.
If we explain the benefits, people will rationally engage.

But real-world behavior shows otherwise: digital services are often undermined by predictable behavioral pain points all along the user journey, from awareness to repeat usage. (based on p. 18)

Seven steps, seven ways to drop out

The report follows the user journey from awareness to repeat usage. Some users drop out at each stage, leaving only a small proportion to complete transactions and return to the service. (based on p. 6, 7 and 18–19)

Awareness and Discovery
The Empty Store
Pain point
People don’t realize a service exists, or assume it isn’t relevant to them. (p. 18)
What to do
Use clear, benefit-focused messaging; highlight peer usage. (p. 34)
Case study Finland, digital tax filing
Consideration and Motivation
The Crossroads with No Signs
Pain point
Too many options, unclear information: deferring action feels safer than making the wrong decision. (p. 18)
What to do
Send personalized nudges emphasizing time savings; offer incentives for first-time users. (p. 37)
Case study Argentina, telemedicine
Access and Onboarding
The Closed Gate
Pain point
Too much effort, too soon: users drop out before seeing any benefit. (p. 7 and 18)
What to do
Minimize requirements; start online, finish in person. (p. 41)
Case study Moldova, criminal records
Navigation and Transaction
The Lost Path
Pain point
Small, cumulative barriers: unclear instructions, unexpected requirements. (p. 18)
What to do
Simplify interfaces, use progress indicators, enable data sharing to pre-fill forms. (p. 48)
Case study South Korea, consistent design
Payment and Finalization
The Last-Minute Wobble
Pain point
At the point of payment or submission, trust falters. (p. 18)
What to do
Integrate secure payment systems, provide instant confirmations, use visible trust cues. (p. 54)
Case study Mexico, COVID‑19 apps
Confirmation and Receipt
The Panic Loop
Pain point
Unsure it worked, users repeat steps or contact support. (p. 18)
What to do
Send automated confirmations; provide dashboards for status tracking. (p. 57)
Case study Philippines, the 8888 hotline
Follow‑up, Reuse, Maintenance
The One-Time Wonder
Pain point
Without follow-up, first-time users become last-time users. (p. 7 and 19)
What to do
Store user data for easier repeat use; send reminders for follow-up tasks. (p. 59)
Case study Brazil, bond reinvestment

Step names are the report’s own.

Running throughout the user journey is the trust gap: without signals of competence, care, and transparency, small doubts can become big barriers. (based on p. 19)

A diagnosis for every barrier

The COM-B Model, at the heart of the report’s method: for a behavior to occur, the user must have the capability (skills and knowledge), the opportunity (external factors and environment), and the motivation (internal drivers and emotions) to carry it out. (based on p. 24–25)

Seven steps, seven case studies

One case per step, from the report: four experiments against a control group (one of them a survey), an A/B test, an observation with no comparison, and a reform described without figures. Each comes with its caveat. During the game, each one appears in the night, right after the step it sheds light on.

According to the World Bank · FinlandIn the game: Step 1: The Empty Store

Show that others already do it

1 icon = 1 taxpayer in 1008 more file online

+8percentage points

The Finnish Tax Administration tested a social norm message: “A majority of people in your age group already file online.” This localized peer reference increased adoption of digital tax filing by 8 percentage points over the control group. (p. 93)

Notes 71 and 72 on this page appear to be swapped: the Finnish case’s note (72) points to an Indian smart cities award contest, while the one before it (71) points to Hyytinen, Tuimala and Hammar (2022), “Enhancing the adoption of digital public services: Evidence from a large-scale field experiment.” We could not check the figure in the article itself, so it remains attributed to the report.

According to the World Bank · ArgentinaIn the game: Step 2: The Crossroads with No Signs

Nudging the first click

1 icon = 1 patient in 1006 more try telemedicine

+6percentage points

Telemedicine is an experience good: its value is hard to assess until users have tried it. In a randomized field experiment, about 2,000 households with no prior telemedicine usage received periodic emails explaining the availability and benefits of the service. Patients in the treatment group were six percentage points more likely to use telemedicine at least once. (p. 40)

That initial use led to a lasting shift: eight months later, the treatment group had six times as many virtual consultations as the control group. The results suggest that prompting first use through low-cost informational nudges can overcome inertia and unlock sustained engagement. (based on p. 40)

Measured use, in a randomized trial. The report also cites 12 points for those who opened an email: a self-selected group, not to be compared with the control group.

According to the World Bank · MoldovaIn the game: Step 3: The Closed Gate

Check identity at pickup, not at sign-up

Online applicationno accountPickup or deliveryidentity checked

In the early 2010s, Moldova wanted to put criminal record certificates online, but lacked a national digital ID system or a mechanism for secure online payments. It did not require digital ID or user log-in: identity was checked only at the point of pickup or delivery of the certificate, and a public verification process allowed easy verification of records. (p. 44)

The report sees a foot-in-the-door moment here: an easy first action that increases follow-through. Reducing frictions at sign-up boosts engagement. (based on p. 44–45)

The report’s executive summary (p. 9) gives a daily volume of checks that is hard to reconcile with p. 44 and is unsourced: we don’t use it.

According to the World Bank · South KoreaIn the game: Step 4: The Lost Path

One look and feel across government sites

Average session time for search (A/B test)

28seconds per search

Despite South Korea’s recognized leadership in digital government, many citizens experienced frustration and confusion: with no uniform standards across agencies, user experiences were fragmented. For some services, up to 83% of comments were negative. (p. 51)

In 2024, the Ministry of the Interior and Safety and the National Information Society Agency (NIA) developed UI/UX Guidelines for Digital Government Services, in January 2025, the Ministry launched the Korea Design System (KRDS), a pan-government design system. At the Ministry of Employment and Labor, whose content was scattered across 48 labor office websites, a consistent layout and search filters cut average session time for search from 51 to 28 seconds, according to A/B testing. (p. 51–52)

Usability testing on pilot websites showed an average satisfaction increase of 13.2 percent. (p. 53)

The report says drop-off rates decreased and trust rose, without figures. Its executive summary (p. 9) cites a 44% reduction in average session time; we keep the seconds measured on p. 52.

According to the World Bank · MexicoIn the game: Step 5: The Last-Minute Wobble

When reassurance backfires

1 icon = 1 adult in 1004 fewer say they’re willing

−4percentage points

During the COVID-19 pandemic, a survey experiment of over 23,000 Mexican adults tested different promotional messages for diagnostic and contact-tracing apps. The most common approach, highlighting government efforts to protect user data, reduced willingness to adopt them: by 4 percentage points for the diagnostic app, 3 for the tracing app. (p. 56)

Rather than providing reassurance, vague or defensive privacy messaging appeared to trigger suspicion. The report suggests, tentatively and without figures, that subtle cues, like logos of trusted online security brands, may build trust better than explicit reassurance. (based on p. 56)

Stated willingness in a survey, not measured use.

According to the World Bank · PhilippinesIn the game: Step 6: The Panic Loop

Fast feedback on every complaint

32minutes

The 8888 Citizens’ Complaint Hotline refers complaints to the appropriate agency within 32 minutes, with a required response within 72 hours. One year after launch, it was fielding over 1,000 complaints daily. (p. 95)

The report cites it as an example of fast feedback: immediate confirmation and initial processing that highlight the time saved. It also cites its confirmation messages with reference numbers. (based on p. 55 and 95)

Response times and a volume, with no control group: nothing here measures the effect of speed. The executive summary (p. 9) says 30 minutes; we follow p. 95. The volume comes from a 2017 news article (note 75).

According to the World Bank · BrazilIn the game: Step 7: The One-Time Wonder

A reminder, a simpler choice

On Tesouro Direto, the government-run platform for purchasing treasury bonds, roughly 80% of investors did not reinvest when their bonds matured: many forgot or procrastinated, others were overwhelmed by choosing a new bond among dozens. (p. 62)

The National Treasury and B3, the stock exchange, tested three email reminders before bonds matured. Compared with the control group (no email): (p. 62–64)

×5scheduled reinvestments, with the Forced Choice email: reinvest or stop earning
×4.3reinvestment rate, with the Active Choice email: a simplified menu of tailored options

(p. 64)

The Attention Only reminder led to a moderate increase but was substantially less effective. The Forced Choice email had the highest open and click-through rates, but once they clicked through, users who received the Active Choice email were more likely to complete the reinvestment. (based on p. 64)

Ratios as the report gives them, from the case study published by Irrational Labs (note 19). The control group was much smaller than each of the other three.

What the report concludes

Digital services often fail not because they’re poorly built, but because they’re out of sync with how people actually think, feel, and act. (p. 79)

These barriers are systemic and cross-cutting: piecemeal fixes won’t solve them. (based on p. 79)
Modest changes in design, communication, or delivery can significantly reduce drop-out and boost adoption. (based on p. 80)
The report recommends integrating behavioral science into digital service teams, mapping the behavioral user journey, investing in iterative testing and learning, and adapting KPIs to the key points of the journey. (based on p. 98)

Game rules

Rules made up for the game, not measurements. The crowd, the pass rates, and the “out of 100” counter describe no real service.
Each step lets part of the crowd through, and the counter multiplies these shares: one neglected step thins out the rest of the path. The report doesn’t state this rule; it describes drop-out at each stage (p. 6) and says piecemeal fixes won’t solve systemic frictions (p. 79).
The report’s figures read differently: in percentage points, multiples, or seconds, most often against a control group. That’s why no game number is ever placed next to a report figure.
The seven moves borrow ideas from the report (a store sign for the social norm, the benefit shown at the crossroads, a gate moved for the ID check, a guided trail, a subtle seal for trust, an idea the report puts forward tentatively and without figures, an instant confirmation, a timely reminder) without reproducing their conditions.

Where do your users drop out?

Awlers maps your users’ journey, uses AI to help pinpoint the steps where they drop out, and helps you test small fixes.

Let’s talk about your user journey