One label, many working lives

Platform work is not one job or one experience. Delivery apps are a familiar image of the gig economy, but the ABS survey captures several types of services and very different patterns of participation. The story moves from what people do, through who participates and how regularly, to what they want from work and what protections they report. Each section adds a different dimension to the same national picture.

The ABS estimates about 179,000 people engaged in digital platform work during the previous 12 months in 2024–25. This is a national survey estimate for all platform workers, not DoorDash drivers. Error bars are approximate, and the report does not claim statistically significant changes between years.

1. What work is visible—and what is less visible?

The six service categories are not mutually exclusive. Delivery has the highest point estimate in 2024–25, but we cannot establish a clear ranking over transport from these figures alone. Use the dropdown to switch survey periods; the intervals and caution labels update with the data.

Chart key: Teal/blue dots = estimates with RSE below 25%; gold dots = estimates with RSE 25% or higher (interpret cautiously). Horizontal whiskers = approximate 95% uncertainty intervals.

Text alternative: In 2024–25, delivery services account for 50% and personal transport for 39% of platform workers. The approximate uncertainty intervals overlap, so the difference is not established. Six service categories are shown with horizontal intervals; respondents may report multiple categories. Gold markers flag estimates requiring caution.

2. Who is represented?

Age, sex and country of birth are separate marginal distributions, not combinations of the same people. They should not be interpreted as evidence about, for example, the age distribution of workers born in India. The selector changes the demographic dimension rather than mixing incompatible denominators.

Chart key: Teal/blue dots = estimates with RSE below 25%; gold dots = estimates with RSE 25% or higher (interpret cautiously). Horizontal whiskers = approximate 95% uncertainty intervals.

Text alternative: The dropdown displays age (15-34 years: 57%; 35-54 years: 36%; 55 years and over: 8%), sex (Males: 71%; Females: 29%), or country of birth. The country-of-birth categories have estimates ranging from 5% to 37%, with caution required for some small groups. These are separate distributions, not joint age-by-sex or birthplace-by-age observations.

3. How regular is the work?

A single annual headcount can hide different patterns of participation. The ABS also reports frequency and duration. These are separate questions, so their percentages should not be added together across the two panels.

Chart key: Teal/blue dots = estimates with RSE below 25%; gold dots = estimates with RSE 25% or higher (interpret cautiously). Horizontal whiskers = approximate 95% uncertainty intervals.

Text alternative: Frequency of platform work: At least once per week: 65%; At least once per month: 13%; At least once every 3 months: 8%; Once every 6 months or less: 14%. Duration of platform work: Less than 1 year: 51%; 1 year to less than 2 years: 19%; 2 years to less than 5 years: 17%; 5 years or more: 13%. The two panels are separate distributions and cannot be combined into a joint frequency-by-duration breakdown. Error bars are approximate 95% intervals.

4. A main job is not the same as a preferred arrangement

The distinction between current status and preference matters. The survey separately asks whether platform work is a main job and what work arrangement people prefer. These measures cannot be used to infer that any particular respondent is dissatisfied.

Chart key: Teal/blue dots = estimates with RSE below 25%; gold dots = estimates with RSE 25% or higher (interpret cautiously). Horizontal whiskers = approximate 95% uncertainty intervals.

Short labels are used for readability; hover over a marker for the full ABS category and uncertainty details.

Text alternative: Current main-job status: Main job: 39%; Not main job: 61%. Preferred work arrangements: Digital platform work through apps or websites only: 24%; Wage-based employment only: 48%; Combination of digital platform work and wage-based work: 25%; Some other kind of work: 2%. These separate questions cannot be linked at the respondent level. The published table does not explain why the 2023–24 main-job categories do not total 100%.

5. What do safety responses reveal—and not reveal?

Insurance and feeling unsafe are distinct survey questions for people who had undertaken platform work in the previous four weeks (about 99,000 estimated workers in 2024–25), not the full 12-month population. The figures cannot establish whether lack of insurance causes workers to feel unsafe.

Chart key: Teal/blue dots = estimates with RSE below 25%; gold dots = estimates with RSE 25% or higher (interpret cautiously). Horizontal whiskers = approximate 95% uncertainty intervals.

Text alternative: Personal accident insurance: Had personal insurance: 39%; Did not have personal insurance: 61%. Feeling unsafe: Felt unsafe: 16%; Did not feel unsafe: 84%. These estimates describe recent platform workers, not all workers in the 12-month population. The questions cannot be cross-tabulated from the published aggregates; intervals are approximate.

Short labels are used for readability; hover over a marker for the full ABS category and uncertainty details.

What the five views reveal

The same label—platform worker—covers different services, different degrees of commitment and different preferences about employment. The national estimate alone cannot show that variation. The five views suggest that a useful discussion of platform work needs to ask what kind of work, how sustained it is, what arrangement is preferred, and what protection is reported—not simply how many people use an app to earn money.

This is a story about heterogeneity, not a trend. The ABS workbook does not link individual answers across questions, so we cannot claim, for example, that delivery workers are the people without insurance, or that those who prefer wage employment are currently unhappy. Nor do overlapping survey estimates establish changes over time.

Data preparation and uncertainty notes

  1. Source: One original public ABS Excel file, 63360_Table09.xlsx, containing Table 9a (characteristics) and Table 9b (digital platform work). The workbook remains unmodified in data/.
  2. Import: readxl::read_excel() reads explicit worksheet ranges. Each chart extracts documented Excel rows and columns: Table 9a C/D for 2024–25 share/RSE and I/J for 2023–24; Table 9b C/D and F/G respectively. The displayed estimates are loaded from the file at Knit time.
  3. Cleaning: Trim category whitespace; coerce reported numeric values; check for missing cells and implausible percentage/RSE values. No categories are imputed or aggregated.
  4. Uncertainty: Approximate standard error = estimate × RSE / 100. Approximate 95% interval = estimate ± 1.96 × SE, clipped to 0–100%. These are illustrative normal approximations, not ABS-published confidence intervals. The workbook does not provide the covariance required for formal pairwise tests. In particular, the upper interval for “Did not feel unsafe” is clipped at 100%; this is a mathematical boundary, not a precise upper confidence limit.
  5. Quality: ABS marks RSE 25–50% estimates as use with caution and RSE over 50% as unreliable. Hover labels report RSE; gold markers flag estimates with RSE ≥25%. Intervals are visible without hovering.
  6. Population: Tables 9a and 9b report different questions and sometimes different reference populations. The insurance/safety questions concern the last four weeks; other charts describe people who worked on platforms in the last 12 months. Birthplace, age, gender, frequency, preferences and work type are marginal distributions, not individual-level cross-tabulations.
  7. Precision and interpretation: Survey figures are rounded in the prose. The two periods are displayed descriptively, not as evidence of a trend. Multiple work types can be reported; their percentages need not sum to 100%. The unexplained 2023–24 main-job percentage discrepancy is not attributed to an invented cause.
  8. Category order and scales: Categorical axes use explicit, reversed category arrays so horizontal charts display the workbook order from top to bottom. Quantitative axis limits are based on the largest approximate interval in each panel; axes may therefore differ between panels.
  9. Interaction and accessibility: Charts provide hover details and zoom; Charts 1 and 2 add period/dimension selectors. Frequency and duration are shown as two compact, independently scaled widgets. Text alternatives follow each plot. Colour is reinforced by labels, intervals and explicit caution descriptions.

Reference

Australian Bureau of Statistics. (2025, December 12). Working arrangements, August 2025 (Catalogue No. 6336.0, Tables 9a and 9b) [Data set]. https://www.abs.gov.au/statistics/labour/earnings-and-working-conditions/working-arrangements/aug-2025

OpenAI. (2026). ChatGPT [Large language model]. https://chat.openai.com

AI assistance disclosure

The story idea, choice of data, and the R code and written text in this report are my own. I used generative AI tools only to get suggestions for fixing problems in my work. ChatGPT (OpenAI, 2026) and Claude (Anthropic, 2026) were used to suggest fixes for chart layout, plotly settings and syntax errors. I reviewed each suggestion, applied the changes myself, re-ran the report and checked the results against the ABS workbook. I did not use AI to choose the story, generate the charts, or write the report text.