This interactive matrix codes studies according to their engagement with twelve elements of an individual’s experience of platform livelihoods, crossed by nine livelihood types. You can also filter by eight crosscutting factors and by country.

How to use the map

There are two views of the Map. The first view, Studies, shows the total number of studies that mention each element, by livelihood type.

The second view, Results, presents these reports with valence. The color codes circle show the direction of sentiment (green is positive, red is negative and orange is neutral) and the number within the circle shows the number of observations. Most studies reported more than one experience element. and many studies looked at more than one kind of livelihood. Thus, the number of claims in the results view may exceed the number of studies.

In both views, you can filter by different variables to gain a more precise picture of how these experiences and types link to various digital development issues. You can click on any circle to access further information on the studies.


This map was last updated in October 2020.

We are tracking several studies that are in progress, or have been identified but not yet coded. These will be folded into the next version of the map.

If you have questions on the map, or would like to discuss research priorities or know of relevant platform livelihood studies that you would like us to include, please contact us at

The Knowledge Map 1.0 was developed in 2020 by Caribou Digital in collaboration with Qhala, with the support of the Mastercard Foundation.

We were animated by the question of how people in the global south experienced platform livelihoods, and we drew on anchor studies by Heeks (2017) and Zollmann and Wanjala (2020) to inform our design.

Study identification and selection

Given the interdisciplinary nature of our topic, we utilized an adaptive approach to identifying studies. Most studies were identified using a combination of Google Scholar and the bibliographies of the studies.

By design, we filtered for studies from the Global South. Studies from 2020 about COVID-19 were almost entirely from the grey literature.

With greater density of attention to ridesharing and crowd work/freelancing it was more difficult to identify studies associated with the experience of selling via platform markets in agriculture and e-commerce. We will add more studies in these areas in later iterations.

Not all studies addressed the whole issue of experience. But if the study was able to capture first-person reports of some dimensions of the platform experience, we included it.   Studies that did not contain new primary research (ethnographies, surveys, observational designs) were excluded.


Frankly, we had begun with a plan to do fully inductive coding, and to include material from the Global North, but it quickly became apparent that the scale of that task and the number of possible themes emerging would overwhelm the resources we had allocated to the project. We elected after a couple of weeks of exploration to draw more directly on the Heeks (2017) and Zollmann and Wanjala (2020) typologies.

We read dozens of studies using Heeks’ and Zollmann and Wanjala’s (2020) typologies side-by-side. Then, later, we created a superset of elements drawing on the two frameworks, compressing and merging where possible. Heeks’ (2017) framework has positive and negative experiences as separate items, but in ways that often resemble opposite sides of a coin.  In our combined coding schema we selected language that was not inherently positive or negative, and then tagged studies as making “positive”, “neutral” or “negative” claims on that theme.

The four authors shared reading responsibilities, and there was considerable back-and-forth and refinement as we became comfortable with the coding scheme. However, we did not assign more than one reader to a study. Thus there is some risk that coder subjectivity would result in different evaluations of a given paper. 

We also added several “watch items” to our coding plan, given the interest of the Mastercard Foundation, on crosscutting factor: gender, youth, COVID-19 and rurality. These codes are not elements of the livelihood experience, per se, but rather important moderating and mediating variables that are helpful in filtering and contextualizing the experience findings. 

Throughout the process our team also identified several new themes, which we think are best described as emergent dynamics rather than elements of the experience.

  • Filters:
  • Sector
    • Freelancers
    • Microworkers
    • Ride Hailing Drivers
    • Delivery & Logistics Drivers
    • Trades & Services
    • Asset Owners
    • MSEs
    • Farmers
    • Creatives
  • Outcome
    • Access to Work & Markets
    • Inclusion
    • Objectivity & Professionalism
    • Earnings
    • Upskilling & Growth
    • Flexibility
    • Health & Safety
    • Betweenness & Protection
    • Association, Organization & Support
    • Social Acceptability
    • Purpose & Passion
    • Entrepreneurial Drive
  • Crosscutting Theme
    • Amplification
    • COVID-19
    • Contestation
    • Fractional Work
    • Gender
    • Hidden Hierarchies
    • None
    • Rurality
    • Youth
  • Country
    • Bangladesh
    • Cambodia
    • Canada
    • China
    • Colombia
    • Ghana
    • Global
    • India
    • Indonesia
    • Kenya
    • Malaysia
    • Mexico
    • Mozambique
    • Nepal
    • Nigeria
    • Pakistan
    • Philippines
    • SSA
    • Saudi Arabia
    • South Africa
    • Sri Lanka
    • USA
    • Uganda
    • Ukraine
    • Vietnam
  • Publication Year
    • 2010
    • 2012
    • 2014
    • 2015
    • 2016
    • 2017
    • 2018
    • 2019
    • 2020
Sector Outcomes
Access to Work & MarketsInclusionObjectivity & ProfessionalismEarningsUpskilling & GrowthFlexibilityHealth & SafetyBetweenness & ProtectionAssociation, Organization & SupportSocial AcceptabilityPurpose & PassionEntrepreneurial Drive


Key: Number of   studies; and of which, number of tests with a   positive impact   negative impact   no impact
The EGM was last updated in September 2020.

evidence points from studies