What does Tattle pay?

Uli annotation exercise	Pay rate	Per annotator payment	Context
July 2022	8/post	16000 (2000 posts)	First annotation exercise creating a dataset of gendered abuse detection.
April -July 2024	1200/hr	10800 (6 90 min sessions)	AI safety benchmark dataset
April 2026	2500/hr	15000 (4 90 min sessions)	Expansion of annotated slur database in 8+ Indian languages.

In order to understand how Tattle has decided its pay rate for data annotation work over the years, it is important to first define data annotation work. Firstly, Tattle primarily works on producing safety datasets in multiple Indian languages that reflect the cultural variety and nuances of how safety is experienced by people in daily living. Secondly, we consider safety data to be a unique category within broader data work that requires skilled labor for annotation as it combines lived experience and language expertise to produce the best results.

How should I pay ethically?

When I was given the mandate to determine wages for the data annotators to be employed for a crowdsourcing exercise under Uli, I was told to go big. This felt like a dream and a big responsibility. I wanted to determine a rate that would not be just adequate or cover essentials, I wanted to start from the question of, “if an amount of money could represent feeling safe, excited, and secure in offering my labour, what would that amount be?” In other words, how did I want someone to feel when they came on board to work with us? Could I somehow encapsulate that amount in an hourly rate? This process was experimental and I am sharing below the different threads of theory, imaginative thinking, and calculations that went into it.

What are existing measures for calculating wages in India?

What is minimum wage?

Minimum wage in India is determined by type of employment by the central government and thresholds are determined by each state. Wages are divided by skilled, semi-skilled, and unskilled based on standard labour department definitions.

The current minimum daily wage for an unskilled worker in Maharashtra is ₹563¹, which comes to about ₹14,638 a month (calculated by multiplying the daily rate by 26 paid days). The daily rate for skilled labor is ₹690, which amounts to ₹17,490 per month. At an hourly rate of assuming 8 hours per work day, these minimum wages amount to ₹70.37 for unskilled workers and ₹86.25 for skilled workers. This rate is far below what is required for meeting even essential needs like rent and food as we will see in the calculations done for a living wage in the next section.

What is the living wage?

According to the Anker Research institute, a living wage is a standard rate that affords a decent standard of living defined as including food, water, housing, education, healthcare, transport, clothing, and other essential needs². As per their updated reports published annually since 2023, the monthly living wage adjusted for inflation in Delhi was ₹23,086. This amounts to about ₹887.92 daily, compared to the state minimum wage rate of ₹710. This living wage still does not account for savings or wants, and by using the pricing indices used by the state to calculate minimum wage, it continues to grossly underestimate the needs value of workers living in these states.

The difference between minimum and living wage is the difference between a person eating cheaper, less nutritious food and thereby increasing their healthcare costs and someone eating slightly better quality food and having some income for healthcare. However, neither measure ensures a sense of financial security that cannot be erased by a single emergency. Hence the need for a calculation of wages that help workers to thrive rather than merely survive.

How does one calculate an ethical wage?

There are many models one can use to determine an ethical wage. At different stages of the Uli project different considerations have gone into determining a payment model³. In this article, I am sharing one particular model that was used to determine pay for one instance of Uli data annotation work.

I started with thinking about how much I want to make for my work. What feels just, comfortable, and nourishing? How much value does one hour of my labour really produce in the world?

But thinking about value as only coming from what my labour gives to an enterprise is in itself unethical. It divorces my body and my wellbeing from my labour. So what is the investment I need in myself and my wellbeing to keep doing the labour that I am doing is also something that I want included in compensation for my work.

If we consider existing indices, minimum wage covers only the basics of food to stay alive. A living wage expands on this with food, rent, and healthcare as necessary components to keep a person alive. But radical challenges to the concept of work under capitalism now demands that we not only provide for bare necessities but for all the essentials that make life meaningful. Therefore, a wage that helps us thrive and not just survive. A thriving wage then covers wants and savings in addition to needs. It provides for fun and the ability to invest in our futures.

I decided that since I was being given the freedom to set an ethical wage for this project, I wanted people who worked with us to thrive. My work is in safety and making online public spaces more inclusive, accessible, and safe for marginalised groups. It counts as community care and labour that betters society.

Considering this I added a new component to the wage - the value of its social impact. How much does this labour improve life in society? So now we have a number of components to think about: Wage = rent + food+ utilities + healthcare + education + disposable income for wants + savings + value of social impact

I began with researching average hourly wages for data annotation work online. The median wage for data annotation work was around ₹345 while the median hourly wage for a data analyst was ₹696. Considering that we were hiring people for safety data annotations that required exposure to harmful content and hate speech, it was important to pay a higher salary to account for the social value produced by their work. Additionally, if the hourly wage in the market was around ₹696 at the higher end, it was important to consider the expertise being brought to the labour by our annotators. Under Uli, we have always hired annotators with expertise in combatting gender based violence either through their research, activism, writing, or lived experience. In addition, our annotators bring in language expertise as they share the specific nuances of harm as it occurs in different Indian languages. Referring to the daily living wage of ₹887.92 calculated by Anker Research Institute for a big metropolis such as Delhi, I had to raise the hourly wage to incorporate expertise and basic living expenses. This hourly rate needed to be adequate to cover rent, food, utilities and healthcare in metropolises like Mumbai, Delhi, Bangalore, where a majority of our annotators resided.

Raising the hourly rate above the daily wage rate for unskilled labour, if we take ₹1200/hr we arrive at a monthly income of ₹1,92,000/- for a regular 40 hour work week. This is an adequate salary to afford rent and the basic expenses in the bigger cities when considering the average median rents range from ₹25k to ₹50k in the biggest metros. Therefore, I decided to calculate a reasonable hourly living wage as follows:

Rent+Food+utilities+healthcare in a big city = ₹1200/hr

This living wage still lacks components listed above such as educational and self-improvement expenses, adding the value of the social impact of one’s labour in society, and components for fulfilling personal desires and wants.

This is where I employed the 50-30-20 rule for budgeting⁴, which states that for managing a budget healthily, about 50% of one’s income should be spent on needs, 30% on wants, and 20% on savings for the future. This would mean that the above hourly rate only makes up half of an income that allows one to thrive and manage our budget well.

Therefore, an hourly wage encompassing these aspects would be approximately ₹2400/hr.

I tweaked this number a little as we had the budgetary flexibility to add in an additional ₹100 to account for the fact that this work was producing the social good of increased safety in the digital commons.

The final thriving wage was therefore ₹2500/hr.

Footnotes

1. https://www.labourcodecalc.in/minimum-wage-2026 ↩

2. Anker Methodology | Discover accurate wages—Get Informed Now. (n.d.). Anker Research Institute. Retrieved 3 July 2026, from https://www.ankerresearchinstitute.org/anker-methodology ↩

3. To read more about the models of payment used in Uli for data annotation work see our blog "Our Approach to Data Annotation" written by Srravya Chandiramowuli. ↩

4. How do we quantify a thriving wage? (n.d.). Vega Mala Consulting. Retrieved 15 June 2026, from How do we quantify a thriving wage? (n.d.). Vega Mala Consulting. Retrieved 15 June 2026, from https://www.vegamala.com/blog/how-do-we-quantify-a-thriving-wage ↩