Why Predictable Pay May Be Gig Platforms’ Cheapest Retention Tool

 By Saurabh Kumar Roy

A delivery rider’s average earnings can look healthy while the income feels impossible to plan around. Platforms design pay to balance supply and demand; workers experience it as a puzzle. The gap between the two may be costing more than it appears.

How a delivery partner is paid

Pay in food and quick-commerce delivery is layered:

  • Base pay for distance and estimated time. Media reporting on Zomato and Swiggy puts it at roughly ₹20–50 for deliveries within about five kilometres, plus ₹8–10 for each additional kilometre.
  • Surge for demand–supply mismatch, bad weather or less desirable areas.
  • Incentives or “quests” triggered by hitting targets, with common slabs at 10, 20 or 25 orders in a day.
  • Tips, which are small: about 5% of Zomato orders are tipped, only about 1 in 25 users tip on Zomato and Swiggy, and tips across all platforms came to about ₹150 crore in 2025.

On levels, Eternal reported average hourly earnings, excluding tips, of ₹102 in 2025, up from ₹92 in 2024. For partners logging at least eight hours a day for 26 days, average monthly earnings excluding fuel were ₹23,709 in 2021, ₹27,109 in 2023 and ₹27,726 in 2024, so growth has moderated even as order volumes rose. At ten hours a day, figures shared publicly by Eternal’s founder put gross earnings near ₹26,500 and net near ₹21,000 after about 20% for fuel and maintenance, comparable with entry-level formal jobs paying ₹18,000–22,000. As reported, daily earnings depend less on individual order payouts than on time logged in, incentive thresholds and peak-hour work.

The worker’s side: earnings that are hard to read

  • Understanding. In the RIGHTS survey (Janpahal, 2024; 5,220 workers across 32 cities), about half of workers said they did not fully understand how their payments are computed and only 35% said they did, with little difference by tenure. Meanwhile 78% felt underpaid and about 70% said they earned more in previous jobs, although most said payments arrive promptly.
  • Preference. In NCAER’s 2023 survey, 44.9% of workers preferred a fixed amount with no incentive.
  • Net earnings. In my own discussions with riders, people could not say clearly how much they earned versus saved after costs. Weekly payouts did not help, because the amounts were too small to feel material. Mental accounting, where surge and incentive money is treated differently from base pay, and vehicle EMIs and depreciation blurred the picture further.
  • One walkthrough. In my walkthrough of one quick-commerce platform’s rider app, an incentive unlocked only at 25 orders and was worth far more than base pay; reaching it would have meant 12–14 hours of work. A “guaranteed earnings” feature paid only just above its threshold. Indicated earnings varied by as much as 50% between time slots, and the rider saw an estimate rather than the surge itself. That is one platform seen by one observer, so I treat it as a generator of hypotheses, not evidence.

What the research says

  • Workers value predictability. Cameron and Meuris (2026, PLOS ONE) report that the constant hustle to reach minimum pay thresholds exhausted ride-hail workers and led some to leave for predictable salaried jobs at lower pay. The shape of the variability mattered as well as its size, and attrition fell when workers felt they had agency over their pay.
  • Uncertainty has a cost. Corgnet and co-authors (2026, Journal of Applied Psychology) report that pay uncertainty reduces engagement in recovery, with perceived financial scarcity as part of the mechanism.
  • Targets can backfire. Benjaafar, Ding, Kong and Taylor (2022) show that when many workers income-target, a higher surge can reduce supply, because workers who hit their target log off.
  • Timing and tangibility of money matter. As I read Li Ding’s 2025 study of about 4,000 drivers and 2.8 million deliveries, cash collected on delivery worked like immediate liquidity: the probability of ending a shift rose sharply as drivers neared a platform cap on cash they could hold, yet drivers handling more cash were also less likely to quit and more likely to log in the next day.
  • Opacity changes behaviour. Cameron (2021, Organization Science) found that “efficiency” drivers kept their own ledgers because they did not trust the algorithm, and Rahman (2021, Administrative Science Quarterly) found that opaque ratings pushed workers into trial-and-error and superstition. Opaque pay plausibly does the same.
  • Predictability can pay for itself. In a field experiment at Gap, Kesavan and co-authors (2022, Management Science) found that responsible scheduling raised store productivity by 5.1%, with sales up 3.3% and labour down 1.8%. These were retail employees, not gig workers, but the result shows that predictability need not cost money.

The platform’s side

Dynamic pay is a legitimate operational tool. Surge and incentive structures help match supply to demand (see Cachon, Daniels and Lobel, 2017; Besbes, Castro and Lobel, 2021), and platforms point out that most workers earn above minimum wage and value the flexibility that fixed schedules cannot offer. Both points deserve weight.

But complexity has a price. Workers who cannot read their pay cannot respond to it as designed, and unpredictable income feeds the churn described in the previous article. The question is whether the cost of that churn exceeds the matching benefit of pay that is hard to read.

Two experiments I want to run

  1. Weekly versus daily payout. Compare log-in and delivery patterns (inputs) with net monthly savings (outputs). Workers who choose a payout schedule differ from those who do not, so assignment would need to be randomised among new riders. Outcomes: 30, 60 and 90-day retention, hours worked and comprehension of earnings.
  2. Time-block pay. Pay for committed blocks rather than per delivery, with predictable minimum hours, drawing on scheduling research (core hours, guaranteed minimum hours, self-scheduling, shift swaps). Compare effective wage and platform cost with per-delivery pay, and test whether it narrows the seasonal supply dip described in the first article.

Limits and trade-offs

  • Predictability can cost flexibility. In my notes on platform best practices, workers weighted flexibility above support, and some safety features added complexity.
  • The survey evidence is self-reported and uses different sampling frames, and my app walkthrough covers one platform.
  • Results from US retail and New York or Chinese ride-hail settings may not transfer to Indian delivery work.
  • Nothing here is yet causal. The experiments above are how I would test it.

Sources

  • Moneycontrol reporting on delivery-partner pay models and on tipping at Zomato and Swiggy; Eternal disclosures and public statements by its founder.
  • Janpahal (2024). The RIGHTS Survey. NCAER (2023). Food delivery platform workers survey.
  • Cameron, L. D., & Meuris, J. (2026). The perils of pay variability. PLOS ONE, 21(5), e0347194.
  • Corgnet, B., Hernán-González, R., Sayre, G. M., & Gächter, S. (2026). No rest for the weary: Pay uncertainty reduces engagement in recovery. Journal of Applied Psychology, 111(5), 638–657.
  • Benjaafar, S., Ding, J.-Y., Kong, G., & Taylor, T. (2022). Labor welfare in on-demand service platforms. MSOM, 24(1), 110–124.
  • Ding, L. (2025). The impact of payment on delivery on gig drivers’ behavior in developing economies.
  • Cameron, L. D. (2021). “Making out” while driving. Organization Science, 33(1), 231–252.
  • Rahman, H. (2021). The invisible cage. Administrative Science Quarterly.
  • Kesavan, S., Lambert, S. J., Williams, J. C., & Pendem, P. K. (2022). Doing well by doing good: Improving retail store performance with responsible scheduling practices at the Gap, Inc. Management Science.
  • Cachon, G. P., Daniels, K. M., & Lobel, R. (2017). The role of surge pricing on a service platform with self-scheduling capacity. MSOM.
  • Besbes, O., Castro, F., & Lobel, I. (2021). Surge pricing and its spatial supply response. Management Science, 67(3), 1350–1367.