Most Indian SMBs end up on the wrong commercial model for their software work, and they stay there for years. The retainer-versus-project decision is usually made once, in a hurry, often on price, and then never revisited. The cost shows up later as scoped-out requests, an agency that goes quiet between invoices, or a fixed-price project that drags six months past its end date while nobody admits it is dead. Founders and ops leads feel this most, because they are the ones chasing updates at 9pm.
What actually separates a retainer from a project? A retainer buys capacity and continuity; a project buys a defined outcome at a fixed price. That is the whole difference, and everything else follows from it. On a retainer you are paying for a block of hours or a standing team each month, and the work queue is expected to change. On a project you are paying for a specification, a deadline and a handover, and changes are supposed to cost extra. The mistake is buying a project when what you actually want is someone on call, or buying a retainer when you need one shipped thing.
Which metrics tell you the model is wrong? Four metrics will tell you within one quarter: backlog age, change-request volume, rework hours, and decision latency. Track them monthly and keep them boringly visible.
- Backlog age: median days from request raised to work started. Above 30 days on a retainer means you are under-buying capacity; near zero on a project means scope is creeping.
- Change-request volume: count of formal change requests per month. More than three a month on a fixed-price project is a signal the specification was never right.
- Rework hours: hours spent fixing something already invoiced as done. Above 15% of monthly hours, the model is hiding a quality or briefing problem.
- Decision latency: days your side takes to approve a decision. Above 5 days, no model will save you; the bottleneck is internal.
If you run a B2B portal or an internal tool used daily by your team, backlog age and decision latency are the two that matter most, because the tool decays the moment changes stop.
When does a project make more sense? A project is the right call when the outcome is genuinely definable up front and will not need to change for six months. Migrations, a one-time portal build for a known set of users, GST or e-invoicing compliance changes with a hard statutory deadline, or a website rebuild with a frozen content plan all fit. The test is simple: can you write the acceptance criteria on one page and get sign-off from everyone who will use it? If yes, fixed price protects you, because the vendor absorbs the estimation risk. If the page keeps growing new bullets every review, you are about to pay for that risk anyway, just as a change request instead of a monthly line item.
When does a retainer pay for itself? A retainer wins once the work is continuous, multi-threaded and hard to specify in advance. The moment your software has live users, support questions, seasonal peaks and a steady trickle of "can we also just..." requests, fixed-price scoping costs more in lost weeks than it saves in fees. A worked example: a distribution business with a 400-user order portal raises roughly 9 to 14 small requests a month. Priced as individual projects at 8-12 hours each, with two weeks of quoting and approval per item, they get through maybe four a month. On a 60-hour monthly retainer they clear the queue and keep 15-20 hours for planned work. The monthly fee is often similar; the throughput is not.
What thresholds should trigger a switch? Set the trigger before you need it, in writing, so the conversation is about a number and not about trust. In our experience these are the thresholds worth agreeing on:
- Move project to retainer when change requests exceed three a month for two consecutive months, or when backlog age crosses 30 days on ad-hoc work.
- Move retainer to project when under 40% of retainer hours are used for three months running, or when the work has narrowed to one definable deliverable.
- Renegotiate retainer size when utilisation sits above 90% for two months, or rework exceeds 15% of hours.
- Escalate either model when decision latency passes 5 working days or the same defect is reopened twice.
Review these quarterly, on one page, with your vendor in the room. If you want a starting point for what the numbers should look like, our pricing page sets out how retainers are sized and what happens when a month runs hot or quiet.
How do you handle the commercial mechanics? Get the boring details into the contract, because they decide whether the model actually works. For Indian vendors, retainers are invoiced monthly with GST at 18%, and TDS under Section 194J applies at 10% on professional or technical services, so budget for the cash-flow gap between invoicing and credit. Ask for a written rollover policy (typically 0-25% of unused hours, capped at one month), a notice period of 30 days on either side, and an out-of-scope rate fixed in advance. On projects, tie 20-30% of fees to acceptance, not to a launch date. Paying by UPI or NEFT is fine; what matters is that the payment milestone is written down. If you are evaluating a partner before you commit to either model, a short scoping engagement is usually cheaper than signing a year of the wrong one.
Where to start Pick the four metrics above, pull the last three months of data, and see which threshold you have already crossed. That single exercise usually settles the argument. If you want a second pair of eyes on it, Maysan Labs runs this review with clients before recommending a retainer or a fixed-price build, and we will tell you when a project is the cheaper answer.
Buying software or a website?
Twelve questions to ask any vendor before you sign: price, your data, fit for an Indian business and support.
Get the free checklistRelated case study
Storefront we build and run ourselves
We needed a storefront we could sell and run, not only build: stock, orders, payments and customers in one place.
Written by Sanjay Mandal at Maysan Labs, a software studio in Gurgaon, India. Follow us on LinkedIn