A clothing brand in Delhi came to me last year with one problem. Every time a customer placed an order, the brand was paying ₹7 to send WhatsApp updates about it. That ₹7 went to a notification reseller and bought four messages per order: confirmed, packed, shipped, delivered. Customers still wrote in to ask where their parcel was.
We connected their store straight to the Meta WhatsApp Business Platform and wired it to shipping updates from Shiprocket. Each order now costs ₹1.10 and sends ten messages instead of four. On a per-message basis the price fell by about 94%.
That looks tiny on a single order. A brand shipping 3,000 orders a month keeps around ₹18,000 every month that used to go to the reseller, and its customers hear from it often enough that they stop opening support tickets for basic questions. Most of that saving came from removing the reseller who was sitting in the middle.
I want to walk through a few of these projects and what they actually cost and saved. “Automate your operations” gets sold as one big expensive project. For a small business it is usually a set of small fixes.
Start with one job that wastes money every week
The common mistake is to go shopping for a system. An owner decides the business needs automation, asks around, gets a quote for a large custom build, and then either overpays or gives up.
Better to start small. Find the one task that eats money or hours every week. Write down how bad it is right now. Fix that task and nothing else. Once it works and you can point at the saving, pick the next one. The projects below all started from one small task.
A daily chore that now runs on its own
I built a receipt-tracking product called AMRIOP. In the first version every user signed up by hand and then uploaded each receipt themselves. That was four or five steps for every receipt, and it went on all day.
We moved signup to single sign-on and asked the user for one mailbox permission during that step. Receipts now arrive in their email and get pulled in and sorted into a separate tab on their own. Steps per receipt dropped from four or five to zero. People used to skip this job because it was dull. Now it happens without anyone touching it.
The cost a cheap build hides
Let’s Calendar is a scheduling tool that needed a way to import contacts in bulk. The quick version handled about 5,000 contacts and then gave up, timing out or running out of memory after roughly fifteen minutes. It got worse when a few people used it at the same time.
We rebuilt that import to run in the background in small chunks. On the same server it brought in 101,096 contacts in under a minute, and it held steady when we tested past 500,000. It also emails the user a list of any rows that failed, with the row numbers.
Same feature, same hardware, more than twenty times the capacity. The first version had been quietly losing users who tried it once and walked away. A low quote often hides this gap, and you find out about it after the money is spent.
Here is a quicker one. An Australian retailer, ArcticBlue, was on a setup where the average server CPU sat near 60% and spiked to 100%, which took the shop offline during working hours. We moved it onto a properly indexed Laravel build and added bot blocking at the edge. Average CPU now sits around 5%. That was on the same server, with no hardware upgrade.
Why these projects paid off
We measured the starting number before touching anything, every time, which is what let us prove the saving later. Where a reseller or a middle layer had been taking a cut, we went around it and connected the systems directly. Most of the effort went into making things dependable. And each fix sat on top of email, WhatsApp, or whatever the business was already running, so staff did not have to learn new software.
At an earlier job with Manomay Informatics we cut manual work for enterprise clients by around 80% with the same approach, built up from small pieces over time.
You can see the before and after figures for these projects, with live client links, on my portfolio.
How to brief a developer so you get the same result
The brief you give a developer decides how much you save.
Bring the task and its numbers. “We send about 3,000 order updates a month, it costs us this much, and staff spend this many hours on it” tells a developer far more than “we want a dashboard.”
The next question is what the thing will cost to run each month, not just what it costs to build once. A cheap build that breaks often or runs up bills is not cheap.
Get the target number in writing before any work starts. If nobody can name the number that is meant to move, the scope is still vague.
Then pick a single workflow and stop there. Get it running, watch it for a month, and decide what comes next. Most of the businesses I work with reach useful automation by repeating that loop and checking each change against the number it was meant to move.
Prashant Singh is a freelance web developer and AI automation specialist based in Kanpur, Uttar Pradesh, with over ten years building production software for clients in India, the US, the UK and Australia. His work and client case studies are at prashantwebdeveloper.in.
