Two lakh rupees. That was the number I typed into Gemini this week. If a 50 to 100 person IT company in India spent ₹2 lakh a month on Google Ads, could it actually win software projects from the US? Or would the money just disappear into clicks from students and job seekers?
I started with a broader question. Everyone I talk to in this industry says the same thing: US projects have dried up. So I asked Gemini whether there is still a way in. Eight questions later I had a positioning strategy, five types of projects to chase, a landing page layout, a word for word sales call script and two budget plans, one at ₹2 lakh and one at ₹6 lakh.
It was a lot of good material. It was also confident about numbers it could not have known, and at one point it wrote a client success story that never happened. So I went back through all of it, checked what I could, and wrote down what I would keep and what I would throw away.
Is it really that hard right now?
Yes, and the numbers back up the mood. In January, Reuters reported that brokerages expected India's top six IT companies to grow revenue by only about 4% in the December quarter. Indian software exporters had not posted double-digit growth since the March 2023 quarter. US clients were nervous about tariffs and holding back on big programmes. Then there is the $100,000 fee on new H-1B petitions, and a Senate bill called the HIRE Act that proposes a 25% tax on US payments to foreign workers. The bill has not passed, and many people think it never will. But a proposal like that is enough to make a US finance team hesitate before signing an offshore contract.
The big firms are finding their way around it. TCS reported this month that its AI business has crossed $3.1 billion a year, and it signed $9.6 billion in new deals in the quarter. That is not a company in trouble. The pain is lower down, in the 50 to 200 person firms that have spent fifteen years selling one thing: good developers at a fair hourly rate.
Gemini's first answerStop selling hours
Gemini's opening argument was simple. US companies are not spending less on software. They are buying it differently. They have stopped buying coding hours and started buying outcomes: lower risk, faster launches, a problem that goes away.
I agree with this completely, and I think it is the most useful thing in the whole conversation. Every CTO in America gets the same email every week. Dedicated React developers, 50 engineers, ISO certified, great rates. It does not matter how true it is. It reads exactly like the hundred other emails in the same inbox. Gemini's suggestion was to sell a fixed piece of work with a clear finish line instead. Here is the shift it described, in my words.
| The old pitch | The pitch Gemini suggested |
|---|---|
| We have 50+ developers | We build HIPAA-compliant portals for small US clinics |
| Hire React devs at low rates | We modernise your old billing tool in four weeks, fixed price |
| Staff augmentation | A small delivery team that owns a milestone end to end |
| 5,000 cold emails a week | 30 researched, personal problem audits a month |
It also told me to stop chasing Silicon Valley. Funded startups there either have big in-house teams or a list of vendors they already trust. Gemini pointed at mid-sized companies in places like Texas and the Midwest instead: freight brokers, medical billing firms, regional manufacturers. Businesses making between $20 million and $250 million a year, running on old portals and spreadsheets, with nobody inside who has time to fix them. I like this. Those companies are not impressed by your logo wall. They are impressed when you understand how a freight load gets booked.
The Google Ads questionPeople do not google "software company". They google their problem.
When I asked what kind of projects could come from Google Ads, Gemini said something that changed how I thought about the whole thing. Nobody searches Google to find a long-term technology partner. They search when something is broken, when a deadline is close, or when a framework they depend on is about to stop being supported.
So it listed five kinds of projects that people search for in a hurry. I found this list genuinely smart, because each one is built around pain rather than a service name.
| Project type | What triggers the search | Deal size Gemini gave |
|---|---|---|
| Codebase rescue | The last vendor vanished or left messy code before a launch | $8,000 to $35,000 |
| Migrations | The product runs on a framework that is no longer supported | $20,000 to $80,000+ |
| API integrations | The company has outgrown Zapier and needs real data pipelines | $10,000 to $40,000 |
| Compliance portals | A clinic or finance firm needs software that passes an audit | $25,000 to $60,000 |
| No-code to custom | An app built on Bubble or WordPress cannot handle growth | $15,000 to $50,000 |
It was just as clear about what does not work on Google Ads. "Hire dedicated developers" is a fight against Toptal, Turing and Upwork, who can outbid any small firm. "Mobile app development" attracts people who want an Uber clone for $2,000. And nobody buys a six-figure enterprise consulting contract from a search ad. Those go through referrals and vendor lists.
Then I asked the obvious question. Do these search terms even have volume? Gemini admitted that each exact phrase gets maybe 20 to 150 searches a month across the US. Its answer was that low volume is the point. If one project is worth $30,000, you do not need thousands of clicks. You need one right person. So instead of one keyword, you group 10 to 25 variations of the same problem into a single campaign. "Take over existing web application", "emergency React developer", "fix legacy code developer". Each one is tiny. Together they add up.
From a sales point of view, this is exactly right. I would rather have five calls with people who have a real problem this week than fifty calls with people who are just curious.
The budgetWhat ₹2 lakh a month actually buys
This is where I gave it my number. Gemini's advice was firm: put all of it on one problem, with 10 to 15 related keywords and one landing page. Do not split ₹2 lakh across rescue, migration and mobile apps, because each campaign ends up with about $15 a day, which is barely one click. You would learn nothing from any of them.
Its estimate was $18 to $28 per click. I could not verify that figure. The most recent broad benchmark I found is WordStream's 2026 report, which looked at more than 13,000 US search campaigns and put the average across all industries at $5.42 a click and $66.69 a lead. Niche B2B software keywords almost always cost a lot more than that average, so Gemini's range sounds believable. But the only number you should plan with is the one Google Keyword Planner shows you for your own keyword list, before you spend a rupee.
I ran Gemini's funnel again with today's exchange rate, and the arithmetic holds. Here is how it plays out.
| Stage | Assumption | Per month |
|---|---|---|
| Budget | ₹2,00,000 at ₹94.75 per dollar | About $2,110, or $70 a day |
| Clicks | $18 to $28 per click | 75 to 117 |
| Enquiries | 3.5% to 6% of visitors fill the form | 3 to 7 |
| Real calls | Some enquiries are not a fit | 2 to 4 |
| Deals | 25% to 33% of calls close | 0 or 1 |
Zero or one. That is the honest answer, and I respect Gemini for saying it plainly. A first project of $4,000 is about ₹3.8 lakh, which pays back the month on its own. A rescue that turns into a monthly team of three or four people at $12,000 to $18,000 is ₹11 to ₹17 lakh a month. So one win can carry several months of ads. But some months will have no win at all, and a company needs to budget for at least three months before deciding whether this works.
Gemini said a well-matched landing page lifts your Quality Score and gives you a 30% to 50% discount per click. Google's own help page says Quality Score is a diagnostic tool and is not an input in the ad auction. What actually affects your cost is what the score summarises: how likely people are to click, how well the ad matches the search, and how useful the landing page is. So the advice is right. Build a matching page. But the percentages are someone else's estimate, not Google's.
Great layout. Do not copy the case study.
Gemini was very clear that the ads should never go to the company homepage. If someone searches for an AngularJS to React migration and lands on a page that says "we build digital experiences for enterprises", they hit back immediately. One page per problem group, it said, with a headline that repeats the problem in the visitor's own words.
The rescue page it designed is good. No top menu, so there is nowhere to wander off to. A short form asking for the tech stack and the biggest pain. A promise of a code review in 48 hours. A mutual NDA signed before anyone looks at a single file. And a line saying all the code and IP belong to the client. Those are exactly the fears a US founder has about an offshore team, and the page answers each one before they ask.
Then, in the middle of the page, there was a case study. A US freight broker. A Django and React dashboard. "Fixed 42 broken API endpoints." "Reduced database load times by 68%." And a quote from a CTO.
None of that happened. Gemini invented the client, the numbers and the quote to fill a template. It is a placeholder, but it looks so real that a busy team could publish it as is. In a pitch built on trust and NDAs, one made-up testimonial that a prospect checks is enough to lose the deal and the reputation. Use your own real project, even a small one, or leave the section out until you have one.
I think this is the biggest lesson of the whole conversation, and it goes beyond IT sales. AI tools are very good at the shape of a thing. They will happily fill that shape with details that sound specific and are completely fictional. The specificity is what makes it dangerous.
Will this work for a 50 to 100 person firm?Small at the front door, big behind it
This was my sixth question, and Gemini's answer had my favourite line of the day.
"Boutique Front Door, Factory Back-End."
Look like a small expert team on the first call. Bring in the full bench once they trust you.The idea is that a US buyer with a broken codebase is scared of two kinds of vendors. The freelancer who might disappear, and the giant IT firm that brings six people to a fifteen-minute call and asks for a six-month commitment. A mid-sized Indian firm sits in between. It can act like a small specialist team on day one and then quietly roll in an architect, three engineers and a tester once the client sees the first results. A freelancer cannot do that second part. That is a real advantage, and I had not thought of it that way before.
Gemini also said this will fail if the company runs it through its normal sales process, and here I fully agree from experience. Do not put a junior business development person on the first call to ask about budget and timeline. A founder whose product is broken wants to talk to an engineer. The person on that call should be the director of engineering or a principal architect, someone who can ask whether the app even builds outside one developer's laptop.
The part Gemini mentioned but did not really spell out is the time difference. 8 AM in New York is 6:30 PM in Kerala in winter. A 6 PM close in California is 7:30 AM the next morning here. If these leads are in a panic, a reply after 18 hours is a lost lead. So the senior engineer who takes these calls is working evenings, every week. That is a people decision, not a marketing one, and I would sort it out before spending the first rupee on ads.
The sales callCharge for the first look
The last thing I asked for was a script for a twenty-minute discovery call. Gemini's version never quotes a full project. It asks a few sharp questions, explains that anyone who quotes a rescue without seeing the code is guessing, and offers a paid 48-hour code audit for a flat $4,000. If the client then hires the team, half the fee is credited to the first month.
I like the paid audit a lot. It filters out people who only want free advice, it gets you inside the code, and the client walks away with something useful either way. My only worry is that many Indian firms are used to giving the first review away to win the deal. Charging ₹3.8 lakh for two days of work will feel strange the first few times. I think it is worth getting used to. A free review tells the client your senior people's time is worth nothing.
The follow-up questionWhat if I triple it to ₹6 lakh?
After the call script, I asked one more thing. What happens at ₹6 lakh a month? Gemini's answer was better than I expected, mostly because it did not tell me to pour three times the money into the same ten keywords. Those searches have a ceiling. Past that point you just pay more for the same few clicks.
Instead it split the money three ways. ₹2.5 lakh stays on urgent rescue searches. ₹2.5 lakh goes to planned migrations, the AngularJS to React and monolith to microservices kind of work. And ₹1 lakh goes to retargeting, which means showing ads again to people who visited the landing page but left without booking a call.
The migration track is the smart part. A rescue lead is in a panic and decides in days. A migration lead has a budget that was approved months ago and a project manager comparing three vendors. Gemini said those deals take three or four weeks to close but are worth $40,000 to $120,000. I could not verify that range, but the logic is sound. One track brings fast cash. The other brings bigger, calmer projects.
| ₹6 lakh plan | Gemini said | When I checked |
|---|---|---|
| Budget in dollars | $6,500 to $6,700 | About $6,330 at ₹94.75 |
| Daily search spend | About $220 | About $174, since ₹1 lakh goes to retargeting |
| Search clicks | 240 to 320 | 188 to 293 at $18 to $28 a click |
| Enquiries | 10 to 16 | 7 to 18 at a 3.5% to 6% form rate |
| Two audits at $4,000 | ₹7.4 lakh | ₹7.58 lakh, before the 50% credit its own script offers |
Most of the gap is the exchange rate again. Gemini seems to be working with about ₹90 to the dollar. At today's rate, ₹6 lakh buys a little less than it thinks. The overall picture still holds, though. Roughly three times the budget gives roughly three times the leads, plus a second kind of project to sell.
The retargeting money is where I would slow down. Gemini wants ₹1 lakh spread across LinkedIn, Google Display, YouTube and brand search. That is about $1,050 a month across four places. Retargeting also only works once enough people have visited. Google now lets an ad run to a list once it has 100 active users, which this campaign would reach in a few weeks. LinkedIn needs at least 300 matched members before it shows a single ad, and it can only match visitors who are logged into LinkedIn. With 200 to 300 ad visitors a month, that could take two months or more. I would start retargeting on Google and add LinkedIn once the list is big enough.
The retargeting ad Gemini suggested reads: "How we stabilized a 400k-line Django/React build in 3 weeks with zero downtime." Like the freight broker story, there is no such project. This time the made-up proof would run as a paid ad, in front of the exact people deciding whether to trust you. Use a real project, or a plain explanation of how your audit works.
The operational part of the answer is what I agree with most. At ₹6 lakh, Gemini expects two or three serious calls a week. It says you need two senior engineers trained on the call script so no US lead waits more than a day for a slot, and form submissions sent straight to a Slack channel with a phone alert. Its example was an enquiry at 9 PM in India that needs a reply within 10 minutes. That is the same time difference problem from before, with more leads riding on it. One small correction: 9 PM in India is 11:30 AM in New York only during US daylight saving time. From early November to mid-March it is 10:30 AM.
Would I spend ₹6 lakh? Not on day one. I would run ₹2 lakh on rescue for three months, learn what a click really costs and how many calls turn into paid audits, and only then add the migration track. Gemini called ₹6 lakh "a predictable client acquisition engine". It might become one. But you only learn whether something is predictable after you have watched it work at the smaller size.
What I actually thinkAds are the smaller half of the answer
If I were running a 50 to 100 person company and had ₹2 lakh a month, I would do what Gemini said with one change. One problem, rescue or a single migration path. One honest landing page with a real project on it. An engineer on every first call. A paid audit as the first sale. And a promise to myself to run it for three months before judging it.
But I would not rely on ads alone. Some of the best ideas in the conversation had nothing to do with Google. Partner with small US design agencies that win app projects but have no developers to build them. Get to know fractional CTOs, who each advise four to six companies and need a team they can trust. Watch US job boards for senior engineering roles that have been open for three months, because that usually means a project is slipping. Record a short video pointing out real problems on a prospect's website instead of sending another cold email. None of these cost ₹2 lakh. All of them build the kind of trust a search ad cannot.
So, can an Indian IT firm still win US projects in 2026? I think yes. Just not the way most of us were taught to sell them. The US buyer has not stopped spending. They have stopped buying people. They are buying the problem going away, and they want to talk to the person who will actually fix it.
Common questions