A founder showed us a Meta dashboard last month. ₹2.4 lakh spent in three weeks. ROAS of 0.7×. He wanted to know if changing the creative would fix it.
It would not. The problem wasn’t the ad. The problem was upstream of the ad. We told him to pause everything and go fix the website first.
This is a much more common scenario than most agencies will admit, because most agencies make money when the ads are running. So here’s our list of times we tell people not to run paid.
1. When the landing page is slower than 3 seconds
Indian mobile traffic on Meta is real-world flaky. If your site takes four seconds to show anything useful, you’re paying to acquire a click that bounces before the headline even loads.
Open your site on a Reliance Jio 4G connection in any non-metro city. If you’d close the tab as a normal person, your customers definitely will.
Fix this first.
Run it through PageSpeed Insights. If your Largest Contentful Paint is over 3 seconds, no amount of ad budget is going to make the math work. We’ve seen brands turn ROAS from 0.8× to 4.2× without changing a single ad — just by fixing the site.
2. When you haven’t proven organic demand exists
Paid ads are good at amplifying demand that already exists. They are bad at creating demand that doesn’t.
If nobody is searching for what you sell, nobody is talking about it organically, and you’ve never had a sale that wasn’t from a friend — paid ads will mostly burn money proving that nobody wants the thing yet.
Start with three weeks of organic. Post on Instagram, post on LinkedIn, talk to ten people who fit your target. If even one stranger asks how to buy, that’s a signal you can amplify. If not, you have a product question, not an ads question.
3. When you can’t tell us what a customer is worth
This one is almost a trick question. Ask any founder “what’s the lifetime value of your average customer?” and watch how confidently they answer. If the answer is hesitant, paid is risky.
Without an LTV estimate, you have no way to know whether ₹450 to acquire a customer is great, fine, or catastrophic. You’re flying blind. Stop, run the math, then come back.
4. When the brand is brand new and the product isn’t shipping yet
We’ve watched founders run pre-launch “awareness campaigns” and end up with 8,000 site visits and zero email captures because the site said “coming soon” and didn’t even ask for an email.
If you’re not selling yet, paid is for one thing only: list-building. And even then, only after the email-capture experience is solid. Otherwise you’re paying for traffic that has nowhere to go.
5. When the agency proposes “retainer-based optimisation” before testing anything
Self-serving, but true. If someone wants ₹40,000 a month to “manage and optimise” paid campaigns before they’ve tested even one creative against another, they’re charging for the office, not the work.
A reasonable agency runs three or four small tests cheaply, looks at what worked, and then proposes a retainer. We’ve done it the other way and watched ourselves spend more time defending the retainer than running the ads.
When paid actually works
It works when:
- Organic demand already exists and you want to scale it
- The site loads fast and converts at a known rate
- You know what a customer is worth
- You can lose two weeks of budget without panicking, because that’s roughly how long it takes to find a winning creative
Most brands we audit are missing at least two of these. The right answer is almost never “spend more.” It’s usually “spend differently, somewhere upstream.”
Paid is a lever, not an engine. It amplifies what’s already working. If nothing’s working yet, no amount of pulling on the lever will start it.
If you’re not sure whether your brand is ready for paid, we do audits. Fixed fee. Honest answer. Pour us a brief — or read more about how we run paid media and our fixed-fee marketing audits.
What "fixing the landing page first" actually means
When we say "fix the site before running ads," founders sometimes hear "do a redesign." That's not what we mean. Nine times out of ten, the fix is smaller than that.
The most common problems we find on Indian D2C landing pages that are eating ad spend:
- The hero image is a 4MB PNG. On a Jio connection, this takes six seconds to render. The customer bounced before the page said anything.
- The call to action is below the fold on mobile. Sixty-two percent of Indian mobile traffic scrolls less than one full screen on an unfamiliar site. If your "Buy Now" or "Learn More" is three scrolls down, it might as well not exist.
- The headline describes the product, not the problem it solves. "Premium Cold-Pressed Argan Oil" versus "Stop washing it twice — one oil that actually absorbs." The second headline gets clicked. The first gets scrolled past.
- There's no social proof visible without scrolling. Indian buyers are highly review-sensitive. If a first-time visitor has to look for reviews, they won't.
- The checkout has three extra steps. Each step in a mobile checkout that requires a tap loses 15–20% of buyers in India, where UPI is instant and any friction is genuinely painful to compare.
Most of these fixes take a developer a day, not a month. We've seen ROAS go from 0.8x to 3.1x by fixing just the first two on this list, with zero change to ad creative or budget.
The real cost of running ads too early
It's not just the money you spend on ads that don't work. It's what that spend teaches you.
When you run ads on a broken landing page, you learn nothing useful about whether your product or your offer is appealing. All you learn is: "this site doesn't convert." You've spent ₹2 lakh to discover something a five-minute PageSpeed test would have told you for free.
Worse, if you ran those ads at low ROAS for more than a few weeks, your Meta pixel now has a history of high-cost-per-conversion data attached to your account. The algorithm has learned your ad account is "hard to monetise." You'll pay a higher CPM for the same audience than a fresh account would — sometimes for months, until you generate enough positive conversion data to reset the signal.
Starting paid ads on a site that converts at 0.5% instead of 2% doesn't just waste today's budget. It contaminates the learning you'll need to scale tomorrow.
A pre-launch paid ads checklist
Before we turn on paid for any new client, we check four things. All four need to pass:
- LCP under 3 seconds on mobile, Indian 4G. Test at PageSpeed Insights. If it fails, pause and fix before spending.
- Organic conversion rate above 1%. Some real humans who found you without ads need to have bought the thing. If zero strangers have converted from organic traffic, you don't have enough signal to amplify.
- Customer lifetime value is estimated. Even a rough number (₹800 first purchase, 2.3x repeat rate = ₹1,840 LTV) is enough to set a maximum acceptable CAC. Without this, you're spending without a ceiling.
- A testing budget you can afford to lose. The first two weeks of any new paid campaign is a learning exercise. You're buying data, not conversions. If losing ₹50,000 would cause a genuine cash-flow problem, this is not the month to start paid.
If you pass all four, paid ads become a lever. If you fail any one of them, you're not ready — and no amount of creative optimisation will change that.
When paid actually works in India — specific examples
We've managed campaigns with ROAS above 4x. Here's what those brands had in common that the others didn't:
They all had repeat buyers before they started ads. Not a lot — sometimes just 20 or 30 people who'd bought twice. But the existence of a repeat purchase told us the product had a real fan, not just a first-time curiosity buyer. Ads to cold audiences work much better when the offer has been proven on a warm audience first.
They all had at least one organic Instagram post that had saved above 2% before we started paid. We use that post as the first creative test. The format, message, and thumbnail that already worked organically almost always outperforms the "professional creative" that looks like an ad from the first frame.
And they all had a clear answer to the question: "what should the customer do after they land on the page?" It sounds obvious, but a surprisingly large number of Indian brand sites have a landing page that is beautiful and says nothing specific about the next step. Paid ads are very good at getting someone to a page. They cannot fix a page that doesn't know what it wants from the visitor.