I have spent most of my career on the agency side. So take this with whatever salt you need: if you've just started something, you probably shouldn't hire one of us yet.

Not because agencies are bad. Because there is a set of things nobody can do for you, they all take one evening each, and until they're done, every rupee you hand to anyone — an agency, a freelancer, an ad platform — works at a fraction of its potential. I have watched founders spend six months of budget discovering this.

Here are the six.

1. Know what one order actually leaves you

Selling price, minus product cost, minus shipping, minus packaging, minus payment gateway fee, minus the cost of the ones that come back. What's left is the only number that decides whether advertising can ever work for you.

From it you get your break-even ROAS — the return below which every additional sale loses money. If that number is above about 4, paid advertising is going to feel impossible, and no agency on earth will fix it, because the problem isn't in the ad account. It's in your pricing, your product cost or your return rate.

This is one evening with a spreadsheet. It is the single highest-return thing a new founder can do, and it is the thing most often skipped.

2. Verify that your tracking matches reality

Place a real order on your own store. Then check: does it appear in your analytics? Does it appear in the ad platform? Do the rupee values match what the customer actually paid?

In a surprising share of new stores, one of those three is wrong. And if your numbers are wrong, every decision made from them is wrong too — including the decisions you're paying somebody else to make. You will spend months optimising toward a number that isn't real.

Twenty minutes. Do it before you spend anything.

3. Write down what you sell, in a sentence a stranger understands

Not a positioning statement. A sentence. What it is, who it's for, and why it's different from the cheaper thing next to it.

Founders who built the product themselves are the worst at this, and it's not for lack of intelligence — it's that you're too close to it. You know why the stitching matters. Your customer doesn't know stitching is a thing that varies.

Until this sentence exists, every ad you or anyone else writes is guessing at it. And an agency will guess at it too, more expensively, and then charge you for the rounds of revision while you tell them it's not quite right.

4. Talk to ten customers. Actually talk.

Phone calls, not a survey. Ask what nearly stopped them buying, what they thought it would cost, and what they compared it against.

You'll get your objections, your price anchors and your real competitors — which are almost never the competitors you assumed. Every one of those goes straight onto your product page and into your ad copy. This is the cheapest research in existence and almost nobody does it, because it's uncomfortable and it doesn't feel like work.

It is the most valuable afternoon in this entire list.

5. Fix the obvious leaks yourself

Not a conversion optimisation programme. The obvious things, on your own phone, off wifi:

  • Does the site load in under three seconds?
  • Is the delivery date visible on the product page, as a date?
  • Can someone check out as a guest, without making an account?
  • If a payment fails, are they immediately offered another method?
  • Is the returns policy answerable without leaving the page?

Every one of those is free and takes an afternoon. Paying for traffic to a store that fails three of them is the most common way I see new budgets disappear. You are buying visitors and pouring them into a bucket with holes you already know about.

6. Spend your first ₹30,000 yourself

Not to succeed. To learn what the machine feels like. Set up one campaign, one clear offer, three or four creatives, and let it run for two weeks without touching it daily.

You'll lose most of that money. That's the tuition, and it's cheap. What you get for it is the ability to tell, six months later, whether the person you're paying is doing something sensible or something expensive. Founders who have never run a campaign cannot evaluate the people they hire to run campaigns, and that asymmetry is where most of the bad agency relationships in this industry come from.

So when should you hire someone?

When you've done all six, and the constraint has genuinely become time or scale rather than knowledge. Concretely:

  • You know your break-even ROAS and you're beating it consistently
  • You have a channel that works and you can't personally give it enough hours
  • You can describe what you want done well enough to tell whether it was done

Hire before that, and you're outsourcing a problem you can't yet specify to someone who will interpret it however is convenient. Hire after, and you're buying capacity — which is the thing agencies are genuinely good at selling.

Never hire out a job you can't evaluate. Learn just enough to judge the work, then hire the person who does it better than you.

I would rather you came back to me in eighteen months knowing your own numbers than sign you today and spend our first quarter together discovering them at your expense.