When to spend money on ads

You've got £200 sitting there and a feeling that this is the bit you're missing. Everyone selling a course says ads are how you scale. The platform keeps offering to boost your post for £6 a day and telling you it'll reach 3,400 people.

For most sellers reading this, the honest answer is: not yet, and probably not for a while.

Not because ads don't work. Because ads multiply whatever your shop already does, and if your shop currently turns 100 visitors into one order, ads turn 1,000 paid visitors into ten orders and a bill. The arithmetic that decides this takes about ten minutes and almost nobody does it before spending, which is why so many small sellers have a story about the £200 that vanished.

Work out your number first. Here's how, and what to do instead if the number says no.

The number: the most you can pay for one customer

It's your gross margin on one order. Not your price. Your margin, after materials, packaging, delivery and your own time.

If you make £5 on an order, then £5 is the absolute ceiling on what you can pay to acquire that customer, and paying exactly £5 means you did all the work for nothing. To actually earn something, you need to pay well under it. Half of it is a reasonable target.

That gives you a second number: the most you can pay for a click. Multiply your margin by your conversion rate, because most clicks don't buy.

Margin per order	Converts at 1%	At 2%	At 3%	At 5%
£5	5p	10p	15p	25p
£10	10p	20p	30p	50p
£25	25p	50p	75p	£1.25
£50	50p	£1.00	£1.50	£2.50
£100	£1.00	£2.00	£3.00	£5.00

Those are break-even click prices. Pay that and you've earned nothing.

Look at the top-left corner. If you sell a £15 item on a £5 margin and cold traffic converts at 1%, you can pay 5p a click to lose no money. There is no ad platform in any country where you buy clicks at 5p from an audience that buys. That's not pessimism, it's arithmetic, and it applies to a large share of small physical-goods sellers.

The third number, if you prefer thinking in return: break-even ROAS is 1 divided by your margin. 30% margin means you need £3.33 back for every £1 in, before you've made anything. 50% margin means £2. Every "we got 4x ROAS!" screenshot you've seen means something completely different depending on which of those the person was working with, which is why the number on its own is meaningless.

Three things that have to be true before you spend anything

One: something already sells without paying for traffic. You need proof a stranger will buy, at your price, without you personally messaging them. Not friends. Strangers. If you can't point at three of those, you're about to pay to find out something you could learn for free by asking forty people, which is what getting your first ten orders is about.

Two: your page converts the traffic you already have. Go and count. How many people opened your shop last month, and how many bought? If it's 200 and zero, ads make it 2,000 and zero. Fix the shop first. What to do when nobody is buying is a six-step diagnostic that costs nothing.

Three: your margin can carry it. From the table above. If the answer is 5p a click, stop here.

There's a fourth condition that's less obvious. Ad platforms optimise by learning from conversions, and a shop doing nine orders a month doesn't generate enough of them for the optimiser to learn anything at all. You'll be funding its education with a sample size too small to teach it. This is the reason a £6-a-day boost so reliably produces reach and no sales: it's not that the money is too little, it's that the signal is.

What Tomas spent in Bristol, and what it bought

Tomas makes small-batch hot sauce in Bristol. He sells a three-bottle set for £19 with £3.95 postage, mostly through Instagram, mostly paid by card or by bank transfer for people he knows.

His costs on a three-bottle set:

Line	Cost
Bottles, caps, labels (3)	£3.30
Ingredients (3)	£4.80
Box and mailer	£1.20
His time, 20 min at £14/hr	£4.67
Total	£13.97
Price	£19.00
Margin	£5.03 (26.5%)

He'd been going eight months, was doing about eighteen sets a month, and decided to put £200 into ads to break out of it.

What he got: 640 clicks at an average of 31p, and seven orders. £133 of revenue against £198 of spend.

Cost per customer: £28.29. Margin per customer: £5.03. He lost about £163 on the campaign, and the seven customers would have to buy from him five more times each before that spend broke even.

The number he should have calculated first: to break even at £5.03 a customer on £198 of spend he needed 40 orders from 640 clicks. That's a 6.2% conversion rate from cold traffic.

His own link, from people who already followed him, was converting at about 4.4%.

He needed strangers to buy at a higher rate than the people who already knew him.

Written down like that, it's obviously impossible. Nobody writes it down.

What he did with the next £200: nothing. He spent it on ingredients and put the hours into two things instead. He messaged every past customer at the four-week mark, since a bottle of hot sauce lasts about that long, which is the whole idea in how to get repeat buyers. And he asked twenty customers whether they knew one person who'd like it.

Three months later he was at 34 sets a month with no ad spend. The margin on those extra sixteen sets, about £80 a month, is roughly what the campaign lost him in a fortnight.

None of that means his ads were badly made. They were fine. The economics were wrong before the first impression served.

Where the £200 goes further

Ranked by what they tend to return for a small seller, best first:

1. Messaging past customers at the right moment. Costs nothing but time, converts at maybe one in three, and the customers are already yours.
2. Asking for referrals properly. One specific ask to one specific person, in the week after their order lands. Asking customers for referrals has the wording, and the traffic it produces converts better than anything you can buy.
3. Fixing the page. A hidden postage cost or a first product nobody wants is costing you more than any campaign will earn.
4. Better photography of your best seller. One product shot properly beats twenty shot badly, and it improves everything downstream including your eventual ads.
5. Product samples to ten strangers. Ten sets of hot sauce cost Tomas about £93 in materials and would have produced feedback, photos and probably four customers.
6. A free-delivery threshold. Raises average order value, which raises your margin per customer, which is the number that decides whether ads ever become possible.

That last one matters more than it looks. Every one of these makes the ads maths better later. Doubling your average order value roughly doubles what you can pay for a click, and that's the difference between a channel that can't work and one that can.

When ads genuinely are the right answer

They exist. All of them are situations where the maths works, not where the seller is ready to grow.

High margin per order. Digital products, services, anything over about £60 with 60%-plus margin. A £180 commission at 70% margin can pay £2 a click at a 1.5% conversion and still profit.

Proven repeat purchase. If you know that a customer buys 3.2 times over a year, your acquisition budget is 3.2 times your first-order margin, not one. This is the single biggest lever and it requires you to have actually measured it, which is why it comes after a year of selling, not before.

A local service with a tight radius. A cleaner, a tutor, a photographer advertising to one city. Small audiences, high intent, and the lifetime value of one client is enormous compared to a click.

One product, one clear buyer, and stock you have to move. Seasonal goods with a deadline, where the alternative to selling at any acquisition cost is unsold stock.

Retargeting people who already visited, with a hard caveat below.

If you're not in one of those, ads are a tax on impatience.

The retargeting trap

Retargeting always looks brilliant in the dashboard, and it's the one place small sellers most often fool themselves.

You show ads to people who already visited your shop. They buy. The platform reports a 9x return. What actually happened is that a large share of those people were coming back anyway, and the ad claimed credit for a sale that was already going to happen.

This isn't a conspiracy, it's how attribution works: the ad was the last thing they saw, so it gets the sale. At small volumes you can't run a holdout test properly, so you can't separate the two.

Use it if you like, and treat the reported return as an upper bound rather than a result. If your total revenue isn't up by roughly what the dashboard claims the ads produced, the ads didn't produce it.

What the platforms are actually good at

Short version, no rankings.

Boosting a post is the worst use of money available and the one the app pushes hardest. It optimises for engagement, which is not purchase, and it will happily spend £6 getting you likes from people who will never buy anything.

Meta ads properly configured, with a purchase objective, need conversion volume to work. Below a certain order count they can't learn. They're strong for visual products with a broad audience.

Google search ads are different in kind: you're buying people who typed a specific intent. Much better conversion, much higher click prices, and only worth it if there's real search volume for what you sell. If you sell something people search for by name, this is usually the highest-intent traffic available, and it's worth finding out whether anyone is searching for it at all before you bid on it.

TikTok is cheap to reach people with and hard to convert directly. Discovery happens there and the purchase usually happens somewhere else, days later, which attribution handles badly.

Local groups and WhatsApp communities are free and outperform all of the above for a lot of sellers in Lagos, Nairobi and Manila, where a recommendation in a community group does what a paid impression can't.

The cost stack nobody counts

If you do run ads, the ad spend is not the only new cost, and this is the part where I'll be plain about the tool I work on.

Cold traffic from an ad won't send you a bank transfer. A stranger who's never heard of you wants to pay by card, right then. So running ads usually means adding card payments, and on Sailo that's 1–3% of the goods on each card order, depending on your plan, on top of whatever Stripe charges you. That's three new costs at once: the ads, the subscription, and the processing.

On Tomas's seven orders, the 1–3% would have been about 10p a set. Trivial. The $49 a month is not trivial against £35 of margin, and it doesn't stop when the campaign does.

There's no mobile money rail either, so if you're advertising into a market where buyers expect M-Pesa or GCash, the shop will carry the catalogue and the order and you'll still be confirming the money yourself from your own notifications.

The point isn't that the plan is expensive. It's that "start running ads" is never one decision. It's three, and the recurring one outlasts the campaign.

Do this before you spend anything

Two sums and one count, tonight.

1. Margin on one order, with your own time in it at a real hourly number.
2. Your current conversion rate, from whatever traffic you already have. Orders divided by visitors, over the longest period you have. If you don't have enough traffic for that number to be stable, what conversion rate you should expect explains why the published benchmarks won't help you either.
3. Multiply them. That's the most you can pay for a click and break even. Halve it. That's your actual ceiling.

If that number is under about 20p, close the ads manager and go and message ten past customers instead. If it's over 50p, you have a channel worth testing, and the test should be £50 over two weeks with the purchase objective, not £6 a day on a boosted post, and you should decide the kill criteria before you start.