Few Buyers, Big Orders: Is SEO Worth Buying?

A Singapore machine shop with forty customers and a seven-figure turnover gets pitched the same monthly content programme as a nail salon. The pitch is wrong, and not because the machine shop is too small. It is wrong because the category has almost nothing left to write about after six weeks.

Here is the test we would run before spending anything, with the arithmetic shown. It is two questions long, and one of them disqualifies our own entry package for some of the firms that ask us about it.

The two numbers that decide this

Not traffic. Not keyword volume. These two:

1. How many distinct questions does your category actually contain? Questions, not keywords. "cnc machining singapore" and "cnc machining in singapore" are one question.
2. What is the gross margin on one new account in its first year? Margin, not revenue, and one account, not your best year.

The first tells you whether you are buying a programme or a project. The second tells you whether you can afford either. Most arguments between a Singapore SME and its agency in month five are really a disagreement about one of these two numbers that nobody wrote down in month one.

Step one: count the questions, not the keywords

Do this with your own trade and it takes an evening.

We did it for Singapore precision engineering on 8 October 2026. Google Autocomplete returned 47 suggestions across three Singapore-anchored seeds, plus 30 more on a cost seed. Forty-seven suggestions sounds like a content calendar. It is not. Collapse them into the questions underneath and you get something like this:

1. Who machines parts in Singapore?
2. Is this particular firm any good?
3. What is precision machining, or sheet metal fabrication?
4. What drives a machining price?
5. How do I estimate a machining cost myself?
6. How do I bring a machining cost down?
7. What tolerance can you hold?
8. Will you take one part, or a small batch?
9. What are you certified to?
10. Where are you, and can you deliver to me?
11. How do I send a drawing, and when do I get a quote?
12. Do you have a job for me?

Twelve. Four of those came out of the suggestion data, five came from reading what 10 Singapore suppliers' pages leave out, and one is not a buyer at all but still needs a page.

Twelve real questions is a good website. It is not twelve months of articles. Count yours honestly and the number will land somewhere between eight and thirty for most specialist trades in Singapore — and somewhere past two hundred for a consumer category with towns, services and price points to multiply.

Step two: what one account has to be worth

Now the money, using our own published price so you can check the sum.

Our SEO, AEO and GEO programme starts from S$500/month as the live page states today, with a three-month minimum. Twelve months of that is S$6,000.

Divide S$6,000 by the first-year gross margin on one new account and you get the number of accounts the work has to win before it has paid for itself:

First-year gross margin per new account	New accounts needed in year one
S$2,000	3
S$6,000	1
S$20,000	1 covers three years
S$50,000	1 covers eight years

Read the top row and the bottom row as two different businesses, because they are. A firm making S$50,000 of margin on a new account needs the work to produce a single customer in eight years to be ahead on cash. A firm making S$2,000 needs three a year, every year, from a category with twelve questions in it.

We would take the first firm's brief. We would tell the second firm to keep its money, and we mean that literally — not as a negotiating position.

One number is missing from that table on purpose, and it is the honest gap in all of this: we cannot tell you how many accounts the work will produce. Nobody can. Google's own guidance on whether you need an SEO puts it bluntly — "No one can guarantee a #1 ranking on Google." The table tells you the bar. It does not tell you that the bar gets cleared.

Where our own package fails this test

Our entry tier buys 10 articles a month. A category with twelve questions is finished in about six weeks.

So for a specialist Singapore supplier, month three of a monthly content programme is the month it starts producing pages nobody asked for. You get articles on adjacent topics, articles aimed at other countries, articles that repeat earlier articles with a different noun in the title. Every agency in this industry knows that drift. Volume packages are sold by the month because retainers are easier to run than projects, not because every category needs them.

If your count comes back under about twenty questions, what you need is a project: write the twelve pages properly, fix the capability page, set up measurement, then stop and let it run. That is a smaller invoice for us than a year of retainer, and it is the right answer more often than our pricing page implies.

Google says something adjacent and worth repeating: "If you run a small local business, you can probably do much of the work yourself." Twelve pages of plain answers about your own shop is work you can do, if you have someone who writes clearly and the discipline to finish.

The read we would rather sell you first

Before any of this, there is a cheaper purchase: finding out whether the demand is there at all.

A feasibility read is a few days of work. Pull the live suggestions for your category, count the questions, search each one and count how many of the first results are operators like you, and check whether any of it is rising or dying. The output is a yes, a no, or a "not this way". It costs a fraction of a year's retainer and it sometimes ends the conversation.

That is the same test we set out for people who have not started trading yet in before you incorporate, check if anyone is searching. It applies just as well to a thirty-year-old engineering firm deciding whether to put money into search for the first time.

Two cases where our answer is no

Your buyers are a known list. If you could write down every company in Singapore that could place your kind of order, and you already know most of them by name, search is not your growth channel. It is your reference check — the page people read after someone gives them your name. Fix that page, then spend the rest on the sales call. We measured the reference-check pattern in the suggestion data: eleven of forty-seven searches named a specific firm, and eight of those wanted its reviews.

Your margin per account cannot clear the annual fee. If one new account produces less gross margin than a year of the work costs, the arithmetic needs two or three wins a year from a category that might not contain them. We set out other versions of this in three times SEO is the wrong thing to buy.

There is a third case that is not a no but a wait: you cannot yet tell a search enquiry from a referral. Buy the attribution before the traffic, or you will spend a year arguing about whose customer that was. Our note on proving search pays covers the setup.

What this costs, in full

The three-month minimum means S$1,500 leaves your account before you can judge anything, and three months is not long enough to judge search work. Our own realistic timeline says as much. So the real commitment you are weighing is closer to nine or twelve months than three, and you should price it that way in your own head even though the contract does not.

The other cost is attention. Twelve good pages about your own shop need your engineers to check the tolerances, your sales lead to confirm what you actually quote, and someone to answer our questions for a few weeks. Firms that cannot spare that get generic pages, which are worse than no pages.

And the cost to us of publishing this: it talks a share of the people reading it out of a retainer and into a one-off project, or out of buying anything at all. We would rather carry that than spend month nine defending output that the category never needed.

One next step

Send us your trade and your main category phrase. We will pull the live suggestions, collapse them into a question count, and tell you whether what you need is a programme, a project, or nothing from us yet. That is a free visibility check through our contact page.

If the count comes back large and the margin clears, our programme starts from S$500/month as the live page states today. We do not promise rankings, AI citations or leads.

Prices described here are the prices on our live services page on 8 October 2026 and may change. This is commercial and marketing guidance, not legal, tax or financial advice.

FAQ

How do I know if my market is too small for SEO?

Count the distinct questions your buyers ask, not the keywords. If the whole category fits inside two months of a content package's output, you are buying a project, not a programme. Then check whether one new account's first-year gross margin covers a year of the fee on its own.

What is a reasonable number of articles a month for a specialist B2B firm?

Fewer than most packages sell. If your category holds about a dozen real questions, the right plan is to answer those twelve properly and stop, then revisit when your market or your capability changes. Paying monthly for volume in a thin category produces pages nobody searched for.

Can an agency tell me not to hire it?

A good one can and sometimes should. The honest version is a feasibility read first: count the demand, count who holds the results, and report the answer even when the answer ends the sale. Google's own guidance notes that a small local business can probably do much of the work itself.

How many new customers does S$500 a month have to win?

That depends entirely on your margin, which is why the sum matters. At S$6,000 a year, one account with S$6,000 of first-year gross margin covers it; an account worth S$2,000 means you need three a year. We cannot tell you whether the work will win any — nobody can promise that.

Is a three-month minimum long enough to judge search work?

No, and we say so. Three months tells you whether the work is being done, not whether it pays. Judge the output in month three and the result somewhere between month nine and month twelve, and decide in advance which numbers you will look at.

What should I spend on instead if the answer is no?

Usually the page people reach after someone gives them your name, and the route from that page to a quote. In our October 2026 sample of Singapore machining sites, only one of ten had a form that would accept a drawing. That is a cheaper fix than any retainer and it works on demand you already have.