Micro-budget marketing: what works under $100 a month?

Below about $100 a month, micro-budget marketing stops being a budget-allocation problem and becomes a routine-design problem. There is no meaningful way to split that money across platforms, so the question "where should the budget go?" has almost no leverage in it. The question with leverage is "what do I do every week?" — and the answer to that can be worth a great deal more than a hundred dollars. Here is a complete monthly system that fits inside $100, usually inside $50. (Disclosure: published by BidSurvivor, which appears in week three, labelled.)

Why splitting the money is the classic mistake

Take $100 and divide it across Google, Meta and LinkedIn and you have created three rounding errors. WordStream's benchmark data puts the all-industry average cost per click at roughly $5.42, which makes a $33 slice of a search budget about six clicks. Six clicks tell you nothing: at any plausible conversion rate the expected number of conversions is zero, and zero is exactly what you would see if the campaign were perfect and if it were broken.

Every large platform also has an implicit minimum before its optimisation does anything useful — a learning phase measured in conversions, not dollars — and a micro-budget is below all of them simultaneously. The full arithmetic is in how much it costs to advertise a startup.

So the rule underneath everything that follows: at this budget, money buys tests, not campaigns, and a test needs all of the money at once to mean anything.

The monthly routine

Week 1 — free visibility maintenance ($0). One batch of new directory listings, or one pass refreshing stale ones: dead links, an old tagline, a screenshot from two versions ago. This is unglamorous and it compounds, because a listing that ranks keeps sending people for years. The set worth doing is in free advertising for startups.

Week 2 — one piece of writing ($0). One article that answers a question a buyer actually types. Not four thin posts; one that is specific enough to be wrong. Comparisons, prices, procedures and post-mortems still earn clicks when generic overviews no longer do. This is the slowest line in the system and the one that eventually makes the others unnecessary.

Week 3 — one measured test ($0–50). Alternate months. In one, take a free slot on our board — the first brand into an empty twelve-hour slot pays $0, no account and no card, every visitor starts with $100 of house credit, and each brand's opens and click-throughs are published — and use it as a message-testing lab rather than a traffic source. In the other, buy the smallest warm-audience placement in advertising: roughly $50 in a niche newsletter (how to spend exactly fifty dollars). One tests the sentence; the other tests the audience.

Week 4 — community presence ($0). A handful of genuinely useful answers in the places your users complain. The slowest-burning channel in the system, and the one that produces the users who stay longest.

Total: $0–50 in most months, never over $100, and roughly six hours of work. The six hours are the real cost, and the reason most people fail at this is not the budget.

What the money must do when you finally spend it

One test at a time, at full size, with a question written down before it runs.

A useful test at this budget has three properties. It has a single variable — one pitch line, one audience, one landing page, changed against a known baseline. It has a pre-registered success threshold, written before the data arrives, because a threshold chosen afterwards is a story. And it produces a number you can carry into the next channel: a click-through rate on a headline, a signup rate on a page. Those two numbers make every future quote from any seller legible in ten seconds — price divided by (visitors × your rate) is cost per customer, and you can compare that against anything.

That portability is the whole reason to run cheap tests first. The channels that cost real money get cheaper once you know your own numbers, and they stay ruinous if you buy them to find those numbers out.

What to measure, and what to ignore

One number per channel, checked monthly, twenty minutes in a spreadsheet:

• Writing: search impressions, then clicks.
• Directories and communities: referral visits.
• Slot tests: click-through rate on the pitch line.
• Newsletters: signups per placement, and their opens four weeks later.

Ignore followers, impressions on a feed, and anything a platform reports that you cannot trace to a visit. That distinction is worth being pedantic about, and we were pedantic about it in vanity metrics vs real clicks.

After three months, one channel will be visibly ahead. That is where the next hundred dollars goes — not spread evenly, all of it — and doubling down on one working channel is how a micro-budget graduates into a small one.

The honest limits of this system

It is slow. Three of the four weeks produce nothing measurable in month one, and that is not a flaw to be optimised away — it is what compounding looks like from the inside. If you need customers this week for a bill due next week, this system will not deliver them, and no honest version of it would.

It is also fragile in one specific way: it depends entirely on you doing the six hours. A hundred dollars a month is a decision you make once. Six hours a month is a decision you make again every month, and the failure mode of every micro-budget plan we have watched is not a bad channel choice, it is week three going missing in April and never coming back.

The trick underneath it all is that three of the four weeks cost nothing and compound, while the fourth buys information rather than volume. The money is a garnish. The routine is the meal.