Back to journal
Advertising5 min read

How much should you spend on Meta ads before you judge them?

By · Published · Updated

Person using smartphone for social media
Advertising

Meta ads need time and budget to learn. Here is a practical framework for setting a sensible budget, judging results fairly and knowing when to stop.

Why most people judge Meta ads too early

The most common mistake I see with Meta ads is judging them too early. A business runs a campaign for a week, spends $200, gets no enquiries, and decides that Facebook advertising does not work for its type of business. In most cases the ads were never given a fair chance. The campaign did not have enough time or money to find out who was likely to respond.

I understand the instinct. $200 is real money for a small business, and a quiet week feels like a verdict. But a week of data at that spend is not enough to tell you anything reliable. It is like judging a new employee after their first morning. You might be right, but you have not seen enough to know.

Before you decide whether Meta works for you, decide what a fair test looks like. Set the budget, set the run time, and agree in advance on the measure you will use to judge the result. Writing this down before the campaign starts stops you from moving the goalposts when the early numbers look uncomfortable.

What the learning phase means for your budget

Meta's advertising system is a machine learning model. It needs data to work out which people are most likely to respond to your ad. When a new campaign or ad set starts, it spends its early days in a learning phase. It tests your creative against different audiences, gathers signals and works out what gets results. During this phase your cost per result is usually higher and less steady than it will be later.

The rule of thumb we use is at least 50 conversions per week per ad set before the system leaves the learning phase. A conversion here means the action you care about, such as an enquiry form submission or a phone call. The number matters because a model with only a handful of results each week is guessing. It needs enough outcomes to learn from.

If your target cost per lead is $20, then 50 leads a week means about $1,000 a week, or roughly $4,000 a month for a single ad set. That is more than most small businesses want to spend on a test. This is why the budget question matters so much, and why a small budget so often produces a verdict before the system has had a real chance to work.

How much to spend each month

For a Meta campaign that can be judged fairly, we recommend a minimum ad spend of $1,500 a month. Where the budget allows, $2,000 or more is better. Below that level the system struggles to gather enough signal to leave the learning phase in a sensible time, and most of what you see is noise.

That figure is not a promise of results. It is the smallest amount that gives the campaign a reasonable chance to produce a clear answer. If you can only afford $300 a month, Meta may still be worth a try, but treat it as a small experiment. Do not expect a conclusion from it, and do not pay someone to manage a campaign of that size on the hope of a result.

Keep the management fee separate from the ad spend in your head. The ad budget goes to Meta. The fee pays for setup, creative, testing and reporting. Ask whoever runs your campaign to show you how the spend is split and what you are paying for in each part, and check that the answer makes sense before you agree to anything.

How long to run a test before deciding

For a new campaign, I recommend a minimum run time of six weeks before you make a judgement. That is long enough for the system to learn, for the team to make useful changes, and for you to see results that are steady rather than random.

Week one is for learning. The system is testing and gathering data, and the results will look uneven. Weeks two and three are for optimisation. We pause the ads that are not performing, put more budget behind the ones that are, and refine the audiences. Weeks four to six are when the campaign settles, and that is when you can see the true cost per lead.

Judging in week one is a bit like judging a recipe after tasting only the batter. Some things are worth knowing early, such as whether the ads are being shown at all, or whether people are clicking through. But the verdict on whether the campaign pays for itself needs the full run.

Measure leads and cost per lead

What should you judge? Not clicks, not impressions and not reach. Those numbers tell you that people saw the ad or tapped on it. They do not tell you whether the business is growing. Judge enquiries, bookings or calls, and the cost of each one.

A campaign that produces 15 leads at $22 each is working, even if the click-through rate looks modest. A campaign that produces 200 clicks and no leads is not working, even if the click-through rate looks excellent. The only question that matters is whether the ads are bringing in business at a cost that makes sense for your margins.

Make sure your tracking is set up before the campaign starts. If you cannot see which enquiries came from the ads, you cannot judge the ads. A simple approach is a dedicated phone number or a form with a source field, plus a monthly check that matches the leads to the work that actually turned into paid jobs.

When to stop a campaign

Does a six week test mean you should keep spending regardless? No. If a campaign is clearly failing, stop it. In my view the worst outcome is spending money on a campaign that is not working, without a clear reason and a plan to fix it.

After two weeks and $1,500 with zero leads, something is wrong. It is usually one of three things: the offer, the landing page or the targeting. The offer may not be compelling enough to act on. The landing page may be slow, confusing or missing a clear next step. The targeting may be too narrow or aimed at the wrong people. Each of these can be diagnosed and changed.

We will tell you when a campaign is not working, and we will tell you what we think is wrong. We do not keep taking management fees on a failing campaign without a diagnosis and a plan. If the diagnosis says Meta is not the right channel for your business, that is a fair answer. It is better to learn that after one honest test than after six months of hope.

Why cost per lead should fall over time

A well-run campaign should get steadily cheaper per lead over about three months. The system learns which people respond, the creative matures as you learn which messages land, and weak ads are replaced with stronger ones.

If the cost per lead is not falling after several weeks of proper management, something needs to change. That may be the creative, the audience, the offer or the way leads are followed up. Changing one thing at a time makes it much easier to see what helped.

The businesses that do well with Meta treat it as an ongoing process rather than a one-off experiment. They run campaigns, measure them, refine them and run them again. Meta is one channel among several. Used well, it can be a steady source of enquiries. Used as a quick test with a tiny budget, it will usually disappoint.