Facebook ads for a small business: budget, setup and the first month

Facebook and Instagram ads cost the budget you set, because Meta sells each impression in an auction rather than from a price list. The price per impression moves. In two 2026 benchmarks, 1,000 impressions cost €9–€11 in the UK, €4–€9 across Germany, France, Italy and Spain, and €14–€18 in the US.

Each time there is room to show someone an ad, Meta holds an auction and picks the ad with the highest total value: the bid, Meta's estimate of how likely this person is to act, and an ad-quality score that counts, among other things, how often people hide the ad (how Meta's ad auction works).

A more relevant ad can beat a higher bid, which is why two businesses aiming at the same town pay different prices.

Cost guides that two UK agencies published in May and June 2026 run wider. Between them they quote £5–£18 per 1,000 impressions (€6–€21), £0.45–£1.50 a click (€0.52–€1.75) and £8 to more than £50 a lead, roughly €9–€58.

Industry moves the price of a lead further than country does. In the American firm's 2025 benchmarks, which cover 726 lead campaigns, the median lead cost $3.16 (€2.71) for restaurants, $41.26 (€35.41) for home improvement and $76.71 (€65.83) for dentists. Twenty-four times the restaurant figure.

For a local service business I would start at €20–€30 a day, €600–€900 for the first month, on one campaign. At the US figures of €24–€40 a lead, that buys roughly 15 to 37 enquiries, enough to see which ad pulls and whether the enquiries become customers. A UK agency's 2026 cost guide starts local businesses at £800–£2,000 a month (about €930–€2,330).

Neither budget gets an ad set out of learning. Meta calls that “Learning limited”, says it is not a penalty, and suggests fewer ad sets, a wider audience, more budget or a more frequent event (About learning limited). Expect bigger swings. Read a week of figures, never a day.

The optimisation event should match the result you want to measure. Meta discusses the volume needed for conversion optimisation in its landing page view guidance. With a small budget, compare qualified enquiries and their cost as well as visits. A cheaper visit is useful only if it contributes to the campaign goal.

An online shop needs more, since a sale asks more of a visitor than an enquiry does; the same UK guide starts conversion campaigns at £1,500 a month (about €1,750). With less, optimise for add to cart, Meta's own example of a more frequent event.

Boost a post that has already drawn comments and shares, to put it in front of people who don't follow you yet. Build a campaign in Ads Manager when you want enquiries or sales you can count: a boost has fewer targeting and creative options, while a campaign can optimise for a conversion on your own website.

Where you press the button matters. Since February 2024, a boost paid for in the Facebook app on an iPhone or iPad goes through Apple, which adds a service fee of 30% of the ad payment (Meta on the Apple fee), and Instagram's iPhone app charges it too. A €100 boost from the phone costs €130. From a browser, €100.

I use boosts for one job: pushing a post that already works to more people nearby. Which posts earn that is the subject of social media marketing for a small business.

Set up one campaign with the Leads or Sales objective, one ad set limited only by location, and three to five ads. Before launch, install the Meta Pixel and the Conversions API so Meta can see each enquiry or sale, and in the UK and the EU get cookie consent before the pixel loads.

Ads Manager offers six objectives: Awareness, Traffic, Engagement, Leads, App promotion and Sales (Meta's objective guide). A service business wants Leads, which can collect the enquiry as a form, a message or a call. A shop wants Sales. Traffic optimises for clicks, the one result on that list you can buy cheaply and still lose money on.

Leads and Sales campaigns open in Meta's Advantage+ setup, which automates the budget, the audience and the placements (what turns Advantage+ on and off). I leave the placements alone, so the same ads reach Instagram's Reels and Stories, and constrain one thing: where the customers are.

In Advantage+ audience, locations, minimum age, languages and excluded custom audiences are controls Meta will not cross. Age, gender and interests are only suggestions (audience controls and suggestions). So a dentist in Leeds draws a radius around the practice, adds no interests and leaves “Reach more people likely to respond to your ads” unticked, because that option lets delivery go past the chosen locations.

An ad set can only optimise for what it can see. The Meta Pixel is a snippet on your site that reports actions such as a submitted form or a purchase back to Meta; browser restrictions and network errors lose some of those reports, so Meta recommends sending the same events from your server through the Conversions API too, removing duplicates and aiming for server events to match 75% of the pixel's (Conversions API best practices). The pixel sets cookies, and in the UK the ICO's guidance on PECR says non-essential cookies cannot be set before the visitor consents; Article 5(3) of the ePrivacy Directive sets the same condition across the EU.

The Leads objective can take the enquiry on your own page or in an instant form inside Facebook or Instagram. The default form, More volume, is built for speed. Higher intent adds a review screen that Meta says keeps out people “only marginally interested”, and it runs only in the Facebook and Instagram feeds on phones (instant form types). Where every lead means a phone call, I would pay more per lead for Higher intent or a landing page before I took the default. Calls are the expensive part.

Make each ad vertical first, 9:16 at 1440 × 2560 pixels: Meta recommends that shape for Reels and Stories on both apps and for video in Instagram's feed. Then crop a 4:5 version at 1440 × 1800 for Facebook's feed and for still images in Instagram's. On the vertical one, keep text out of the top 14% and the bottom 35%.

Those bands, plus 6% on each side, are where the profile icon, the caption and the button sit, and Meta's Ads Guide asks for them to stay clear of text and logos.

For Facebook's feed, Meta suggests 50–150 characters of text and a 27-character headline (Facebook Feed specs).

I start with three to five ads, each with a different reason to get in touch: the price, a before and after, the question customers always ask on the phone, an offer with an end date. Adding an ad restarts learning, as do changes to targeting, creative or the optimisation event (significant edits), so new ads go in as a weekly batch.

Cost per lead is the price of a click divided by the share of visitors who enquire, and the landing page decides the second half. At about €1.30 a click, close to the £1.11 median a UK agency gives for 2026, a page that turns one visitor in 20 into an enquiry costs €26 a lead. At one in 50, the same ad costs €65.

The page also feeds the auction. For an ad optimised for leads, the action Meta is estimating happens on your page, so a page that converts better raises that estimate and brings the ad set nearer its 50 a week.

The pages I would send Meta traffic to have four things in common:

• The offer from the ad, repeated in the first screen in the same words.
• One action, as a form of three or four fields or a phone number people can tap.
• A cookie banner that leaves the form visible.
• Speed on a phone. A click on a Meta ad opens the page inside Facebook's or Instagram's own browser, so test it there, on an actual phone, before the campaign starts.

That is why our ads management builds a landing page for each campaign before any ad goes live. Off the page, call every lead the day it arrives and note which ad brought it; that note, kept for a month, is what later tells the ads that bring customers from the ads that bring only forms.

Spend the week before launch on tracking and the page, leave the ad set alone for its first seven days, switch off the weakest ad in the second week and judge the month on cost per customer. Three numbers carry the verdict: cost per result, link click-through rate and the share of leads that became customers.

1. Before launch: a test event from the pixel and the Conversions API in Events Manager, a working consent banner, the page checked on two phones, a figure for what a customer is worth, and a look at your competitors' ads in the Meta Ad Library.
2. Days 1 to 7: one campaign, one ad set, three to five ads, €20–€30 a day. Change nothing.
3. Days 8 to 14: switch off the ad with the weakest link click-through rate.
4. Days 15 to 30: add new ads if needed, fix the page if landing page views are not becoming enquiries, and work out the cost per customer. If the sums work, raise the budget a fifth at a time. Meta's own example: $100 to $101 is unlikely to restart learning, $100 to $1,000 may (significant edits).

Cost per result is Meta's figure for what each lead or sale cost; set it beside the margin one customer brings you, not beside a benchmark. Link click-through rate says whether an ad stops the scroll. The 2025 US benchmarks put the median click-through rate at 2.59% for lead campaigns and 1.71% for traffic campaigns, but the comparison that counts is inside your own ad set, where an ad far below its siblings is the one to replace. The third number is not in Ads Manager at all. It is how many leads became paying customers, which only your records hold, and it stops a campaign full of cheap leads from looking better than it is.

Ads Manager counts generously: under Meta's standard attribution, a lead or sale can be credited to the ad up to seven days after a click, or a day after someone merely saw it, even if that person then arrived through a Google search (attribution settings).

If the month ends with 20 leads and two customers, the number to work on is that 10%, and a bigger budget will not move it.