What a Facebook ad agency actually does all day

The actual job, in order of how much it matters

A Facebook ad agency spends most of its useful time on creative and measurement. Budget management, the thing people imagine they are paying for, is maybe fifth on the list, and Meta's automation has been eating that job for years.

1. 1Creative production and testing. New concepts, new hooks, new formats, running against each other. This is where the results come from and where most of the hours go.
2. 2Measurement. Making sure the Meta Pixel and conversions API are firing the right events, so the platform optimizes towards something real.
3. 3Offer and landing page. The ad sells the click; the page sells the thing. Agencies that ignore the page are doing half a job.
4. 4Audience and structure. Less hands-on than it used to be, but still where waste hides.
5. 5Budget and bidding. Largely automated now. Important, but not where your money earns its keep.

What your fee should actually buy

Ask an agency to describe a normal week. The good ones talk about what they made, what they killed, and what they learned. The weak ones talk about "monitoring" and "optimizing", two words that can mean anything and often mean logging in on Friday.

A reasonable month includes new creative in the account, at least one thing switched off for underperforming, a clear read on which concept is winning and why, and a report that explains what changed rather than just what happened.

If the monthly report could have been written without opening the account, it probably was.

Signs you are paying for busywork

• The same three ads have been running for four months. Creative fatigue is real, and this is what it looks like from the outside.
• Reporting leads with reach and engagement. Both are easy to buy and neither pays anyone.
• No one has ever mentioned your landing page. The click is half the transaction; ignoring the other half is a choice.
• Lead volume is up and nobody knows if the leads are any good. Volume without quality is a reporting problem dressed up as success.

That last one is the one to watch. It is very easy to make Facebook produce more leads. It is much harder to make it produce leads your sales team is pleased to receive, and the gap between those two things is where most disappointing paid social lives.

When you should not be running Facebook ads at all

Plainly: if you cannot feed the channel creative, do not start. Paid social consumes ideas. An agency can produce them, but that has to be in the scope and in the budget, and if it is not, you are buying a slow decline with a monthly invoice attached.

And if your website undercuts everything before the visitor gets to the offer, fix that first. We have watched traffic and rankings look perfectly healthy while enquiries stayed flat, purely because the site buried the offer, loaded slowly, and made people hunt for the next step. Fixing the website did more for growth than any new campaign would have.

Same principle as always: if the money is leaking somewhere upstream, more traffic just fills the bucket faster. Start where the leak is, then come back to the ads.

Measurement is the half nobody audits

Creative gets all the attention because it is visible. Measurement gets none, because it is invisible until it is catastrophically wrong, and on Meta, wrong measurement does not look like an error. It looks like a campaign that is working.

The platform optimizes towards whatever event you nominate. If that event is misconfigured, firing on page load, or counting the same person twice, Meta will confidently spend your budget pursuing it. The reports will look healthy. The bank account will disagree.

• Check the [Meta Pixel](https://developers.facebook.com/docs/meta-pixel/) is firing the event you think it is. Not that it is installed, that the specific conversion event is correct.
• Check for duplicates. Pixel plus conversions API without proper deduplication means every conversion counted twice, and every cost-per-result halved on paper.
• Check the attribution window. A 7-day click window and a 1-day window tell very different stories about the same campaign. Neither is wrong; knowing which one you are reading is the point.
• Check what the CRM says. If Meta claims 40 leads and your inbox has 12, believe the inbox.

That last check takes five minutes and almost nobody does it. Compare what the platform reports with what your team actually received last month. If the two numbers disagree, everything built on top of them (the optimization, the budget decisions, the verdict on whether Meta "works" for you) is built on sand.

This is why we will not judge a paid social account until the measurement is trustworthy. Not out of pedantry, but because the alternative is having a confident opinion about a number that is not real. Fix the tracking, then argue about the creative.

The page after the click

Meta traffic arrives cold and impatient. Nobody went looking for you; you interrupted them. That changes what the landing page has to do, and it is the half of the job most agencies quietly leave to you.

• Assume zero context. They saw an ad for a few seconds. Explain yourself immediately, in the same language the ad used.
• Match the promise. If the ad said one thing and the page says another, you have paid for a click and bought a bounce.
• Be fast. Core Web Vitals are a fair bar, and cold social traffic is the least forgiving audience you will ever pay for.
• One action. Not a navigation bar, a newsletter popup and three competing buttons.

Test it the way your customers will experience it: on your phone, on mobile data, standing outside. Most people have only ever seen their landing page on a desktop in the office, and are genuinely startled by how slow and cluttered the real thing feels.

If the page cannot hold that traffic, no amount of creative testing will rescue the campaign. Fix the website first, it is cheaper than the media budget you are about to waste on it.

Frequently asked questions