Train people for long enough and you notice the mistakes are not random. The same five turn up in almost every cohort, and most of them are neither carelessness nor lack of intelligence.
They are logical mistakes. Somebody taught this person something that was true, the thing changed, and nobody went back and told them.
Here they are in order of cost, cheapest first.
1. Learning tools before problems
Beginners start with a course on a tool: Meta Ads Manager, Google Analytics, an SEO platform. They finish able to click every button and unable to decide anything.
The trouble is that tools change constantly and problems do not. Someone who understands how to define the right audience will learn any tool in two days. The reverse is not true.
The fix is not to avoid tools. It is to learn them while solving a real problem, even a tiny project of your own.
2. Believing targeting settings are constraints
This is the most common technical mistake I see, and the most damaging, because it makes someone misread their results for months.
A beginner sets an interest and an age range and assumes those are closed boundaries. Then they see the ad reaching people outside that, and conclude there is a bug or that the account is broken.
The truth is written in Meta's own documentation. When Advantage+ audience is enabled, Meta's developer documentation states that the delivery system resets age_min and age_max to default values, that advertisers can pass age_min values between 18 and 25 only, and that age_max is fixed at 65.
It also names what is not expanded: location constraints, minimum age, language, and custom audience exclusions. Those are the real boundaries.
So interests are a signal and geography and language are constraints. Understanding that distinction on day one saves you a year of wrong conclusions.
3. Never setting up measurement, then losing the history
This is the mistake whose cost appears very late, at which point there is no way back.
A beginner installs Google Analytics and leaves it. But Analytics does not know what matters to you. Google defines a key event as an event that measures an action particularly important to the success of your business, and any collected event becomes one only when you mark it. Without that step the property records activity and measures nothing.
Worse is the data retention setting. Under Google's data retention settings, event data on a standard property can be kept for either 2 months or 14 months, with the longer periods reserved for Analytics 360. User-level data is 2 or 14 months.
The practical consequence is specific. That setting governs explorations and funnel reports, so a property left on the shorter period cannot run a twelve-month exploration however long it has been collecting. Standard aggregated reports are unaffected, which is exactly why the problem stays hidden until the first time somebody tries to look back a year and finds nothing.
Check that setting on the first day you touch an account, not the day you need the history. It is not retroactive.
4. Judging a campaign far too early
One day, a hundred pounds, and a verdict that "ads do not work". This happens constantly.
The problem is not impatience. It is that the person does not know when a number becomes safe to judge at all, and that is calculable from how much data you have rather than a matter of feel.
The rule I give beginners: if the outcome you are measuring has happened fewer than twenty times, you are not comparing anything yet. And before spending at all there are conditions that should come first, which I wrote about in when to start spending on ads.
5. Copying a tactic without its context
Someone sees a success story, copies it precisely, watches it fail, and concludes the person was lying.
Usually they were not. They simply did not mention the part that made the tactic work: an existing customer list, a known brand, or a product with a margin nothing like yours.
Tactics do not transfer. Principles do. When you see something working, ask why it works before asking how to do it.
Why these five in particular
Look at them together and something is shared. Every one happens when a person takes a number or a setting as final fact without asking where it came from and how it is calculated.
The interest you thought was a boundary, the history you assumed was kept, the result you assumed was enough to judge. Three of the five are the same error wearing three costumes.
So the practical advice is not to memorise this list. It is to build one habit: whenever you see a number in any dashboard, ask two questions. Who calculated it, and over what period?
That habit alone protects you from most of the mistakes that are not on this list either.
A checklist for your first week
- Open the data retention setting in Analytics and see whether it says 2 months or 14.
- Confirm at least one key event is actually marked, not merely collected.
- Write down which constraints the platform genuinely respects on your account: geography, language, exclusions.
- Decide before any campaign how many times the outcome must occur before you will judge it.
- When you copy a tactic from someone, write one line on what they have that you do not.
The sixth mistake, which is not on the list
Choosing the wrong source to learn from in the first place, which quietly produces all five above.
I wrote about the order of learning in what to learn first in marketing, about why courses alone change nothing in why courses do not change your life, and about the step after the videos in what to do after learning from videos.
And if you are genuinely at the point of choosing a paid course, there are practical questions worth asking before you pay. Our colleagues at UCourses set them out in six questions for choosing a digital marketing course.
One last thing
None of these mistakes means you are unsuited to the field. They mean you learned from sources that explained how without explaining why and when.
If you take one thing from this, make it the second mistake. Know what the platform treats as a hard constraint and what it treats as a suggestion. That distinction alone changes how you read every number afterwards.