Why learning a live channel differs from studying it
The certifications and self-study guide structures self-study around documentation, practice accounts and a weekly routine, away from live money. That is the right way to learn a platform's vocabulary and its intended use. The digital marketing skills map says a marketer with the transferable method skills can move between disciplines by learning the platform layer, which is the fast part. Both are true, and neither covers the moment when a channel lands on your desk because its owner left, the agency was let go or the business decided to try it, and it is already running.
Three things change when the channel is live. Every change has consequences: it spends money, reaches customers or alters the numbers someone else reports. The channel has a history you did not see, so settings that look wrong may be deliberate and settings that look harmless may be holding something up. And the business keeps asking for results while you learn, so the temptation is to change things quickly to look in control.
The guide to making the most of a first digital marketing role sets rules for working safely in live accounts as a newcomer to marketing. The tutorial on taking over an existing marketing programme inventories a whole programme someone inherits. The guide to being a marketing team of one decides which channels a sole marketer runs at all. This tutorial sits between them: an experienced marketer, in their current job, adding one channel they have never run. If the channel is part of a deliberate move into another specialism, the guide to switching specialism within marketing covers choosing and testing the move itself. The manager's side, leading a specialist in a channel or discipline the manager has never run, is covered in the guide to leading specialists whose craft the manager does not share.
The skills hub notes that professional bodies such as the Chartered Institute of Marketing frame competence as knowledge plus demonstrated application. A live channel is where the application happens, which is why it is worth learning carefully rather than quickly.
Before you start
- Access at the right level. Ask for read access first if the team allows it, and edit rights when you are ready to change something. If your team keeps an onboarding pack for new marketers, read its access and rules sections.
- The team's procedures for the channel. The paid campaign launch SOP and the campaign launch QA checklist are the safety net for paid channels; find the equivalent for others, or ask who approves what.
- The platform's own section of this site and its documentation, such as the Google Ads section or the email marketing section, for the mechanics this tutorial deliberately leaves out.
- Any handover record the previous owner left, and their contact details if they are still reachable.
- An agreement with your manager about what you own from when, what decisions still need approval, and who will review your work while you learn.
- A document for the learning log, set up before you change anything.
- Time for structured study alongside the live work. If the channel has a path in the site's academy, the tutorial on building a learning plan from the academy paths shows how to schedule it around the gap.
The steps
Read the channel's history before touching it
Spend the first days reading, not changing. Read the change history if the platform keeps one, the past reports, the plans and briefs, the invoices and any contracts. Note what was changed, when and, where you can tell, why. Look at how the account or programme is structured and how things are named, and find the team's Naming ConventionAn agreed pattern for naming campaigns, ads, events and tracking parameters so that reports can group them reliably. if it has one. Write down every setting you do not understand and every one that looks wrong; do not fix any of them yet. Many will turn out to be deliberate.
Write down the baseline and how it is measured
Decide which outcome the channel exists to produce and write down its recent level, as a BaselineAn estimate of what a metric would have done without a given change, campaign or promotion, against which the effect is measured. taken from the business's own records where possible: orders, enquiries, bookings or customers rather than only the platform's reported conversions. Note the measure's definition, its source, the period it covers and how much it normally moves from one week to the next. Read the history before the latest month, because a newcomer is especially likely to read a normal wobble as a problem. This baseline is what every later change is judged against, so record it before your first change, not after.
Find who knows the channel and ask for an hour
List everyone who has worked on or relied on the channel: the previous owner, an agency, a colleague who builds the creative, a developer who set up the tracking, someone in finance who pays the invoices, a salesperson who receives the leads. Ask the one who knows it best for an hour and bring your list of questions from the first step. Ask what they would never change, what they always meant to fix, what broke in the past and what they would watch first. If the owner is leaving, the tutorial on handing over work when a marketer leaves describes the record they should leave; ask for it.
Learn the settings that spend money or reach customers
Not every setting matters equally at the start. Find the ones that spend money, such as budgets, bids and schedules; the ones that decide who is reached, such as audiences, lists, locations and exclusions; the ones that send or publish, such as automated flows and scheduled content; and anything automated that acts without anyone pressing a button. Learn what each does from the platform's documentation and its section of this site, and write a line on each in your log. These are the settings where a mistake costs most and where you should be slowest to change anything.
Sort changes into reversible and irreversible
Take your list of things that look wrong or could be better and sort it using the one-way and two-way doors framework. Two-way doors are changes that can be undone easily and cheaply, such as pausing a single item, adding an exclusion or editing a line of copy. One-way doors are hard or impossible to undo, such as deleting history, changing how conversions are counted, sending to a whole list or ending a contract. Make two-way doors your first changes and leave one-way doors until you understand the channel, have talked to your reviewer and have your manager's agreement.
Make one small reversible change and read it
Choose one two-way door from the list that you expect to help and that you can observe. Make it alone, so that you can tell what it did, and follow the team's procedures as you would for any other change. Write in the log what you changed, when, why and what you expect to happen. Then watch the measures it should affect against the baseline for long enough to see past normal movement. Whether it helped or not, the point is to learn how this channel responds to a change and how long it takes to show it.
Run a first planned test
Once a single change has been made and read, plan a first proper test. The experimentation loop gives the shape: an observation from your reading, a hypothesis that names the change and the KPIA metric tied directly to an objective. it should move, a decision on how the test will be read, the test itself and a written learning. Where the channel supports it, run it as an A/B TestControlled comparison of two variants. against a control; where it does not, compare with the baseline and say so in the write-up. Run launches through the team's launch procedure and QA checklist rather than improvising them.
Keep a learning log with a reviewer
The learning log is a dated record of what you read, what you changed, what you expected, what happened and what you now believe about the channel. Keep it in the same place every time and write in it as you go, not at the end of the week. Ask a reviewer, such as your manager, a colleague who knows the channel or an agency contact, to read it on a regular rhythm and question it. A reviewer catches assumptions you cannot see and is the person who will notice if a change you thought was reversible was not.
Report what you are learning, not only results
In your regular reports, add a short section on the channel that says what you have learned, what you have changed, what you are testing and what you expect next, alongside the usual measures against the baseline. This sets expectations while results are still noisy and shows that the channel is being run deliberately. When a test ends, report it in the same way whether it won, lost or was inconclusive. Over time the log and these notes become the material for a case written as the portfolio guide describes.
What to learn first in any channel
Every channel has its own mechanics, which belong in the platform's own section of the site. But the questions to answer first are the same in any channel, and they are ordered by how much damage a misunderstanding can do.
| Area | Questions to answer first | Why it comes early |
|---|---|---|
| Money | What spends, how much, on what schedule and who approves changes? | Mistakes here cost money immediately |
| Reach | Who receives or sees the channel's output, and who is excluded? | Mistakes here reach customers and are hard to take back |
| Automation | What acts on its own, such as rules, flows and automated bidding or sending? | It keeps acting while you are still learning |
| Measurement | What counts as a result, where is it recorded and how does it match the business's records? | Every later judgement depends on it |
| Structure and naming | How is the channel organised and named, and why? | Changes made against the structure confuse reports |
| History | What was tried before, what worked and what broke? | It stops you repeating a past mistake |
| People | Who builds, approves, receives and pays for the channel's work? | They are your sources and your early warning |
The order matters more than the detail. A marketer who understands money, reach and automation in a new channel can make small safe changes while learning the rest; one who starts with optimisation tactics can do damage before they know it.
An email marketer takes on paid search for a tool hire firm (illustrative)
Brixomel learns paid search while it runs
Brixomel is an email marketer at an invented tool hire firm. When the colleague who ran paid search leaves, the channel passes to her, with a monthly budget of £6,000 and an agency that builds the ads but does not manage the account day to day. She has never run paid search.
She spends her first two weeks reading: the account's change history for the past year, the last six monthly reports and the search terms that triggered the ads. From the firm's booking system she writes a baseline of paid search enquiries, an average of 310 a month over the previous six months, ranging from 270 to 350; the platform reported 420 conversions a month over the same period, because it also counted repeat visits to the contact page. She gets an hour with the departing owner in his last week, who tells her never to touch the location settings, which exclude areas the depots cannot deliver to.
Her first change is a two-way door: 26 negative keywords for searches about buying rather than hiring tools, each logged. Spend on those searches stops and enquiries stay within the baseline's range. Her first planned test is a landing page for the most-searched tool category, run against the existing page for six weeks; the new page produces 8% more enquiries, but the range runs from slightly worse to considerably better, so she logs it as inconclusive and extends it. The agency's account director reviews her learning log every fortnight, and her monthly report to the operations director now has a short section on what she has learned and what she is testing.
The example shows the shape rather than the figures: a period of reading before any change, a baseline from the business's own records instead of the platform's count, the one setting that must not be touched found in the handover hour, a reversible first change, a planned test read honestly, and a reviewer outside the firm.
Common mistakes
- Changing things in the first days. Settings that look wrong are often deliberate. Read the history and ask before you fix anything.
- Measuring only with the platform's numbers. Platform-reported results follow the platform's counting rules. Take the baseline from the business's own records where you can.
- Making several changes at once. If everything changes together, nothing can be learned. Change one thing, log it and read it.
- Treating a one-way door as a two-way door. Deleting history, changing what counts as a conversion or sending to a whole list cannot be undone cheaply. Check before you act.
- Learning alone. Without a reviewer, assumptions go unchallenged. Find one, even outside the team.
- Reporting only results while you learn. Early results are noisy. Report what you are learning, changing and testing so expectations match reality.
- Skipping the team's launch procedure because the change feels small. The SOP and QA checklist exist for exactly the changes that feel too small to check.
