Self-paced · no timetable

The Social Media Practicum

Run an account that belongs to someone else, and be able to say why every post exists.

For people managing social media for a small business, starting from one real account.

How long: About 9 weeks: a first week at 2 hours a day, then 6 to 8 hours a week
Cost today: Level 0 free
The rest: Early access, by invitation or request

By the end

Take one real account to a running operation you can price, report on and hand over.
Publish a post whose job you can state, and read what it did against the goal behind it.
Write a positioning sentence, a voice document and content pillars tied to the goals.
Run a four-week calendar that survives the week where nothing goes right.
Set a monthly price you can defend from your own recorded hours.

Social media management looks like posting. Posting is the smallest part of the work and the first part to go wrong when the rest has been skipped. A manager who posts daily and cannot say who each post was for, what it asked people to do, or whether it worked, is producing activity rather than results.

The course runs seven levels over roughly ten weeks. The first three fill the opening week. You first read the ground: what the job is accountable for, and what each of the seven main platforms is actually for. You then aim: who the business is speaking to, and what it wants to happen. You then make and judge: content that has a job, a hook that earns three seconds, a post published to the destinations you intended, numbers read against the goal that produced the post, and accounts held securely enough to survive a lost phone.

Four more plateaus follow, and each takes longer because its gates consume real elapsed time. You make the business recognisable: positioning, a voice another writer can produce from, and pillars tied to the goals. You run the operation: a calendar, batched production, a bank of posts, and the week where nothing goes right. You face the public: the inbox as the job, the complaint that grows, and the statement written before it is needed. You charge for the work: a report an owner reads, what paid money buys, a price derived from your own records, and the handover document that proves the system exists without you.

Every lesson uses one invented business so that examples stay concrete. Every exercise and every gate runs against a real subject you name on the first day. That split is deliberate and it is explained on the practice subject page.

Cadence

Seven days, about two hours each. The two hours divide as follows.

Activity	Time
Lesson	30 minutes
Examples and observation	30 minutes
Practical work	45 minutes
Review and notes	15 minutes
Total	2 hours

Level 0 covers days one and two, Level 1 covers days three and four, Level 2 covers days five to seven. Levels 3 to 6 drop the daily structure for a weekly one, six to eight hours a week, because their gates consume elapsed time that cannot be compressed: Level 4 requires two consecutive weeks run against a calendar and Level 5 requires fourteen consecutive days of inbox duty. Reading faster does not make the weeks pass sooner, which is the point. The stated lesson minutes add up to roughly five and a half hours across the week, which is more than the thirty minute daily lesson slot. That is deliberate. Every lesson carries a Going deeper section holding the sourced evidence behind the lesson, and those sections belong in the half hour set aside for examples and observation rather than in the lesson itself. A learner in a hurry can pass the gates without them. A learner who intends to charge for this work should read them, because they are what separates a practitioner from someone repeating what they read on a marketing blog.

Where a claim in this course rests on published evidence, the source is named in the lesson and listed on the sources page. Where a widely repeated claim turns out to have no evidence behind it, the course says so and names the claim. There are more of the second kind than you would expect.

What competent practice looks like at the end of the week

You can be handed an account for a small business you have never seen, and within one working day produce a platform choice you can defend, three customer characters drawn from people you actually spoke to, three goals with numbers and dates attached, twenty content ideas, and one published post whose performance you can read against the goal that produced it. You can also say what you would not post, and why.

What competent practice looks like at the end of the course

You can take that same business and, within one month, produce the full working set: a positioning sentence the owner has agreed to, a voice document another writer can produce from, pillars tied to the goals, a four-week calendar with finished content in it, published posts, a one-page report the owner has read, and a monthly price you have said out loud and can defend from your own recorded hours. You can hand the whole system to another manager in a day, because every part of it exists as a document rather than as a habit in your head.

Where the rest of the course goes, and what it leaves out

The first week deliberately excludes most of the job, and the later levels pick the exclusions up in order. Brand identity, voice and content pillars are Level 3. Scheduling at scale and the operating calendar are Level 4. Community management and crisis handling beyond a single complaint are Level 5. Paid amplification and reporting to a client are Level 6.

What the whole course leaves out, deliberately: influencer work, advertising beyond a small boosted budget, graphic design as a craft in its own right, and reporting at agency scale across many clients. Each is a further course, not a missing chapter of this one.

The account is the evidence

A lesson is finished when something exists that another person could inspect: a table, a recorded interview answer, a draft post, a screenshot of a settings page. Reading a page and agreeing with it does not pass a gate.

The levels

7 levels · 1 free

Practice subject

One invented business to learn on, one real business to be judged on.

Every lesson in this course uses the same invented business, BrightBite Snacks, so that examples stay concrete and comparable. Every exercise and every gate uses a real business that you name before you start Level 0.

The split exists because an invented business teaches cleanly and proves nothing. BrightBite has no owner who disagrees with your platform choice, no customer who complains in public, no missing sales figures, and no week where nothing was posted because the person who takes the photographs travelled. A real subject supplies all four, and those four are most of the job.

BrightBite Snacks, the worked example

BrightBite sells chin chin, plantain chips, cookies, small snack packs and party snack boxes. Its customers include students, parents, office workers, event organisers and people buying gifts. It is based in Abuja. Every worked example, comparison and knowledge check in this course refers to it.

Your real subject

Name one before Level 0. It must have something to sell or something to ask people to do, at least one person you can speak to about it, and a reason to be seen by strangers. Any of these qualify: a shop, a restaurant, a tailor, a salon, a school, a church, a charity, a professional association, a tradesperson, or your own professional page.

You do not need permission to publish in order to start. Most of Week One produces plans, profiles, goals and drafts. In the first week only one gate requires something published, and a draft shown to five people satisfies it. From Level 4 the calendar runs on the real account, so publishing access is needed by then. If you have it, use it. If you do not, ask for it in the first week and record the answer, because asking is part of the job.

The subject you cannot get information about

If nobody will answer questions about the business, the audience work in Level 1 becomes invention and the goals in Level 1 become decoration. That is a signal to change subject in the first two days, not to proceed and fill the gaps yourself. A worse business you can ask questions about beats a better business you cannot.

Rules of practice

Ten rules. Breaking any of them fails the level regardless of the quality of the work.

Do not buy followers, likes, views or engagement, and do not use applications that supply them.
Do not copy another person's work. Reposting with credit and permission is not copying; lifting a caption is.
Do not use photographs, video, music or graphics without permission or a licence that covers the use.
Do not publish information you have not checked with the business, including prices, delivery times and availability.
Do not publish a customer's name, order, address, telephone number or photograph without their agreement.
Do not argue with followers in public. Answer once, move the exchange to a message, and record what happened.
Do not delete criticism because it is uncomfortable. Delete only what is abusive, false about a named person, or a privacy breach, and record what you deleted and why.
Do not promise what the business cannot deliver, including delivery windows you have not confirmed with the person who delivers.
Do not judge a post by likes when it was built to produce orders.
Do not publish anything whose purpose you cannot state in one sentence before it goes out.

Rules six and seven are the two most often broken by people who are otherwise good at the work, because both are broken under pressure and both feel justified at the time. The simulation in Level 2 exists to make that failure happen somewhere it costs nothing.

Assessment

Four parts and one hundred marks for the first week, and a portfolio that grows from eleven artefacts to twenty-two by the end.

The first week is assessed on what exists at the end of it, not on how much of the material was read. Three of the four parts require a real subject. The later levels are assessed by their gates alone, which is why each one produces artefacts the portfolio table at the foot of this page lists.

Part one: explain

Explain each of these in your own words, in two sentences or fewer, without notes: social media manager, target audience, content, platform, algorithm, engagement, conversion, call to action, hook, analytics.

Part two: analyse

Choose one real business account you did not study during the week. Identify its audience, its main platform, three content categories it uses, one strong post, one weak post, the likely goal of each of those two posts, and one improvement you would recommend with the reason it would help.

Part three: create

Produce one complete post for your real subject, carrying a named audience, a stated goal, a hook, useful content, a format suited to the platform, one call to action, correct spelling, and a completed publishing checklist.

Part four: present

Present the post in three minutes. State who it is for, what problem it addresses, why you selected the format, what action you want, and how you will measure success. Three minutes is the constraint that matters: a post whose purpose takes six minutes to explain does not have one.

Marks

Area	Marks
What the job is accountable for	10
Platform knowledge	10
Audience definition	15
Goal setting	10
Content ideas	15
Hooks and openings	10
Final post quality	15
Presentation	5
Account security	5
Portfolio organisation	5
Total	100

Score	Assessment
90 to 100	Excellent foundation
75 to 89	Strong foundation
60 to 74	Developing well
50 to 59	More practice required
Below 50	Repeat the weak sections

A high score is not the point. The learner who scores 78 and can create, explain and improve a real post is further along than the learner who scores 94 on paper.

The portfolio folder

Create a folder called Social Media Management Portfolio. By the end of Week One it holds eleven items, each naming the real subject it was made for and the date it was made.

Item	Produced in
01 Job description	Level 0
02 Platform plan	Level 0
03 Customer profiles	Level 1
04 Social media goals	Level 1
05 Content idea bank	Level 2
06 Hook library	Level 2
07 The post	Level 2
08 Caption	Level 2
09 Publishing checklist	Level 2
10 Feedback report	Level 2
11 Performance tracker	Level 2

The later levels grow the folder from evidence of a week into evidence of an operation.

Item	Produced in
12 Positioning sentence, with the owner's recorded verdict	Level 3
13 Voice document, one page	Level 3
14 Pillars, tied to goals	Level 3
15 Four-week calendar	Level 4
16 Minutes-per-post record, by format	Level 4
17 Deviation log, two weeks minimum	Level 4
18 Inbox log and response-time norm	Level 5
19 Approved holding statement	Level 5
20 Monthly report, as delivered	Level 6
21 Retainer arithmetic	Level 6
22 Handover document	Level 6

Over the full course this folder becomes the thing you show instead of describing what you can do. Keep the weak early versions. A folder that shows Week One beside the handover document is more persuasive than one that shows only the finished system, because the distance between them is the demonstration.

The daily record

At the end of each day, answer seven questions in writing: what you learned, what you created, what confused you, which account you studied, what that account did well, what mistake you noticed, and what you will practise tomorrow. The third and sixth questions are the ones that produce the next week's improvement. The others are records.

Worked exemplars

‡ Worked to standard

Four working artefacts in their working format, annotated line by line. They are the documents this course asks you to produce, shown at full strength rather than as blank templates, and every one of them is for BrightBite Snacks, the invented business the lessons already use, so their numbers are the lessons' numbers.

The voice document. One page, in the form a writer keeps open beside a draft. Annotated for where each rule came from, and for the two rules that were not on the page until a second writer broke them.

The deviation log. Two consecutive weeks of a calendar run against reality, in the four-field form. Annotated for the register that keeps a log alive, the pattern that only appears in week two, and the three decisions next month's planning took from it.

The monthly report. One page for a poor month, in the order an owner reads: goal units first, the shortfall stated, the hours visible, one decision with a date. Annotated for what each line is doing and what it refuses to headline.

The holding statement. Two prepared versions for the worst plausible situation, and the four sentences the crisis draft always wants, struck on the quiet day. Annotated for what each sentence is allowed to claim.

Sources

What the evidence in this course rests on, and which popular claims were checked and found to rest on nothing.

Every figure quoted in a Going deeper section is listed here with enough detail to find it. Full addresses are in the sources file kept beside the course, because this page cannot carry working links. Where a source is vendor research rather than independent work, that is stated, because it changes how much weight the number carries.

What the course cites

Claim	Source
Informative content lowers engagement alone, raises it when combined with personality	Lee, Hosanagar and Nair, Management Science, 2018, 106,316 posts across 782 companies
Office work switches spheres about every 11 minutes, 57.1 per cent of segments interrupted	Mark, González and Harris, CHI 2005, 24 workers observed roughly 25 hours each
Brand keyword search adverts were 99.5 per cent substitution; measured paid search ROI was negative while attribution reported it enormous	Blake, Nosko and Tadelis, Econometrica, 2015, 60-day field experiment across US markets
Specifying when, where and how raises follow-through by a medium-to-large effect and shields goals from derailment	Gollwitzer and Sheeran, Advances in Experimental Social Psychology, 2006, meta-analysis of 94 studies
Thesis predicted in 33.9 days, took 55.5; the worst-case estimate of 48.6 still fell short; 29.7 per cent finished on time	Buehler, Griffin and Ross, Journal of Personality and Social Psychology, 1994
A well-handled complaint raises satisfaction above pre-failure; no reliable effect on repurchase, word of mouth or image	De Matos, Henrique and Rossi, Journal of Service Research, 2007, meta-analysis
The party making the first offer obtains the better outcome; the effect is eliminated when the other party focuses on their own alternatives and target	Galinsky and Mussweiler, Journal of Personality and Social Psychology, 2001, three experiments
Trust is a purchase criterion alongside quality and value	Edelman Trust Barometer brand report, 2026, 17,688 respondents, 15 countries
Nigerian platform reach figures	DataReportal, Digital 2026 Nigeria, published November 2025
Small business capacity and confidence	Talker Research for Adobe Express, 2026, 1,000 owners; Constant Contact, 2025, 2,500 decision makers
Inconsistent posting underperforms, replies outperform	Buffer, State of Social Media Engagement 2026, 52 million posts, Buffer users only
Rival brands have near-identical customer profiles	Ehrenberg-Bass Institute, law of brand user profiles; Anesbury, Winchester and Kennedy, Marketing Letters, 2017
On-target delivery falls as targeting narrows	Nielsen Digital Ad Ratings, 2016, 44,000 campaigns, 17 countries
Third party audience data is roughly 24 per cent accurate	Neumann, Tucker and Whitfield, Marketing Science, 2019, 19 data brokers
Context-matched advertising raises intent, obtrusiveness cancels it	Goldfarb and Tucker, Marketing Science, 2011, 2.46 million respondents
Personas with nine or more attributes match nobody	Chapman, Love, Milham, ElRif and Alford, HFES, 2008; Chapman and Milham, HFES, 2006
Personas built then left unused	Salminen, Jansen, An, Kwak and Jung, Persona Studies, 2018
One verb changes a recalled estimate and manufactures memory	Loftus and Palmer, Journal of Verbal Learning and Verbal Behavior, 1974
Acquiescence bias affected 52 of 53 questions	Hill and Roberts, Political Analysis, 2023
Intention converts to action about half the time	Sheeran and Webb, Social and Personality Psychology Compass, 2016, 422 studies
Stated willingness 66 to 73 per cent, actual choice 14 per cent	Lehmann and Sheffi, MIT Center for Transportation and Logistics, 2019
Past behaviour predicts better only for frequent, stable actions	Ouellette and Wood, Psychological Bulletin, 1998, 64 studies
Specific difficult goals outperform, but not on novel tasks	Locke and Latham, American Psychologist, 2002
SMART originates in a trade magazine article	Doran, Management Review, 1981
Measures corrupt when they become targets	Goodhart, 1975; Strathern, European Review, 1997; Campbell, 1979
Liking a brand does not cause purchase	John, Emrich, Gupta and Norton, Journal of Marketing Research, 2017, over 14,000 participants
Marketers cannot demonstrate social media's impact	The CMO Survey, Duke University, 2016 and 2024 readings; 2025 budget share
Vanity metrics defined	Ries, guest essay, May 2009
Client and agency tenure has doubled	ANA and 4A's, 2025
Showing preparation raises perceived quality by 22.2 per cent	Buell, Kim and Tsay, Harvard Business School working paper, 2015
Only 19 per cent of diners want behind the scenes content	Toast, 2024, 850 respondents
About one third of well-designed experiments improve the target metric	Kohavi, Crook and Longbotham, Microsoft, 2009
Content spread has a prediction ceiling near half the variance	Martin, Hofman, Sharma, Anderson and Watts, WWW, 2016; Salganik, Dodds and Watts, Science, 2006
Instagram and TikTok ranking signals	Instagram ranking explainer, 2023; Mosseri, January 2025; TikTok support and newsroom documentation
The eight second attention span is false	Microsoft Canada, 2015, via Statistic Brain; traced by the Wall Street Journal and the BBC, 2017
Parallel prototyping beats serial iteration on real click-through	Dow, Glassco, Kass, Schwarz, Schwartz and Klemmer, ACM TOCHI, 2010
Surgical checklist cut deaths and complications	Haynes and colleagues, New England Journal of Medicine, 2009
A larger checklist study found no effect	Urbach and colleagues, New England Journal of Medicine, 2014, 101 hospitals
Choice overload is contested and moderated	Iyengar and Lepper, 2000; Scheibehenne, Greifeneder and Todd, 2010; Chernev, Bockenholt and Goodman, 2015
Two to three calls to action performed best in the largest dataset	Omnisend, 229 million emails, 2020
Instagram media rules are stricter than Facebook's	Adobe Content Scheduler specifications; Buffer publishing documentation
Post performance is heavy tailed	Neda, Varga and Biro, PLOS ONE, 2017, over 150,000 posts
Sample size rules for skewed metrics	Kohavi, Deng, Longbotham and Xu, KDD, 2014; Kohavi, Tang and Xu, Cambridge University Press, 2020
Advertising return is usually unresolvable even with millions of people	Lewis and Rao, Quarterly Journal of Economics, 2015
Experimenters stop as soon as results look significant	Berman, Pekelis, Scott and Van den Bulte, 2018, 2,101 experiments
Two factor authentication blocks most attacks	Google security research, 2019; Meyer and colleagues, Microsoft, 2023
The 24 hour Page deletion phishing pattern	Kaspersky, 2024; Better Business Bureau warning, 2024
Departed staff retain access	Beyond Identity, 2023, vendor survey
Least privilege is a formal control	NIST SP 800-53 AC-6 and SP 800-171
Page recovery obstacles, and Meta's unpublished figures	Letter from Senator Baldwin and Attorney General Kaul to Meta, 2023; Meta Transparency Center

Claims this course checked and rejected

These are repeated widely in social media training. Each was searched for a primary source and none was found. If you are taught any of them as fact, the person teaching you has not checked.

The 80/20 rule, the rule of thirds and the 4-1-1 rule for content mix. The 4-1-1 rule has a known originator and no study. The other two have neither.
The eight second human attention span, and the comparison with a goldfish.
That David Ogilvy wrote 104 headlines for the Rolls-Royce advertisement.
That the pottery class quantity experiment was a study. It was a photography teacher's classroom gambit, retold with acknowledged literary licence.
That a single call to action produces 371 per cent more clicks.
That saves and shares indicate more purchase intent than likes. The platforms say they affect distribution, which is a different claim.
That a password change silently unlinks a Facebook Page from its Instagram account.
That asking an owner to name their three best customers is an established method.
That WhatsApp dominates Nigerian small business commerce. Very likely true, and not currently supported by any dated survey.
Any specific figure for how many business Pages are lost to compromise each year. Meta publishes none.

What is open, and what is not

Level 0 of every programme is free: its lessons, checks and gates need no account, its tools need a free account, and the other levels and the worked exemplars are for early-access members.

Create a free account