AfterDesk
EN
Apply

The toolkit

SEO and content operations

This course covers the work clients actually hire you for: finding what people search for, building pages that answer it, formatting and publishing them cleanly, and checking the result. You will run keyword research and on-page checks with free tools only, ship through a CMS without breaking anything, and read Search Console without overclaiming. It also draws the line you never cross to make a page rank.

9
lessons
~60
minutes
12
exam questions

Free · No paid tier · No certificate fee

After this course

Everything, and what is in it.

Where Content Operations Sits

~6 min

The pipeline, end to end

Content work moves through the same stages almost everywhere. Someone decides a topic. Someone researches the query behind it and writes a brief. A draft gets written, then edited. The draft is formatted for the web, illustrated, and loaded into a content management system. Someone checks it before it goes live. It is published, linked from other pages, and shared. Weeks later someone looks at what it did and decides whether to update it. When a client hires a virtual assistant for content, they are usually buying the middle and the end of that chain: research, formatting, images, publishing, checking, reporting. This course is about doing those stages well enough that nothing breaks and nothing has to be redone.

Operations, not strategy

Strategy is deciding what to publish and why. Operations is making sure it ships, correctly, on the day it was meant to. Most clients keep strategy and hand over operations, and we think that suits you: operations is measurable, teachable, and the part that quietly fails when nobody owns it. Know which decisions are yours. If the brief names a target query, that strategy decision is already made. If it does not say which category the post belongs in, that is an operations decision, so make it, note it, and move on. When a real strategy question appears in the middle of your work, do not settle it silently. Name it and ask.

What a usable brief contains

Before you write or publish anything, check that the brief answers these: the target query, the intent behind it, who the page is for, the angle, roughly how long, the points that must appear, which existing pages to link to, what the reader should do at the end, the publish date, and where it goes live. A brief missing two or three of these is normal. A brief missing all of them means someone has to make those calls, and if that someone is you, say so before you start rather than after. Writing the missing pieces down and getting them confirmed takes ten minutes and saves a rewrite.

Access and client data

Content work hands you keys: a CMS account, a media library, sometimes analytics. Treat all of it as borrowed. Use your own named account rather than a shared login, keep the password in a password manager, turn on two-factor authentication where the client allows it, and never pass access to anyone else. Client drafts, files and screenshots stay inside the job: no uploading them to converters, cloud drives or AI tools unless the brief says so, no keeping copies once the work is approved, and no using the finished page as a portfolio sample without written permission. On AfterDesk the shape is simpler. Files come with the task and go back through us. No client logins, no client contact.

Remember

  • Clients usually keep strategy and hand over operations: research, formatting, publishing, checking, reporting.
  • A brief answering query, intent, angle, length, links and deadline is workable.
  • Make operations decisions yourself. Name strategy decisions and ask.
  • Client files stay inside the job. No outside tools, no copies, no portfolio use.

Keyword Research With Free Tools

~7 min

A keyword is a real question

A keyword is not a magic phrase. It is what a person actually typed, in their own words, at a moment when they wanted something. Your job in research is to find those phrasings, not to invent clever ones. Short phrases like running shoes carry enormous demand and enormous competition. Longer, more specific phrases like running shoes for flat feet wide fit carry less demand each, but there are far more of them, they say much more about what the person wants, and a small site can realistically answer them. Most content work for small and medium clients lives in that long tail. Collect queries in the exact words people use, awkward grammar included. That phrasing is data.

The free stack

You can do usable research without paying anything. Google's own search box suggests real queries as you type, and adding a letter after your phrase surfaces more. The People also ask box gives you subtopics and their wording. Related searches at the bottom of the results page gives you neighbours. Google Trends shows relative interest over time and by region, which is enough to compare two phrasings or catch seasonality. Google Search Console, if the client already has a site, shows the queries bringing them impressions today, and that is the highest quality source you will ever get. Forums, question and answer sites, and the contents list of a long Wikipedia article show how people describe a problem before they know its vocabulary.

What the paid tools add

Paid platforms exist and clients sometimes name them. They estimate monthly search volume, difficulty scores and competitor keyword lists in one place, which saves hours. Two things are worth knowing. First, those volumes are estimates, including in the expensive tools, so treat any number as a rough band rather than a fact. Second, you can approximate the important judgement without them. Google Keyword Planner gives broad ranges free if the client already has an ads account. If a client asks for volume data and has no tool, say plainly what free sources can and cannot produce instead of presenting a guess as a measurement. Being clear about the limits of your data is part of the work.

Building the keyword sheet

Put the research in one Google Sheet, one row per query, with columns for the query, where you found it, the intent, the page type that would answer it, priority, the target page once decided, and notes. Then group. Queries that want the same answer belong to one cluster and one page: how to clean suede shoes and cleaning suede sneakers are the same job. Give each cluster a primary query and keep the variants as rows underneath it. This grouping is the actual deliverable. A flat list of two hundred phrases is raw material. A sheet of thirty clusters, each with an intent and a proposed page, is something a client can act on this week.

Remember

  • Collect the words people actually type, awkward grammar included.
  • Autocomplete, People also ask, related searches, Trends and Search Console cost nothing.
  • Volume numbers are estimates everywhere. Never present a guess as a measurement.
  • Deliver clusters with intent and a proposed page, not a flat list.

Reading Search Intent

~6 min

Four things a searcher wants

Behind every query is one of four wants. Some people want to know something: how composting works. Some want to go somewhere specific: a company name plus the word login. Some are comparing before they buy: best budget laptop for students, alternatives, reviews. Some are ready to act: buy, price, book, download, near me. The same words can carry different intent in different markets, so never decide from the phrasing alone. What matters is that the page you produce matches the want. An article aimed at someone ready to buy will lose to a product page, and a product page aimed at someone still learning will lose to a guide. Getting this right before writing saves the whole piece.

The results page tells you

The fastest way to read intent is to run the query yourself in a private window and study the first ten results. Notice the format: list articles, step by step guides, product pages, category pages, tools, videos, forum threads. Notice the depth, the publication dates, and whether large well known sites dominate. Notice what Google adds around them: People also ask, images, a map pack, shopping results, a featured snippet. That mixture is the answer Google has settled on for this query, tested against millions of people. You are not obliged to copy it, but you should be able to explain why you are doing something different. When ten results are comparison lists, a sales page will not win that query.

When the brief and results disagree

This happens often. A client wants a page selling their service to rank for a question people ask while they are still learning. Do not quietly write the sales page and hope, and do not quietly change the target either. Say what you found: the top results are all guides, so a sales page is unlikely to rank here. Then propose the version that works, usually a genuine guide that answers the question and links clearly to the service page. That structure serves the reader and the client at once. Raising this before you write costs one message. Raising it after the page has been live for three months costs the client a quarter.

One page per intent

If two pages on the same site chase the same want with the same answer, they compete with each other, split their internal links, and confuse anyone deciding which one to show. Before adding a page, search the client's own site for the query using the site colon operator in Google, or check the content calendar. If a page already covers it, the right move is usually to improve that page rather than publish a near duplicate. If a brief asks for a new page that duplicates an existing one, flag it with the existing URL attached. That single link makes the decision easy for whoever owns the strategy.

Remember

  • Every query wants one of four things: to know, go, compare, or act.
  • Read the top ten results to see the format that intent rewards.
  • Raise a mismatch before writing, and propose the version that works.
  • Two pages chasing one intent compete with each other. Improve one instead.

The On-Page Checklist

~8 min

Title tag and meta description

The title tag is the clickable line in search results, and it is separate from the headline shown on the page, even when a CMS fills both from one field. Keep it around sixty characters so it is not cut off, put the main query near the front, make it different from every other title on the site, and write it as a sentence a human would click. The meta description is the grey text below it. It does not directly decide rankings, but it decides clicks, so write roughly one hundred and fifty characters saying what the reader gets. Google often rewrites both, which is not a failure. Empty fields, duplicates across pages, and padding with repeated keywords are the failures.

Headings map the page

Use one H1, matching what the page is about. Use H2 for each main section and H3 for subsections inside them, without skipping levels. Headings are structure, not styling: if an H3 looks too small, change the theme, not the tag. Written well, your headings read on their own as a table of contents, and that is the test. Someone should understand the article from the headings alone. Screen reader users navigate by them, and search engines use them to work out what the page covers. Avoid headings that say nothing, like Introduction or More information. Say what the section actually contains, in the reader's words where that fits naturally.

URL slug and internal links

Slugs should be short, lowercase, hyphenated and readable: cleaning-suede-shoes, not post-id-4471 and not a full sentence. Set it before publishing, because changing a live URL breaks every existing link unless a redirect is put in place, and that belongs to whoever owns the site. If you think a live slug needs changing, flag it rather than doing it. Internal links matter more than most people expect. Add two to five links from your new page to genuinely relevant existing pages, and add at least one link from an established page back to the new one so it is not an orphan. Use anchor text that describes the destination, and click every link in preview.

Alt text and the opening

Alt text describes what an image conveys in its context, for someone who cannot see it. Write what matters: a bar chart showing sales doubling between 2021 and 2024, not chart or image of chart. Skip the words image of, keep it to a sentence, and include the query only where it belongs naturally. Images that are purely decorative take an empty alt attribute so screen readers skip them. The opening of the page matters just as much. Answer the query in the first hundred words, in plain language, using the searcher's phrasing once because it is the natural way to say it. There is no keyword density target worth chasing. Write it once, properly, and move on.

Run it as a list

None of this is hard. All of it is forgettable at one in the morning on the fourth article of the night. So build the checklist once, in a sheet or a document, one row per item and one column per page, and run it top to bottom every time. Title present, short, unique. Meta description written. One H1. Heading levels in order. Slug clean. Two to five internal links out, one in. Every image compressed, renamed, with alt text. Opening answers the query. Every link clicks through. The point of a checklist is that it is boring and complete, and that it does not depend on how tired you are.

Remember

  • Title around sixty characters, query near the front, unique on the site.
  • One H1, ordered headings, and headings that read as a table of contents.
  • Set the slug before publishing. Changing a live URL needs a redirect.
  • Alt text describes function. Decorative images take an empty alt attribute.
  • Run the same written checklist every time, no matter how tired you are.

Formatting an Article for the Web

~6 min

People scan before they read

A reader arriving from search decides in a few seconds whether the page holds their answer. Give them that answer near the top, then let the detail follow. Keep paragraphs to two to four sentences, because a wall of text on a phone is a wall. Put a subheading every few hundred words so someone scrolling can find the part they need. One idea per paragraph, and the sentence carrying the point goes at the start, not buried at the end. None of this means writing shallow. Long, detailed pages do well when they are navigable. What loses readers is not length, it is not being able to see the shape of what they are reading.

Lists, tables, emphasis

Use a list when items are genuinely parallel and short. Do not turn every paragraph into bullets, because a page of bullets carries no argument and reads as notes. Use a table when you are comparing several things across the same attributes, and keep it narrow enough to survive a phone screen. Bold the phrase a scanner must not miss, once or twice per section, never whole paragraphs. Avoid all capitals, avoid coloured text as the only signal for something, and avoid underlining anything that is not a link, because readers will try to click it. Emphasis works by contrast. When everything is emphasised, nothing is.

Links and clean markup

Write link text that describes where it goes, not click here or read more, because that text is what screen readers list and what search engines read. Link to external sources you actually opened, and quote them accurately. If a link is sponsored or an affiliate link it needs the right rel attribute, and whether a link counts as sponsored is the client's call, so ask rather than guess. Never strip attributes off existing links. When you paste from a word processor, paste as plain text and rebuild the formatting in the editor, because pasted markup carries invisible font tags and stray styles that break the site's design and are miserable to clean up later.

Check it on a phone

Most readers are on a phone, and most drafts are built on a laptop. Before you call a page done, look at it narrow: use the CMS preview, resize your browser window, or open the preview link on your own phone. Check that the first screen makes the subject obvious, that images are not cropped into nonsense, that tables scroll instead of overflowing, that line lengths are comfortable, and that links are far enough apart to tap. Then read the page once in preview as a reader rather than as the person who built it. Formatting problems are much easier to see when you stop editing and start reading.

Remember

  • Answer near the top, short paragraphs, a subheading every few hundred words.
  • Lists for parallel items, tables for comparisons, bold for the line that matters.
  • Descriptive link text. Paste as plain text and rebuild formatting in the editor.
  • Check every page narrow, on a phone or a resized window, before publishing.

Images That Earn Their Place

~6 min

Where images come from

In order of preference: the client's own library, screenshots or photos you make yourself, then free stock libraries with clear licences such as Unsplash, Pexels or Pixabay, then Wikimedia Commons, where many files are free but require a specific attribution line you must reproduce. Never take an image from a search results page or another company's site. Licensing claims are one of the few ways a small blog post generates a real bill. Record the source and licence of every image in your sheet, next to the page it appears on, so that if anyone asks in a year you can answer in ten seconds. Canva's free plan is fine for simple graphics; check what its licence allows commercially.

Resize, then compress

The most common mistake in content work is uploading a photo straight from a phone or camera. A five megabyte image displayed eight hundred pixels wide makes the page slow for everyone and helps nobody. Resize first, to roughly the width the design actually uses, then compress. Squoosh runs in a browser and is free; GIMP does the same job offline. Use JPEG for photographs, or WebP where the CMS supports it, since it is usually much smaller at the same quality. Use PNG for flat graphics, screenshots and anything needing transparency, and SVG for logos and icons. A full width photograph under a few hundred kilobytes is a reasonable target. Always check the result at full size before accepting the compression.

Names, alt text, captions

Rename files before upload. A name like suede-shoe-cleaning-step-3.jpg tells a human, a search engine and your future self what it is; the name the camera gave it tells nobody. Lowercase, hyphens, no spaces. Write alt text for every image that carries meaning, describing its function on the page rather than listing everything visible in it. Captions are optional and worth using, because captions get read more often than body text, which makes a caption a second chance to land a point. Do not simply repeat the alt text there; the two do different jobs for different people. Set the featured image as well, since that is what appears in listings and when the page is shared.

Screenshots and what they leak

Screenshots are the fastest way to publish something you did not mean to publish. Before one goes into a page, look at every corner: browser tabs, bookmarks bar, notification popups, real customer names, email addresses, account numbers, internal URLs, and other clients' data sitting in a shared dashboard. Crop hard, blur what remains, or rebuild the screenshot with dummy data, which is usually cleaner than blurring. Never publish a screenshot showing one client's information on another client's site. When the work is approved, delete the working images from your machine along with everything else from the job. Assume anything visible in a published image will eventually be read by someone.

Remember

  • Client library first, then free stock with a clear licence. Never scrape images.
  • Resize to display width, then compress. Phone-sized uploads are the usual mistake.
  • Rename files, write alt text, set the featured image, use captions deliberately.
  • Screenshots leak names, emails and other clients. Crop, blur, or rebuild with dummy data.

Publishing and the Pre-Publish Pass

~7 min

Draft, preview, schedule, publish

Every content management system has the same states under different names: draft, preview, scheduled, published, and usually a revision history. Work in draft. Judge the page in preview, which renders it in the real template, rather than from inside the editor. Schedule rather than staying awake to press a button at seven in the morning. Know your role as well: an author can usually write but not publish, an editor can publish anything, an administrator can break the site. Ask for the smallest role that lets you do the job. Publishing rights are trust, not a convenience, and the default rule is that you do not publish anything the client has not approved, unless the brief explicitly puts the calendar in your hands.

The fields nobody remembers

Beyond the body text, a post carries settings that decide how it behaves: slug, excerpt or summary, featured image, category, tags, author, publish date, template, canonical URL, and an index or noindex setting. Use categories and tags that already exist rather than inventing new ones, because an invented tag creates a thin archive page nobody wanted. Watch the two silent killers. A canonical URL pointing at another page tells search engines to ignore this one. A noindex setting keeps it out of results entirely. Both are invisible on the live page, and both are commonly inherited when you duplicate an old post to reuse its layout, which is otherwise a good habit.

The pre-publish pass

Run the same pass every time, in preview, before anything goes live. Title tag present, short, unique. Meta description written. One H1, headings in order. Slug clean and final. Every internal and external link clicked. Images compressed, renamed, with alt text, featured image set. No placeholder text, no leftover comments, no tracked changes. Spelling checked by a tool and then by your own eyes. Mobile view checked. Category and author correct. Publish date correct. Canonical and index settings correct. Then publish. This takes five to eight minutes on an article, and it is the difference between someone a client trusts with the keys and someone they check behind every single time.

After it goes live

Open the live URL in a private window, logged out, and read it as a stranger would. Check that it appears where it should: the blog index, the right category, the navigation if it belongs there. Add the internal link from an older page if you have not already. If the client uses Search Console, you can inspect the URL and request indexing, which is a nudge rather than a guarantee. Record the live URL in the content calendar. If the layout is badly wrong, revert to the previous revision instead of editing live under pressure. Then write down what you changed and when, because in three months that log is the only reliable memory anyone has.

Remember

  • Work in draft, judge in preview, schedule instead of staying up.
  • Ask for the smallest CMS role that does the job. Never publish without approval.
  • Canonical and noindex settings are invisible on the page and inherited by duplicates.
  • Run the same written pre-publish pass every time. It takes under ten minutes.
  • After publishing, check the live URL logged out and record it in the calendar.

Content Calendars and Repurposing

~6 min

The calendar is a status board

A content calendar is not a decoration. It is the one place anybody can look to see what is coming and what is stuck. One row per piece, in a shared Google Sheet, with columns for publish date, working title, target query, intent, format, owner, status, brief link, draft link, live URL and notes. Keep the status values fixed and few: idea, briefed, drafting, in review, scheduled, live, needs update. Fixed values let you filter and count; free text does not. Update the sheet the moment something changes rather than at the end of the week. A calendar three days stale gets ignored, and once it is ignored people start asking you for status by message instead.

Planning a month honestly

Plan against the capacity you actually have, not the capacity you wish you had. Count the hours, subtract the ones already committed, and leave slots empty on purpose for the work that always arrives: an urgent update, a broken page, a piece that comes back from review needing a rewrite. Batch similar work so you are not switching tools all day: research several briefs in one sitting, do a set of images in another. Then plan around the real bottleneck, which is almost never your writing speed. It is review time on the client's side. Build that window into the dates, ask for it explicitly, and mark a piece blocked when it overruns rather than absorbing the delay silently.

One article, several assets

A finished guide is raw material for a lot of other work. It can become a newsletter section, several social posts each built on one point rather than the whole thing, a question and answer block on a service page, a short video or carousel script, a hub that links related posts together, or a slide deck for the sales team. The rule is to rewrite for the channel instead of pasting the same paragraphs everywhere, because a paragraph that works in an article reads wrong in an email and worse in a caption. Always link back to the original. Republishing the full text on another domain is legitimate when handled with a canonical tag or a rewrite, but that decision belongs to the client.

Updating beats writing new

An existing page that already collects impressions is usually a better investment than a new one. Look for pages sitting just below the top results, or pages that have gone stale: outdated facts, dead links, screenshots of an interface that has changed, a topic the results now cover differently. Update the content properly, add the sections that are genuinely missing, fix the links, refresh the images. Change the visible date only if you truly made substantial changes, because a date bumped on an untouched page is a small lie that readers notice. Keep an update log, in the calendar or at the foot of the page, saying what changed and when. That log makes the next update fast.

Remember

  • One shared sheet, one row per piece, fixed status values, updated immediately.
  • Plan against real capacity and build client review time into the dates.
  • Rewrite for each channel when repurposing, and always link back to the original.
  • Updating a stale page often beats a new one. Log what you changed.

Search Console, and the Line

~8 min

What Search Console shows

Google Search Console is free and shows how a site appears in Google search, which is not the same as analytics: it is about the results page, not about what people do once they arrive. The client grants you access to their existing property. You never verify or create a property for a site you do not own. Three parts cover most of the work. The Performance report lists queries and pages with their numbers. The Pages report tells you what is indexed, what is not, and why. URL Inspection checks a single page and lets you request indexing. Start there. Everything else in the tool can wait until somebody asks you for it.

The four numbers

Impressions count how often a page appeared in results. Clicks count how often someone chose it. Click-through rate is clicks divided by impressions, and it mostly reflects how well your title and meta description are doing their job. Average position is exactly that, an average across many queries and many searchers, so it moves for reasons that have nothing to do with you. Compare a period against the previous period of the same length, twenty-eight days against twenty-eight days, rather than reading day to day. Remember that search demand is seasonal, and that a page with forty impressions can swing wildly on noise alone. Small numbers are not trends.

What you may conclude

Some readings are safe. Many impressions and almost no clicks means people see the page and do not choose it, so the title and description are the first thing to change. No impressions at all means the page is new, not indexed, or aimed at something nobody searches, and the Pages report tells you which. Clicks arriving on queries you never targeted is real information about what the page actually answers. Other readings are not safe. You cannot conclude that a ranking change caused a change in sales, and you must never promise a position to anyone. Write reports the way you write a delivery note: what you did, what the numbers show, what you think it means, what you would try next.

The honesty line

Four things you do not do, whatever the deadline. You do not fabricate expertise: no invented credentials, no author bio for a person who does not exist, no statistic without a source you actually opened, no made-up quotes, reviews or case studies. You do not stuff keywords, hide text, or write for a machine at the reader's expense. You do not present generated text as original research, first-hand experience, or testing that never happened. And you do not copy or reword somebody else's article and publish it as the client's. Use whatever tools help you draft, within the confidentiality rules, then verify every fact and every claim yourself and be able to say where each one came from.

When you are asked anyway

Sooner or later someone asks for the thing that crosses the line: fifty spun articles, invented statistics, reviews written by nobody, a claim you cannot verify, links hidden from readers. Answer plainly and in writing. Say what you will not do, say why in one sentence, and offer the version you can do instead, with what it would realistically take. Most of these requests come from pressure rather than bad faith, and a concrete alternative usually ends the conversation. If it does not, decline the work. On AfterDesk you never contact the client: put the problem in your delivery note so we can settle it, or release the task. Your name stays attached to your work long after any campaign is forgotten.

Remember

  • Search Console is about the results page, not what visitors do afterwards.
  • Compare twenty-eight days to the previous twenty-eight. Small numbers are noise.
  • High impressions with no clicks points at the title and description first.
  • Never fabricate expertise, stuff keywords, or pass generated text off as research.
  • Asked to cross the line, say no in writing and offer a real alternative.

Sit the exam

The courses are free. The work is real.

Twelve scenario questions. Pass at ten. Three attempts a day. The bar is the point.

No paid tier. No certificate fee. No upsell. Not now, not later.

Create a free account

You can start the first course tonight.

The curriculum

Everything, and what is in it.