Where we start
Owned channels are the ones the company controls: its website, its articles, its own social accounts. Earned channels are the ones it has to win: a ranking in Google, a creator who chooses to post, a journalist who chooses to write. Neither is bought by the click, which is why both feel free and why both are judged on whatever number is easiest to count. Every channel tonight is one card with the same four beats, so by the fifth you will be reading the shape without the labels. Kettle & Co is the constructed tea brand from Angles session 1, and every Kettle & Co number on this page is constructed.
Reading the channel, not running it. Using AI to do the SEO, write the content or draft the social posts is taught in AI + Marketing ↗. This page is about the other side of the desk: reading each channel's data and asking the question that turns a rising line into a decision. The marketing team runs the channel and owns what it does next; the data team supplies the evidence, because whoever lives with the outcome owns the decision.
★ What you walk out with tonight. A four-beat card for each of five channels you can reuse on any new one, the vanity number for each and the question that replaces it, a working sense of why SEO before/after reads need weeks, and one sentence, built from a constructed quarter of data, that names an insight, the decision it informs and the person who owns that decision.
Search and content: the slow channels people find on their own
6 min liveThe website and its articles are owned: the company writes them and can change them tonight. The ranking that brings people to them is earned: Google decides it, and nobody can buy it. That split is the whole reason search data is slow and noisy, and the reason a data person's most useful move here is a join, from the query a person typed, to the page they landed on, to the order they did or did not place.
LiveSEO4 min▶
1 · What it is. Search engine optimisation is the work of making a site easy for Google to find, understand and show for the queries the company's buyers type. Google describes its search in three stages, crawling, indexing and serving, and says plainly that "not all pages make it through each stage" (Google Search Central, read 10 October 2026). It does not guarantee any page will be crawled, indexed or served, and it does not accept payment to rank a site higher. Its starter guide adds that a change can take effect in a few hours or take several months.
2 · What marketers watch. Rankings for target queries, impressions (how often the site appeared in results), organic clicks and click-through rate, usually from Google Search Console and an SEO tool.
3 · The vanity trap. Total organic traffic, and a page-one ranking for a big query that does not sell. At Kettle & Co (constructed), the article ranking for "matcha benefits" brings 5,100 clicks a quarter, more than any other query, and 21 orders. It is a lovely line on a traffic chart and close to irrelevant to revenue.
4 · Where data adds value. Join Search Console's queries to the landing pages they reach and to the orders those sessions place, then split branded queries (people who already know the name) from non-branded ones (people the channel is genuinely finding). In the constructed quarter, branded search is 610 of 871 organic orders. And because Google says changes can take hours or months, any before/after read of an SEO change needs weeks of data on both sides, never a Monday-to-Friday comparison. Google's behaviour here is as of October 2026; check Google Search Central's current pages before you rely on it.
LiveContent2 min▶
1 · What it is. Articles, guides, recipes and videos the company publishes on its own site. A reader finds a piece through search or social, reads it, and the piece's job is to move some of those readers on, through an internal link, to a product page and an order or a newsletter signup.
2 · What marketers watch. Pageviews, time on page, scroll depth and shares. These say whether people read the piece, which matters to the person who wrote it.
3 · The vanity trap. Pageviews. A guide that a thousand people read and leave is a success by the content team's scoreboard and a cost by the finance partner's.
4 · Where data adds value. Trace the path from each piece to a purchase or signup: how many readers went on to a product page, and how many of those ordered. That splits content into two honest kinds, content that sells and content that informs, and both are allowed. At Kettle & Co (constructed), "how to brew oolong" brings 3,400 clicks and 12 orders in a quarter: useful, well read, and not a sales asset. The decision it informs belongs to the marketing lead: what the next ten articles are for.
Self-studyWhy the branded split changes every search conversation2 min▶
A person who types "kettle and co" already knows the brand. Something else, a friend, a creator, a newspaper, an advert, did the work of getting them there, and search is the door they used. A person who types "tea subscription uk" is being found by the channel. Blending the two makes organic search look like a growth engine when much of its revenue is demand created elsewhere. In the constructed quarter, 610 of 871 organic orders are branded; the non-branded queries bring 13,500 clicks and 261 orders. Neither number is bad. They answer different questions: branded search is closer to a brand-health signal (Angles session 1), non-branded search is the channel doing its own job.
The ranking everyone celebrated. Kettle & Co's content team reaches position one for "matcha benefits" and the traffic chart doubles in a month. The marketing lead asks whether to commission five more articles like it. One join, query to landing page to order, shows the page sells 21 orders on 5,100 clicks while the much quieter "tea subscription uk" page converts at 9.2 orders per 100 clicks. The decision moves from "more of what got traffic" to "more of what buyers type". Constructed example; the pattern is common.
Social and creators: borrowed audiences, owned evidence
5 min liveA brand's own social account is owned in name only: the platform's feed ranking decides who sees each post, and that ranking changes without notice. A creator's audience is borrowed outright. In both, the platform reports generous numbers about reach and engagement, and the only numbers the company truly owns are the ones that arrive on its own site: sessions, orders, new customers, and whether they come back.
LiveOrganic social3 min▶
1 · What it is. Posts on the brand's own accounts, unpaid. The platform's feed ranking decides how widely each post is shown; some of the people who see it engage, and a smaller share click through to the profile or a link and land on the site.
2 · What marketers watch. Followers, reach, and engagement rate (likes, comments, shares and saves as a share of views or followers).
3 · The vanity trap. Follower count. At Kettle & Co (constructed), followers rose 40 percent in a quarter, from 12,500 to 17,500, while social-referred orders went from 121 to 128. The audience grew; the business it sends barely moved.
4 · Where data adds value. Social-referred sessions and orders by post, which only the company's own site data can supply. That separates engagement content from sales content. In the constructed quarter the most-viewed post, a cat knocking over the teapot, drew 41,000 views and 2 orders; the autumn blend launch post drew 8,000 views and 63 orders. Both have a job. Only one of them should be judged on orders, and the brand manager decides which job each post is given.
LiveCreators (organic and gifted)2 min▶
1 · What it is. The brand sends a brief, and often free product, to a creator, who posts to their own audience. Some of that audience uses the creator's discount code or tracked link and orders. Gifted and organic creator work is earned: the creator chooses whether and what to post. Paid influencer deals are read in Angles session 3.
2 · What marketers watch. Views and engagement on the creator's post, and how many discount codes were redeemed.
3 · The vanity trap. Views, and "estimated media value", a figure that prices the creator's reach as if it had been bought as advertising. It prices the space, not the effect on sales.
4 · Where data adds value. Orders and new customers by creator code and by tracked link, with returning customers separated out. Watch for codes leaking to coupon sites, where a code turns up in a browser extension and is redeemed by people who were already checking out (the same pattern as the affiliate coupon sites in bench ask #8). Then follow the customers: in the constructed quarter, 31 percent of creator-acquired customers had ordered again by month 6, a number Angles session 4 sets against every other channel.
This is bench ask #3, "Which creators drove sales?", and it is a split ask: the marketing lead is accountable and responsible, data is consulted with a named deliverable (orders and new customers by creator code and link, 5 modelled hours, constructed), and the brand manager is consulted. The creator choice stays with marketing because marketing lives with the relationship and the result. How that row was drawn is in Partnership session 2.
The 40 percent that sent seven more orders. Kettle & Co's brand manager brings a slide: followers up 40 percent in a quarter, "is TikTok working?" (bench ask #2). The data team puts social-referred sessions and orders by month next to the follower line: orders went from 121 to 128. The answer is not "no". It is "the account is growing an audience; it is not yet growing sales, and here are the two posts that did sell". Constructed example; the pattern is common.
PR: earned coverage and the trail it leaves
4 min livePR is the most earned channel of all: a journalist, editor or podcaster decides to cover the brand, and nobody at the company controls what they say. Its effect rarely arrives as a click with a tracking code. It arrives as people who read the piece and later type the brand's name into Google or come straight to the site, which is why the data team reads PR through branded search and direct visits.
LivePR4 min▶
1 · What it is. The PR lead pitches a story to journalists; some of them cover it; readers see the coverage; some of those readers later search for the brand by name or visit the site directly.
2 · What marketers watch. Coverage count, the total reach of the outlets that ran it, and share of voice against competitors.
3 · The vanity trap. "Advertising value equivalent" (AVE), the price the coverage would have cost if bought as advertising space, and "total potential reach", the summed audience of every outlet. Both measure the size of the stage, not whether anyone in the audience changed what they did.
4 · Where data adds value. Branded search and direct sessions for the two weeks before and after the coverage, set against the same weeks last year, so that a seasonal rise is not credited to the article. This is bench ask #4, "Did the newspaper feature move anything?", which data owns (data accountable and responsible, brand consulted), because the evidence is all internal. What the brand manager does with the answer, pitch that outlet again or not, stays hers.
Self-studyWhat a before/after read can and cannot say2 min▶
A before/after window with a last-year comparison is a reasonable read, not an experiment. It cannot rule out that something else happened the same fortnight: a creator post, a competitor stock-out, a sunny week. Say so on the slide. If the brand wants a sharper answer for a big campaign, that is a lift design (Angles session 6), and attribution models and geo-lift are taught in Marketing Attribution ↗. For a single article, the honest read is usually "branded search rose more than last year's same weeks, and here is what else happened then".
The clippings report with a very large number on it. After a national newspaper feature, Kettle & Co's PR agency sends a report whose headline is the advertising value equivalent and a potential reach in the millions. The brand manager asks the data team whether to renew the agency. The data team writes the window down before looking: two weeks either side of the feature, branded search and direct sessions, against the same weeks last year. Fixing the window first means the read cannot be stretched afterwards to find a bump. Constructed example; the mechanism is the habit.
The one insight worth a meeting
★ 22 min · everyone works it throughNo code tonight. A constructed quarter of Kettle & Co data: six search queries from Search Console joined to orders, and four social posts with their views and orders. A data team could produce twenty charts from it. The job is to produce one sentence: an insight, the decision it informs, and the person who owns that decision. Work the steps in order, out loud for the first two, then alone. Your numbers will differ when you do this on your own company's data; the shape of the reasoning is what travels.
Step 1 · Sort the queries by clicks. This is the order the traffic report already shows, and the order most content plans are made in. Read it top to bottom and notice what you would conclude if you stopped here.
Step 2 · Re-sort by orders per 100 clicks. Same six rows, a different question: of the people this query brings, how many buy?
Step 3 · Split branded from non-branded. The branded row is demand created somewhere else. Take it out and look at what the channel finds on its own.
Step 4 · Put the four posts side by side. Views and orders, nothing else. Mark each post as engagement content or sales content, and do not mark either kind as wrong.
Step 5 · Pick ONE insight. You will have spotted four or five things. Choose the one that would change what the marketing team does next quarter, and drop the rest from the meeting (they can go in the appendix). In this constructed quarter the strongest candidate is that the content that wins clicks is not the content that sells, so the next articles should be built for buying queries such as "tea subscription uk", at 9.2 orders per 100 clicks.
Step 6 · Write it as one sentence with the decision and its owner. The insight, the decision it informs, and the business person who owns that decision. Here the owner is the marketing lead, because she will be judged on next quarter's content plan. The data team recommends what the evidence supports; it does not set the plan.
One insight, not twenty charts. A meeting can act on one finding with a decision attached. It cannot act on a deck that shows every cut of the data, because nobody in the room can tell which cut matters. Data informs the bet; the business places it.
Before Angles session 3
30 min- Run the join on your own search data. Take your top twenty Search Console queries, join them to landing pages and orders, and re-sort by orders per 100 clicks. Note which rows swap ends.
- Split branded from non-branded and write down what share of organic orders is branded. Your numbers will differ; the conversation it opens will not.
- Tag your last ten social posts as engagement content or sales content, then check each one's site sessions and orders against the tag.
- Bring one paid channel report (search ads, paid social or email) to Angles session 3, which reads paid and lifecycle channels in the same four beats.
Three questions
Clear all three and Angles session 2 gets stamped in your passport.
1 · The content team made an SEO change on Monday and asks on Friday whether it worked. What is the honest data-team answer?
The lag is part of how the channel works (Google Search Central, read 10 October 2026). A one-week read mostly measures noise, and ranking cannot be bought, so there is no shortcut.
2 · Followers rose 40 percent in a quarter. Why is that a vanity number on its own?
Follower growth can be real and still decide nothing. The company's own data, sessions and orders by post, says whether the audience is turning into customers.
3 · The cat video drew 41,000 views and 2 orders; the autumn launch post drew 8,000 views and 63 orders (constructed). What is the right read?
Engagement content and sales content both have a place. The mistake is judging one by the other's scoreboard, and the data team's job is to make the difference visible, not to set the content mix.
What this session teaches, and where it came from
Every fact on this page, and its verification tier, is recorded in the course's source map.
Angles session 2 cheat sheet · pin this
You can read the channels a company does not pay for by the click. Angles session 3 turns to the ones it does: paid search, paid social, email and lifecycle, affiliates, influencers and events, in the same four beats, where the platform that sells the ads is also the one that reports how well they worked.