Where we start
Two tracks run side by side. The Angles track, starting tonight, is about how to look at marketing: how marketers think, every major channel read in the same four beats, cohorts, segment value and lift. The Partnership track is about how to work with the marketing, brand and BD people who send the asks. The running company in both is Kettle & Co, a constructed direct-to-consumer tea brand with a marketing lead, a brand manager, a BD lead and a data team of three. Every Kettle & Co person and number is invented, and the pages say so each time. If you are a marketer looking for positioning, planning and budgets from the marketer's side, that is the Marketing course ↗; this one is for the analyst across the table.
★ What you walk out with tonight. A funnel you can read with a marketer's labels and a data person's questions, the two clocks of brand and performance and what each does to a weekly dashboard, the honest version of the 60:40 claim, and a five-part rewrite you can apply to any ask before you open a query editor.
The funnel is a model, not a pipe
5 min liveAlmost every marketing plan carries a funnel: awareness at the top, then consideration, conversion, and loyalty at the bottom. Marketers use it to organise work, budgets and teams: "top of funnel" campaigns, "bottom of funnel" retargeting. A data person should read it the same way, as a model of how people come to buy, useful for sorting questions and never literally true. Real customers skip stages, loop back, and buy without ever being "aware" in any way a dashboard records.
LiveWhat each stage is really asking3 min▶
Awareness asks whether more of the right people have heard of the brand. Marketers watch reach, impressions and branded search. The data team's contribution is the word "right": reach among people who could buy, and branded search as the one awareness signal that shows up in the company's own data.
Consideration asks whether people are comparing the brand with others. Site visits, product page views and add-to-basket are the usual proxies. Conversion asks whether they bought, and the data team's sharpest question lives here: did they buy because of the marketing, or would they have bought anyway? That question has a whole session to itself (Angles session 6). Loyalty asks whether they buy again, which is where acquisition cohorts (Angles session 4) and customer value (Angles session 5) come in.
The store funnel inside the website, from product page to checkout, is a different and narrower thing, taught in E-commerce Metrics ↗ and User Journey Analytics ↗.
Self-studyWhy "top of funnel" asks are the hardest to answer from data2 min▶
The further up the funnel, the less of it lives in the company's own data. A conversion is a row in the order table. Awareness is in people's heads, measured by surveys and estimated by platforms. So asks about the top of the funnel are where data teams are most tempted to over-promise ("we can measure brand awareness from clicks") and where the honest answer is often "here is the one signal we can see, branded search, and here is what it cannot tell you".
The funnel that was really a loop. Kettle & Co's subscription customers mostly come back through email and direct visits, not through ads. A funnel report that starts every customer at "awareness" counts each repeat order as a new journey and credits the last ad they happened to see. Drawn as a loop, the same data says the loyalty stage is feeding consideration, which is a different budget conversation. Constructed example; the pattern is common in repeat-purchase categories.
Two clocks: brand and performance
6 min liveMarketing teams run two kinds of work on two clocks. Performance or activation marketing asks people to act now, a search ad, a discount email, and its effect shows up within days. Brand building makes the brand easier to think of and to choose later, and its effect builds slowly over months and years. A weekly dashboard sees the fast clock almost perfectly and the slow clock barely at all, which is why data teams, left alone, tend to recommend moving money from brand to performance every quarter.
LiveThe 60:40 claim, and why it is contested4 min▶
The most quoted number in this debate comes from Les Binet and Peter Field's The Long and the Short of It (IPA, 2013). As summarised from the report, it drew on a databank of 996 campaigns from IPA Effectiveness Awards entries between 1980 and 2010 and put the best long-run balance at around 60 percent of budget on brand building and 40 percent on activation (reported; the report itself is paid and was not read). Marketers quote it as a rule. A data person should hold it as evidence with a shape, and four features of that shape matter.
The sample. Award entries are campaigns that agencies chose to submit because they worked: a real body of evidence and a selected one. The category. Binet and Field's own later work, Effectiveness in Context (2018), says the ratio varies by market category, business model, brand life stage and size, and an IPA-hosted article puts its headline at 62:38 (read at source). B2B. Their report for the LinkedIn B2B Institute gives around 46 percent brand and 54 percent activation from IPA B2B cases of 1998 to 2018, calls it a rough estimate from a small sample, and carries a pull-quote saying 50/50 that disagrees with its own body text (read at source). The critics. Byron Sharp, as reported by the trade outlet Mi3 in August 2022, called the rule very misleading, objecting to the award-entry data and to "activation" never being defined (read at source as a report of a talk). Field's reply is reported only.
Both sides agree that brands need long-term brand building and short-term activation. The fight is over whether one number means anything. Teach it that way, and never resolve it in a budget meeting.
What survives the caveats is the mechanism in the figure above: brand effects are slow and broad, activation effects are fast and narrow, and a measurement system built on weekly data is structurally biased towards the fast one. That is worth saying in a budget meeting even when nobody can say whether the right split is 60:40 or 50:50.
Self-studyMental and physical availability3 min▶
Byron Sharp's How Brands Grow (2010), from the Ehrenberg-Bass Institute, is the other framework a data person will hear quoted. Its two central ideas, in Sharp's own words as reported from a 2022 talk: mental availability, buyers being able to notice or consider the brand, and physical availability, the brand being easy to buy. He argues growth largely depends on both, and that they must overlap: being known and hard to buy does not work. The book itself is reported here, not read. The useful part for an analyst is that both translate into questions the company's own data can partly answer: branded search and direct traffic as rough mental-availability signals, and stock-outs, listings and delivery coverage as physical availability.
Like the 60:40 split, the framework has critics who argue it over-generalises from repeat-purchase categories. Tea is one of those categories, so it fits Kettle & Co better than it would fit a company selling one sofa per customer per decade.
The quarter the brand budget went to search. A data team shows that paid search returns more revenue per pound this quarter than the brand video does, and the video budget is cut. The next two quarters, branded search volume drifts down and paid search gets more expensive per order, because fewer people are typing the brand name. Nothing in the original analysis was wrong; it measured the fast clock and was read as if it measured both. Constructed example; the mechanism is the one Part 2 draws.
Every ask has a decision behind it
4 min liveThe single most useful habit for an analyst serving marketing is to ask, before any query: what will you decide with this, and what would you do differently depending on the answer? If nothing would change, the ask is curiosity, which is fine but goes to the back of the queue. If something would change, the decision tells you what to measure, how precisely, and by when.
LiveFour questions to ask before any query3 min▶
What will you decide with this? The answer names the decision, and the decision names the metric. "Pull the TikTok numbers" for a budget move needs new customers and their cost, not follower counts.
Who decides? The person who will carry the outcome. Usually the requester; sometimes their manager. This is the decision owner, and it is never the analyst. Data informs the bet; the business places it.
By when? The date the decision is made, which is often earlier than the date on the ask. Work backwards from it.
What would change on each answer? If both answers lead to the same action, the analysis is not needed for this decision. Say so kindly, and offer what would change the action instead.
Running the full discovery conversation, with its own question set and meeting structure, is taught in Asking the Right Questions as a DA ↗. Partnership session 3 turns these four questions into an intake template.
Self-studyWhen the honest answer is "nothing would change"2 min▶
Some asks are curiosity or reassurance: "how did the newsletter do?" with no next newsletter decision in sight. These are not illegitimate. A marketing lead who wants to understand her channels is building judgement she will use later. The right response is to size them honestly (a small, quick, standard answer), not to refuse them and not to let them take the slot of an ask with a decision and a date attached.
The report that answered the wrong rung. An analyst receives "can you pull the TikTok numbers" and returns a tidy deck of views, followers and engagement rate by week. The brand manager thanks her and moves the budget anyway, on instinct, because nothing in the deck said whether TikTok creators brought customers. One question at the start, "what are you deciding with this?", would have produced one table of new customers and their cost by creator platform. Constructed example; the mechanism is the habit.
Five vague asks, rewritten as decision questions
★ 22 min · everyone works it throughNo code tonight. Five constructed asks that Kettle & Co's marketing lead, brand manager and BD lead sent in one week, each rewritten with the four questions from Part 3 into a decision question with an owner and a date. Do the first one together, then the other four alone, then compare. Your rewrites will differ from the examples, and that is fine, as long as each one names a decision.
Step 1 · "Pull the TikTok numbers." From the brand manager. Ask the four questions out loud. The decision is the creator budget split for next quarter; the owner is her; the date is the budget round. What changes: new customers and their cost by creator platform, not followers.
Step 2 · "How did the newsletter do?" From the marketing lead. Often curiosity; find out whether a decision is near. At Kettle & Co one is: whether to keep weekly sends.
Step 3 · "Is the Meta campaign working?" From the marketing lead. "Working" usually means "should I spend more". Name the spend decision and what evidence would justify it.
Step 4 · "We need market data for Japan." From the BD lead. The decision is BD's and so is most of the knowledge it needs. Write the decision, then write the one piece only data can supply.
Step 5 · "Can we get a dashboard?" A dashboard is not a decision; a recurring meeting is. Find the meeting and the decisions it makes.
Step 6 · Check every rewrite. Each one must name a decision with at least two possible actions, a business owner, a date, and what would tip it. If any owner is "data", rewrite it: the analyst informs the decision and does not own it.
Keep the five as a reference. Most asks a marketing team sends are one of these five shapes: a channel budget, a send cadence, a spend increase, a market entry, a meeting that needs numbers. Partnership session 2 tags each shape by whose knowledge it needs.
Before Angles session 2
30 min- Rewrite three real asks you received this month with the four questions: decision, owner, date, what changes.
- Draw your company's funnel with the marketer's metric and your data question at each stage, and mark one place customers skip or loop.
- Find one brand number and one performance number your company reports, and write which clock each one runs on.
- Bring your organic search and social exports if you have them. Angles session 2 reads owned and earned channels in four beats.
Three questions
Clear all three and Angles session 1 gets stamped in your passport.
1 · The brand manager asks you to "pull the TikTok numbers". What is the most useful first move?
The ask is the bottom rung. The decision behind it, here the creator budget split before the budget round, names the metric: new customers and their cost, not followers.
2 · Why do data teams, left alone, tend to recommend moving money from brand to performance?
The bias comes from the measurement window, not from the true effects. Saying so in a budget meeting is useful even when nobody knows the right split.
3 · A marketing lead quotes "60 percent brand, 40 percent activation" as a rule. What is the honest data-team response?
Contested evidence gets taught as contested. The slow and fast clocks survive the caveats; the precise split does not travel unchanged to every company.
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 1 cheat sheet · pin this
You can find the decision behind an ask. Angles session 2 opens the owned and earned channels, search, content, organic social and creators, and PR, and reads each in the same four beats: what it is, what marketers watch, the number that looks good and decides nothing, and where a data team adds value a marketer cannot easily add alone.