Manual Focus

James and Hannah walk through the first Manual Focus AI Shortlist study: six AI engines, 309 UK questions on smart trainers, training apps and smart bikes, each asked three times, for 5,562 answers. Wahoo leads trainers at 83% and Zwift leads apps at 88%, with Wahoo and Wattbike close on smart bikes. But the leader changes with the buying moment, the engines disagree on challenger brands, the same question gives the same brands only 20% of the time, and cycling media supplies over half the citations. They finish with three jobs to hand to AI and what the study can't tell you. The interactive report and a free one-page AI Shortlist card for brands in the study are on manual-focus.co.uk.

Read the article and get the prompts: Which indoor cycling brands does AI recommend in the UK?

A Manual Focus Basics episode. James and Hannah explain above the line (mass-media advertising that builds memory), below the line (direct, targeted activity that prompts action) and through the line (both, under one idea). They cover where the terms come from and the origin story they couldn't verify, why digital sits on both sides of the line, what Binet and Field's IPA work says about short-term metrics, how the split looks in B2B, and who should own the balance between memory and action.

Read the article and get the prompts: Above the line vs below the line marketing, explained

James and Hannah read through the new clause in Meta's Commercial Terms, which takes effect on 30 October 2026 and lets Meta crawl the websites associated with a business's products. They cover why the UK English edition names AI training when the US edition doesn't, which teams carry the exposure, what each of Meta's four crawlers does, why blocking one of them is only a partial control, and three jobs to hand to AI before the deadline.

Read the article and get the prompts: What Meta's new crawling terms mean for UK advertisers

A Manual Focus Basics episode, and the foundation for our three earlier pieces on owned, paid and earned channels in the AI era. James and Hannah explain owned media (what you control), paid media (attention you buy) and earned media (what others choose to say about you), starting from Forrester's 2009 report No Media Should Stand Alone. They cover Seth Godin's permission marketing, why a social profile is closer to rented than owned, Gini Dietrich's PESO model, the UK CAP Code rules that keep paid content identifiable, and how to measure each type on its own terms.

Read the article and get the prompts: Owned, paid and earned media, explained

A Manual Focus Basics episode. James and Hannah go through public relations from the start: the CIPR's definition, the disciplines PR covers, how it differs from advertising, and what makes a story worth covering, with Seth Godin's Purple Cow as the test. Hannah takes listeners through Muck Rack's State of Journalism 2026 survey of 897 journalists, including the 88% who disregard pitches that miss their beat and the four things journalists value in a pitch. They cover how AI has changed both sides of the inbox, why press releases have changed jobs rather than died, and how to measure PR by whether opinion moved.

Read the article and get the prompts: PR 101, a basics guide to public relations

A Manual Focus Basics episode. James and Hannah walk through SOSTAC, the six-stage planning model PR Smith created in the 1990s: Situation, Objectives, Strategy, Tactics, Action and Control. They cover what each stage answers, how to tell strategy from tactics, why most plans fail in the Action stage, and why Control is where a plan gets updated rather than scored. Hannah brings in PR Smith's own 2025 advice on AI and SOSTAC, that any AI-generated plan is a first draft only, and they map the stages onto the three layers of AI.

Read the article and get the prompts: SOSTAC explained, a basics guide to marketing planning

A Manual Focus Basics episode. James and Hannah separate four words that marketing teams use interchangeably: strategy, the choice of who you serve and why they should pick you; a campaign, a time-bound push that carries part of the strategy; creative, the idea the audience meets; and execution, the delivery. Hannah explains three of Seth Godin's ideas from first principles, the smallest viable audience, "people like us do things like this" and Purple Cow, and they bring in Binet and Field's IPA work on short-term metrics. They finish with how to walk back up the chain when a result disappoints.

Read the article and get the prompts: Strategy vs campaign vs creative vs execution, explained

The first episode of the Manual Focus Basics series. James and Hannah explain the three layers of AI from first principles: large language models and chat, agents that take a whole request and deliver it, and orchestration, where several agents run as one system. Hannah covers how a language model predicts text, why the context window matters and why models can be confidently wrong, then Anthropic's distinction between workflows and agents and its published results and costs for a multi-agent system. James grounds each layer in how marketing teams work, why the dull jobs make the best first agent jobs, and why orchestration is a leadership question.

Read the article and get the prompts: The three layers of AI, a basics guide for marketers

James and Hannah talk through the five link changes Google announced for AI Mode and AI Overviews on 15 October 2026. The biggest one for brands is a test: a section at the top of some answers for the one or two official websites that can complete a task, such as a return or a booking. They cover what counts as official, who should care about Preferred Sources, why UK teams should watch the creator test rather than act on it, and three jobs to hand to AI this month.

Read the article and get the prompts: What Google's new AI Mode links mean for brand websites

James and Hannah go through the AI ad tools Meta and TikTok announced at Advertising Week New York on 5 and 6 October 2026, and sort what UK paid social teams can use now from what's still a test. TikTok Search Ads is live in the UK, Meta's video generation from still images is generally available with no markets named, and in-app checkout on both platforms has no UK date. They cover the ads MCP servers that let outside AI agents work inside ad accounts, why both platforms' performance claims need a holdout test, what checkout means for your product data, and three jobs to hand to AI.

Read the article and get the prompts: Meta and TikTok AI ad tools from Advertising Week, a UK guide