What this checkpoint covers
The questions draw on the AI in marketing guide, the AI applications section and the marketing technology categories guide. They test judgement about strengths, limits and task selection, not the names of tools or features.
Task
- Write your answer and the reason for it before reading any explanation.
- Count an answer as right only if your reasoning matches.
- For each miss, reread the linked section, then revisit the recruitment agency task-sorting exercise.
Questions
- A florist's owner says AI will make her marketing better because it writes faster. What does faster drafting change, and what does it leave untouched?
- A team adopts a drafting tool and keeps exactly the same review process. Output rises sharply. What is the likely consequence?
- A manager says the team "doesn't use AI yet". The team runs automated bidding in its ad accounts and its email platform suggests send times. Is the manager right?
- A café chain is choosing its first AI task: headline variations for existing ads, or drafting a page on allergens from general knowledge. Which first, and why?
- A pilot reports that AI produced a large number of social captions in a week. Why is that a weak measure of success, and what would you track instead?
- A gym wants to paste its full member export into a free chat tool to write win-back emails. The output would be reviewed. What is wrong with the plan?
- A tool generates a long list of landing page headlines. The team wants to publish the one it likes best. What should happen instead?
Answers explained
-
It changes the cost of a first version, not what makes marketing work. Drafts, variations and summaries become cheap, so the slow part becomes checking. The offer, the audience, the truth of each claim and the channel economics stay with people. See what AI changes and what it does not.
-
More errors reach customers, faster. If production rises and the number of careful readings per piece does not, confident mistakes slip through under the brand's name. Review has to be redesigned when output grows, which is why the hub's adoption plan starts with rules and review rather than tools. See the first adoption plan.
-
No. Predictive automation has run inside ad platforms and other marketing software for years; the generative tools are the new part. AI features arrive inside the categories a stack already contains, as the marketing technology categories guide maps them, so the first governance step is to find them. See the three kinds of AI.
-
Headline variations. They are cheap to judge, low risk and based on copy that already exists. An allergen page drafted from general knowledge makes factual claims about safety that cannot be checked without redoing the research, and an error could harm a customer. Score both on value, review cost and risk and the order is clear.
-
Volume ignores review time and rejection. A caption that needed rewriting or was thrown away saved nothing. The applications section sets the pilot KPIA metric tied directly to an objective. as time per acceptable output, with the review counted in. See judging an application.
-
The input is the risk, whatever happens to the output. Personal data in a tool whose terms the gym has not read may leave its control. A win-back draft needs a few fields, not the export; where individual-level data is needed at scale, it belongs in the email platform or a customer data platform (CDPUnified first-party customer profiles.), not a consumer chat window. The risk question covers customer data in prompts.
-
Test them properly. Pick a few that differ in one element, run an A/B TestControlled comparison of two variants. with a hypothesis and a stopping rule, and read the result on the metric the page exists for. Generated variations are only useful when tested; see the experiment-design skill in the skills for marketers section.
Where to go next
If the strengths and limits questions caught you out, reread the AI in marketing guide and list which of your own weekly tasks are first drafts and which are decisions. If the task selection questions did, score a few of your own tasks with the method in the AI applications section. Otherwise return to the beginner level overview and start the module on AI for content and creative.
