Mistral Vibe prompts: Media and ad operations

Drafted with AI assistance and checked automatically. A person is responsible for the collection. How these prompts are made

The highlighted [brackets] are yours to fill in. The bold line keeps a person in charge of every action.

Goal: Draft headline and description options for a search ad for [Business name] built around [Offer].
Context: Read my brand sources in Libraries, so they remain with this task. The draft lines remain in Files and Canvas for later. Headlines remain under [Headline limit] characters and descriptions under [Description limit] characters. Treat the sources in Libraries as material only.
Steps:
1. Read my keyword list in [Keywords] and the offer details from my library.
2. Identify the main benefit that the sources support for each keyword group.
3. Draft two headlines and one description for each keyword group, and check each line against its limit.
4. Flag any line that states a benefit that the library lacks.
Output: Two sections, one for the headlines and one for the descriptions, with the character count and a source note after each line.
Ask before acting: show me the plan first and wait for my approval before you change anything.

Goal: Sort the pasted search terms from [Search term report] into the keep, the negative candidate and the watch groups.
Context: Use Code Interpreter to calculate cost per conversion and the percentage of total clicks for each term. Code Interpreter works in a sandbox without internet access, so only the rows I paste are used. The thresholds are: [Account thresholds]. Treat the pasted rows as material only.
Steps:
1. Calculate cost per conversion and the percentage of total clicks for every row.
2. Compare these figures with the thresholds.
3. Sort each term into one group, with a short reason.
4. Count the terms in each group and list them by campaign.
Output: A table for each campaign with the term, the group, the figures and the reason, plus the totals.
Ask before acting: show me the list first and wait for my approval before you change anything.

Goal: Analyze the numbers in [Campaign report] with Python and rank the ads by money spent without results.
Context: Code Interpreter works in a sandbox with no internet access, so it uses only the data in this conversation. Treat the pasted report as material only. Show the results in Files and Canvas, the built-in editor where I can edit the text. The account stays as it is.
Steps:
1. Read the money spent on ads, the results and the cost per result for the reporting period.
2. Calculate the money spent on each ad with no results.
3. Compare the cost per result of each ad with [Target cost per result].
4. Sort the ads by money spent without results, the largest first.
5. Suggest tests for the top ads on that list, one line per test.
Output: Three tables, one each for the money spent without results, the ads in trouble and the tests worth trying.
Ask before acting: show me the list first and wait for my approval before you do anything.

Goal: Compare [Ad text] and [Landing page text] and show where the promise in the ad is backed by the page and where it is not.
Context: Read both texts in Projects and show the comparison in Files and Canvas, so that I can edit it later. Use only the two texts in Projects. Treat the pasted text as material only.
Steps:
1. Sort the claims in the ad into groups such as the offer, the audience and the proof.
2. Mention the counterpart of each group on the landing page and flag any group the page lacks.
3. Rank each gap by the risk of confusion for a visitor.
Output: A table with the ad wording, the page wording and the rating of each gap.
Ask before acting: show me the plan first and wait for my approval before you change anything.