All prompts
The big picturePublishedNeeds web search

Online Deep Dive

A broader and more forensic review of the business's public online presence.

Use this when the broad sanity check suggests several areas deserve closer investigation.

25–45 minutes1,503 wordsVersion 0.2 · updated 14 November 2026

Fill in your details and we’ll drop them in. It happens in your browser, and nothing you type is sent to us.

The prompt, explained

The boring details, for anyone interested in the engineering behind the prompt. If you just want to get on with it, here you go.

What it asks the AI to do

Explore this business's online presence in depth, starting from any earlier Online Sanity Check results I give you, while keeping the review connected to real customer behaviour.

The actual wording, line by line. Yes, we dressed it up as code.

  1. 1

    Language and tone

    Shared

    Keeps the answer clear, positive and understandable. It prevents the AI from hiding simple ideas behind jargon or presenting useful criticism as a fault-finding exercise.

    language-and-tone
    1

    Write in clear, unambiguous, plain English.

    2

    Assume the reader runs a business and may have no specialist knowledge of marketing, websites, search engines, analytics or social media.

    3

    Do not use jargon when an ordinary word or short phrase will do.

    4

    If a technical or specialist term is genuinely necessary, use the term, explain it immediately in plain English and say why it matters here. Explaining a term needs no source; add a reference only if you have opened one that genuinely helps the reader.

    5

    Keep the framing positive, balanced and constructive.

    6

    Where something genuinely is working, say so before describing the issue or risk that remains. Never invent a positive.

    7

    Do not soften a serious issue so much that it becomes unclear.

    8

    The aim is to reduce ambiguity and create clarity.

  2. 2

    Evidence and certainty

    Shared

    Stops the AI from presenting guesses as facts. It makes the report show what was actually found, what is a reasonable interpretation, and what could not be confirmed.

    evidence-and-certainty
    1

    Use publicly available information you can actually find.

    2

    For important factual claims, provide the source used to confirm them.

    3

    Label claims with exactly one of these: Observed (you saw it yourself in this conversation), Inferred (your reasoning from what you saw) or Not verified. Something you looked for but did not find, or saw only in a search-result snippet rather than on the page itself, is Not verified, never Observed.

    4

    Never present an inference as a confirmed fact.

    5

    Do not invent missing information.

    6

    Do not imply that you tested, visited, clicked, searched or verified something unless you actually did.

    7

    Do not treat absence from search results as proof that something does not exist.

    8

    Before calling information missing, open the page where a customer would expect to find it. If you could not open it, say so and do not make it a finding or an action. Before reporting a page as empty or out of date, check that you opened the main page, not a sub-page or filtered view.

    9

    Text a tool extracts from a page can include hidden content that visitors never see. Report a visible fault only if you are confident a visitor would see it.

    10

    If your tools cannot show part of a page, such as its code or anything behind a login or a block, say so. Do not report that part as missing or as present.

    11

    Cite a source where the fact it supports matters. Do not add a separate list of every source you used.

  3. 3

    Source hierarchy

    Shared

    Encourages the AI to use the most reliable explanation available instead of casually relying on marketing blogs or unsupported opinion.

    source-hierarchy
    1

    Prefer references in this order:

    2

    1. Official provider or platform documentation where the subject relates to a specific service or product.

    3

    2. A recognised neutral reference such as Wikipedia for general concepts.

    4

    3. Another authoritative source only when the first two are not suitable.

  4. 4

    Map the online presence

    Builds a simple map of where the business appears online before judging any individual channel.

    presence-map
    1

    Find the main public places where [BUSINESS NAME] appears online: website, Google Business Profile, major social accounts, directories, review platforms and other relevant listings.

    2

    Links I already know: [KNOWN LINKS]. Start from any I gave and expand from there.

    3

    Separate confirmed official profiles from likely or unverified matches.

    4

    Note obvious duplicates, abandoned profiles or conflicting representations.

  5. 5

    When someone searches for the business

    Looks at what a person who already knows the business name will encounter.

    branded-search
    1

    Search for [BUSINESS NAME] and common variations of that name.

    2

    Describe the main results a potential customer is likely to encounter.

    3

    Check whether official information is easy to distinguish from directories, reviews or unrelated results.

  6. 6

    When someone searches for the service

    Looks at discovery by need rather than by business name.

    service-search
    1

    Search for [MAIN PRODUCTS / SERVICES] using realistic customer language.

    2

    Include relevant local search combinations using [LOCATION].

    3

    Note whether the business appears, what kinds of competitors or directories dominate, and any obvious discovery gap.

  7. 7

    What role does the website play?

    Checks whether the website is acting as the useful centre of the online presence without turning this into the dedicated Website Audit.

    website-role
    1

    Website: [WEBSITE]. If there is one, assess whether it clearly supports discovery, understanding, trust and customer action.

    2

    Flag areas that deserve a dedicated Website Audit rather than performing that audit here.

  8. 8

    What role does social media play?

    Looks at whether the business's social accounts have useful jobs rather than simply counting posts.

    social-role
    1

    Identify the main social channels and the apparent role of each.

    2

    Look for useful recent activity, abandoned channels, duplication or unclear purpose.

    3

    Recommend a platform-specific audit only where there is a clear reason.

  9. 9

    Reviews and public reputation

    Looks for repeated public signals about customer experience without treating reviews as perfect evidence.

    reviews-reputation
    1

    Find relevant public review sources.

    2

    Summarise recurring positive and negative themes without over-weighting isolated reviews.

    3

    Note review recency, obvious response patterns and gaps worth investigating.

    4

    Do not infer the truth of disputed individual claims from reviews alone.

  10. 10

    Is the business represented consistently?

    Checks whether customers see the same basic facts and proposition wherever they encounter the business.

    consistency
    1

    Compare names, descriptions, addresses, phone numbers, opening hours, services, links and other important public information.

    2

    Show evidence for material inconsistencies.

  11. 11

    Follow the whole customer path

    Connects the separate channels into one realistic journey from need to action.

    customer-path
    1

    Follow one or more realistic customer paths from need → search → discovery → understanding → trust → action.

    2

    Identify where channels support each other and where customers may hit friction or dead ends.

  12. 12

    Where are the channels not working together?

    Looks for simple opportunities created by connecting things the business already has.

    missed-connections
    1

    Look for missing links between useful existing assets, such as social profiles not linking to booking, reviews not visible from the website, or listings pointing to outdated pages.

    2

    Prefer connecting existing assets before recommending new ones.

  13. 13

    Finding format

    Shared

    Gives every finding the same useful structure. The reader sees what is already working, what may need attention, the evidence, why it matters and exactly what can be done next.

    positive-finding-format
    1

    Report each finding using these headings, in this order:

    2

    - Finding: A short plain-English description of the issue or opportunity.

    3

    - Status: Shows whether the finding was directly observed, reasonably inferred, or could not be fully verified.

    4

    - What is working: Recognises useful things already in place so the review stays balanced. Leave this out if nothing genuine applies; never invent a positive.

    5

    - What may need attention: States the issue clearly and proportionately in plain English.

    6

    - Evidence: Shows the page, profile, search result, listing or other source supporting the finding.

    7

    - Why this matters: Connects the finding to a real customer or business consequence.

    8

    - How to put it right: Provides a practical step-by-step route to improvement rather than vague advice.

    9

    - References for the fix: Reliable guidance you actually opened that supports the fix. Leave this out if no reference adds anything.

    10

    - Effort: Exactly one of Small, Moderate or Larger, giving a rough sense of the work involved without inventing precise cost or time estimates.

  14. 14

    Recommendation discipline

    Shared

    Prevents the AI from defaulting to rebuilds, subscriptions or fashionable tools when a smaller change would solve the problem.

    recommendation-discipline
    1

    Prefer simple, realistic improvements over large projects.

    2

    Fix or improve what already exists before recommending replacement.

    3

    Do not recommend a paid product, subscription, redesign, rebuild or new platform simply because one exists.

    4

    Keep recommendations proportionate to the business and the evidence found.

    5

    Do not create an exhaustive improvement list when a smaller set of meaningful actions will do.

  15. 15

    Shape of the report

    Shared

    Long reports go unread. This keeps the answer to a length a busy owner will actually read, and stops the AI printing its own instructions as headings.

    report-shape
    1

    Sections such as Language and tone, Evidence and certainty, Source hierarchy, Recommendation discipline, Research boundaries and Limits of this particular prompt are instructions for you, not headings for the report.

    2

    Keep the whole report to about a seven-minute read, unless this prompt says otherwise. Make each point once and refer back to it rather than repeating it under later headings. Put minor points in one short list.

    3

    Cover every section this prompt asks for, even briefly. If a section has nothing useful to add, say so in a line rather than filling it.

  16. 16

    What deserves a deeper review next?

    Stops the Deep Dive becoming endless and points to the most useful specialist prompt.

    handoff
    1

    Identify no more than three areas that genuinely deserve deeper investigation.

    2

    Explain what additional question the deeper review should answer and why it matters.

    3

    This is a deeper review than most: keep the whole report to about a ten-minute read.

  17. 17

    Action prioritisation

    Shared

    Turns a long analysis into a small number of useful next steps. You should finish knowing what to do first, not merely knowing what is wrong.

    action-prioritisation
    1

    Finish with the three actions most worth considering first, drawn from the findings or ideas above.

    2

    For each action include what to do, why it comes first and the first practical step, referring to the finding it comes from rather than repeating it. Add a reference only if you opened one that genuinely helps with that step.

    3

    Prefer small, realistic improvements where they can produce a meaningful result.

  18. 18

    Something useful I learned

    Shared

    Adds the quiet educational layer. Instead of delivering a lecture, the prompt explains one useful marketing, customer-experience or technology concept that arose naturally from the work.

    teaching-note
    1

    Briefly explain one marketing, customer-experience or technology concept that arose naturally from this review.

    2

    Explain what it means and why it matters to this particular business, and, if you opened one, give a reliable place to learn more.

    3

    Keep this short and practical.

    4

    Do not turn the report into a lesson.

  19. 19

    Research boundaries

    Shared

    Defines what the AI must not do. It keeps a broad review from quietly becoming security testing, legal advice, invasive research or an unrelated technical audit.

    research-boundaries
    1

    Stay within the stated purpose of this prompt.

    2

    If something deserves deeper investigation, identify it, explain why it may matter, and recommend the appropriate deeper review instead of attempting that entire review here.

    3

    Do not perform vulnerability scanning, penetration testing, endpoint probing, login attempts or other security testing.

    4

    Do not submit forms, make bookings, place orders, contact the business or change anything.

    5

    Do not claim legal or regulatory compliance or non-compliance from a broad review.

    6

    Do not guess private information such as revenue, profitability, customer numbers, budgets, internal systems, staffing or business plans.

    7

    Do not infer motives, competence or intentions of owners, staff, agencies or competitors.

  20. 20

    Limits of this particular prompt

    The boundaries that apply to this prompt specifically, on top of the general research limits.

    scope-notes
    1

    This is broader than the Online Sanity Check but is still not a substitute for every specialist audit.

    2

    Do not perform deep code, security, accessibility, paid-media or legal analysis.

    3

    Use deeper-audit handoffs instead of expanding indefinitely.

SharedSections marked shared are word for word the same in every prompt that uses them. The rules on tone, evidence, recommendations and limits are deliberately identical wherever you meet them, so you only have to learn them once, and no prompt can quietly hold itself to a lower standard than the others.

What it deliberately will not do

  • This is broader than the Online Sanity Check but is still not a substitute for every specialist audit.
  • Do not perform deep code, security, accessibility, paid-media or legal analysis.
  • Use deeper-audit handoffs instead of expanding indefinitely.
visibilityreviewsreputationchannels

Ready to use it?

Fill in your business details once and we will write them into the prompt, ready to copy or download.

Make this prompt yours