Build an SEO Improvement Team with Grok Bot: 5 Responsibilities + 7 Copy-Paste Prompts

Input business context and page URL → use Grok Bot across 5 responsibilities to produce evidence pack, opportunity comparison, brief & draft, QA review, and publication record → output verifiable page improvement brief and change log

Who

Content leads or small teams who need structured SEO improvements for existing pages

You get

A reproducible 5-responsibility workflow with 7 copy-paste prompts for Grok Bot, producing reviewed improvement briefs and publication records

You need

Grok Bot

Steps

1. Kickoff: Define the first improvement assignment

   Open a new Grok Bot conversation and tell the Bot about the business, target audience, and available source files and approved URLs for the page you want to improve. Ask the Bot to first identify the deliverable, evidence needed, reviewer, and completion checks before starting specific work. Do not set 'increase organic traffic' as the task goal, since that is an outcome influenced by many factors, not a task the Bot can verify in a single session.

   Execution prompt · Source text。Grok Bot conversation input, as the kickoff instruction for the first improvement assignment

   Copy prompt

   PROMPT — Help me prepare one SEO page improvement for [business and audience]. Available sources are [files and approved URLs]. First identify the deliverable, evidence needed, reviewer and completion checks. Keep recommendations separate from observations. Do not change the live website. If a required source is missing, explain which part of the task depends on it and continue independent work.
   The point is to create an inspectable workflow before giving it a schedule.

2. Prepare shared brief and source index

   Create a concise business brief with actual product capabilities, customer segments, important pages, approved terminology, and examples of good writing; also list audiences you do not serve, features you do not offer, and claims you cannot support as exclusions. Maintain a source index recording file names, dates, owners, and refresh rules. Keep raw exports separate from working copies. For small internal projects, create inputs, briefs, drafts, reviews, and published-changes folders, and use a stable page identifier in every related file.

3. Responsibility 1: Prepare evidence pack

   Assign the Bot the 'evidence preparation' role: collect and validate material needed by others, without making publication decisions. Ask the Bot to preserve measurement definitions, identify missing data, and give every source a reference. You need to provide search exports, crawl results, and page content to the Bot.

   Execution prompt · Source text。Grok Bot conversation input, to generate a reconcilable evidence pack

   Copy prompt

   PROMPT — Prepare the evidence pack for page [URL]. Inspect the supplied search export, crawl and page content. Record dates, filters, row grain and source locations. Calculate relevant totals reproducibly. List data gaps and possible mismatches. Return a short evidence note and a table another reviewer can reconcile. Do not invent missing conversions or explain changes without evidence.
   Check its work against a few original rows. Confirm that the correct property and date range were used. An elegant brief built on the wrong export is still wrong.

4. Responsibility 2: Compare opportunities and recommend action

   Give the validated evidence pack and business constraints to the Bot, assigning it the 'opportunity selection' role: identify what the reader needs and whether the proposed page can genuinely deliver it. Ask the Bot to compare candidate opportunities across audience relevance, evidence strength, expected usefulness, and implementation effort, recommending one action, one alternative, and one reason to wait.

   Execution prompt · Source text。Grok Bot conversation input, to compare opportunities and output recommendations

   Copy prompt

   PROMPT — Compare these candidate opportunities using audience relevance, evidence strength, expected usefulness and implementation effort. Explain the tradeoffs in plain language. Recommend one action, one alternative and one reason to wait. Do not use a model-generated score as proof of business value. Identify the evidence that would reverse your recommendation.
   Keep the recommendation small. Refreshing a clear section can be better than commissioning a new long article. Sometimes the right action is to improve navigation or repair an inaccessible page before creating more content.
   When live search results are part of the evidence, record the query, location, date and collection method. Results vary. If access fails, mark the observation unavailable; do not replace it with an imagined search-results page.

5. Responsibility 3: Write page brief and draft

   Give the approved opportunity to the Bot, assigning it the 'brief and draft' role. The brief should name the reader, problem, evidence, useful asset, and next action, including what must stay unchanged on an existing page. Then ask the Bot to produce a draft using only approved facts, flagging gaps in a separate review note.

   Execution prompt · Source text。Grok Bot conversation input, to generate page brief and first draft

   Copy prompt

   PROMPT — Write a page brief from the approved opportunity. Include a direct answer, a useful outline, evidence for each material claim, an original example and the asset a person needs to provide. Then produce a draft using only approved facts. Flag gaps in a separate review note. Do not turn a hypothesis into a customer result.
   The original asset matters. It might be a product walkthrough, a calculation, a comparison based on documented criteria, or a small template the reader can actually use. Another generic paragraph about the importance of efficiency does not make the page distinctive.
   Ask for several simple titles, then choose the one that accurately promises the page's value. A title should help the reader recognise their task. It should not imply that the author has tested a workflow that has only been proposed.

6. Responsibility 4: Independent draft review

   Give the draft, brief, and original sources to the Bot, assigning it the 'independent review' role: identify defects, not praise the writer. Ask the Bot to review factual support, product accuracy, reader usefulness, internal consistency, and clarity of the next step. For each problem, quote the short affected phrase, name the source or missing evidence, and propose a correction.

   Execution prompt · Source text。Grok Bot conversation input, for independent draft QA

   Copy prompt

   PROMPT — Check this draft against its brief and evidence. Review factual support, product accuracy, reader usefulness, internal consistency and the clarity of the next step. For each problem, quote only the short affected phrase, name the source or missing evidence, and propose a correction. Mark unresolved items explicitly. A fluent draft is not automatically an approved draft.
   Use deterministic checks where possible: URL existence, calculations, missing sections and duplicate headings. Reserve judgement for questions that need interpretation. A second assistant agreeing with the first is not a substitute for checking the source.

7. Responsibility 5: Record publication and changes

   After human approval and publication, assign the Bot the 'change tracking' role: record what actually changed, and do not mark a draft as live just because the writer finished it. Ask the Bot to save publication time, changed sections, approved hypothesis, baseline period, and next review date; compare the live page with the approved version; and report missing images, broken links, or unexpected differences.

   Execution prompt · Source text。Grok Bot conversation input, to log publication and changes

   Copy prompt

   PROMPT — Record the verified publication of [URL]. Save the publication time, changed sections, approved hypothesis, baseline period and next review date. Compare the live page with the approved version. Report missing images, broken links or unexpected differences. Do not claim performance improvement before measurement exists.

8. Handoff inspect: confirm input version and outstanding items

   Before starting each new assignment, use this prompt to have the Bot inspect the handoff: confirm which input version and accepted decision govern the work, identify only missing items that could change the output, continue sections independent of those gaps, and leave a precise handoff for remaining work. Use with a short handoff form (page ID, task, input version, accepted decision, output file, outstanding issue, next owner) to prevent rework.

   Execution prompt · Source text。Grok Bot conversation input, for handoff inspection before each assignment

   Copy prompt

   PROMPT — Before beginning this assignment, inspect the handoff. Confirm which input version and accepted decision govern your work. Identify only missing items that could change the output. Continue sections that are independent of those gaps, and leave a precise handoff for the remaining work.
   Source notes: Original workflow informed by Klaym AI's AI Marketing Field Guide and Grok marketing resources. Product access and available integrations must be checked in the current Grok Bot account. This is a proposed SEO operating method, not a ranking guarantee or a claim of measured customer results.
   A compact handoff preserves accepted decisions and makes the remaining work visible.

Tell Grok Bot your business context, target audience, available source files and approved URLs, and ask it to first identify the deliverable, evidence needed, reviewer, and completion checks before establishing an inspectable improvement workflow.