Technical SEO prompt for marketers
Give a marketer's technical SEO prompt a fixed evidence table, a business goal and a limit on unsupported claims. Ask for priorities, dependencies and verification steps, then confirm every live signal with a page audit.
Updated 7 October 2026. Use the prompt with evidence you have actually checked.
Copyable prompt
You are helping a marketing team prioritise technical SEO work. Use only the supplied observations and business goal. Return: 1) issues supported by evidence, 2) issues that need a check, 3) recommended order with a reason, 4) copy or design implications, and 5) a verification test. Do not invent traffic, conversions, rankings, crawl stats, citations or competitor findings. Use plain language and keep technical assumptions visible.How to use it
Define the goal first
A lead-generation site, a documentation site and a shop may choose different priorities. State the page type, conversion action and release constraint before asking for a recommendation. The prompt can then distinguish a blocking template issue from a lower-priority improvement.
Use a small sample
Include representative URLs rather than a long export with no context. Note whether each page is a homepage, category, product, article or form. A repeated issue across different templates is more useful than ten rows from one template.
Make the brief reviewable
Ask for a short reason beside each priority and a test that another person can run. This makes it easier to reject a suggestion that sounds plausible but has no evidence. Keep the original observations in the ticket.
Validate the output
Run the anonymous Website Audit for the public sample, compare its returned values with the prompt input and update the brief. The prompt is an organising aid, not a substitute for a crawler or Search Console.
Questions
Should marketers paste a whole analytics export?
Usually no. Start with the evidence that answers the specific decision and remove personal or confidential data.
Can this prompt tell me what Google will rank?
No. It can prioritise observed technical work, not predict ranking outcomes.