[{"data":1,"prerenderedAt":75},["ShallowReactive",2],{"help-article-the-evaluate-stage":3},{"data":4},{"title":5,"excerpt":6,"content":7,"help_category":8,"seo":11},"The Evaluate Stage","How does News Manager decide which articles to draft? Learn how Evaluate scores relevance, the two ways you steer it, and why it runs only on rows you select.","\u003Cp>Evaluate is the AI relevance check that reads each article you select and decides whether it's worth drafting for your brand. It's a worked example of an \u003Cem>evaluative\u003C\u002Fem> prompt — one that judges rather than writes.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>What you'll learn\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cul>\n\u003Cli>What Evaluate decides, what it records, and how scoring works.\u003C\u002Fli>\n\u003Cli>The two ways you steer it: the evaluation prompt and excluded topics.\u003C\u002Fli>\n\u003Cli>Why it runs only on the rows you pick, and why your brand's voice is kept out of it.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Chr>\n\u003Ch2>The concept: a reader that knows your brand\u003C\u002Fh2>\n\u003Cp>\u003Ca href=\"\u002Fhelp\u002Fthe-news-manager-concept\">Scout\u003C\u002Fa> fills your queue with everything your sources publish. Most of it won't fit your brand. Evaluate is the step that reads through and separates the relevant from the noise, so you only draft from articles that match what your brand is about.\u003C\u002Fp>\n\u003Cp>Evaluate runs per brand and uses AI, so it needs your own AI key configured (\u003Ca href=\"\u002Fhelp\u002Fbyok-why-we-do-this\">bring your own key\u003C\u002Fa>). For each article you hand it, the model returns a relevance \u003Cstrong>score\u003C\u002Fstrong> and a decision, and Evaluate does one of two things:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>Approves it\u003C\u002Fstrong> — the article advances to a processing state, ready for drafting.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Skips it\u003C\u002Fstrong> — the article is set aside, with a short reason recorded on the item so you can read why it was rejected.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>That recorded reason matters. When Evaluate skips something, open the item and see why — so you can tell whether the check is reading your brand the way you intended. While it works, Evaluate also tags each article with the topics it detected, which makes the queue easier to scan and filter afterward.\u003C\u002Fp>\n\u003Cp>This is an \u003Cstrong>evaluative\u003C\u002Fstrong> prompt, and m18t treats it differently from a writing prompt: your brand's Voice, Rules, and Glossary are deliberately \u003Cem>not\u003C\u002Fem> injected into it. Only the brand's \u003Cstrong>Positioning\u003C\u002Fstrong> is — because a yes\u002Fno relevance decision should reflect what your brand is about, not how chatty or formal it sounds. See \u003Ca href=\"\u002Fhelp\u002Fthe-brand-ai-profile\">The Brand AI Profile\u003C\u002Fa> for why that split exists.\u003C\u002Fp>\n\u003Ch2>How you tune it\u003C\u002Fh2>\n\u003Cp>Evaluate has two controls, and they work together.\u003C\u002Fp>\n\u003Ch3>1. The evaluation prompt\u003C\u002Fh3>\n\u003Cp>Each brand has its own evaluation prompt (a \u003Ca href=\"\u002Fhelp\u002Fhow-ai-prompts-work\">prompt config\u003C\u002Fa> with the slug \u003Ccode>news-evaluate\u003C\u002Fcode>). This is where "relevant" is defined for the brand — who the audience is, what subjects to favor, what to treat as off-topic. The more specific the brand's \u003Ca href=\"\u002Fhelp\u002Fthe-brand-ai-profile\">Positioning\u003C\u002Fa> and this prompt are, the closer the approve\u002Fskip decisions land to your own judgement.\u003C\u002Fp>\n\u003Cp>If Evaluate keeps approving articles you'd reject (or skipping ones you'd keep), tune here first. Read the skip reasons on a few items to see how the model is interpreting your brand, then tighten the wording. The platform ships a tuned default prompt; you can \u003Ca href=\"\u002Fhelp\u002Fcustomizing-a-prompt\">customize it\u003C\u002Fa> if a brand genuinely needs a different judgement structure.\u003C\u002Fp>\n\u003Ch3>2. The excluded-topics list\u003C\u002Fh3>\n\u003Cp>Alongside the prompt, each brand has an excluded-topics list (managed via the \u003Cstrong>Categories\u003C\u002Fstrong> panel). Any topic you exclude is hard-rejected during evaluation: a REJECT clause is added to the prompt so an article on an excluded topic is skipped regardless of what the rest of the prompt says.\u003C\u002Fp>\n\u003Cp>Use exclusions for clean, absolute cut-offs ("never this topic, ever") and the prompt for the softer judgement calls ("favor this, downplay that").\u003C\u002Fp>\n\u003Ch2>Why it runs only on your selection\u003C\u002Fh2>\n\u003Cp>Evaluate uses your AI key, which means it costs you tokens. To keep that predictable, it runs only on the articles you select in the Feed Items tab — it never works through your whole queue on its own. Tick the rows you want checked, then click \u003Cstrong>Evaluate\u003C\u002Fstrong> in the selection bar. Scouted in a hundred articles but only want twenty evaluated? Evaluate twenty.\u003C\u002Fp>\n\u003Cp>You can also run \u003Cstrong>Evaluate\u003C\u002Fstrong> on a single discovered item from its row actions. If an item errors during evaluation (for example, the model returns an unreadable response), it's flagged as errored and you can \u003Cstrong>Retry\u003C\u002Fstrong> it; a crashed run that strands items in the "evaluating" state can be cleared with \u003Cstrong>Reset stuck\u003C\u002Fstrong>.\u003C\u002Fp>\n\u003Ch2>FAQ\u003C\u002Fh2>\n\u003Cp>\u003Cstrong>Why is the Evaluate button disabled?\u003C\u002Fstrong>\nIt needs your own AI key for the brand. Add an OpenAI, Google, or Anthropic key to the vault and it enables. See \u003Ca href=\"\u002Fhelp\u002Fbyok-why-we-do-this\">Why m18t uses BYOK\u003C\u002Fa>.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Evaluate keeps approving things I'd reject. How do I fix it?\u003C\u002Fstrong>\nSharpen the brand's \u003Ca href=\"\u002Fhelp\u002Fthe-brand-ai-profile\">Positioning\u003C\u002Fa> and the evaluation prompt so "relevant" is defined precisely, and add hard cut-offs to the excluded-topics list. Read the skip reasons to see how the model is reading your brand.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Why doesn't my brand's voice change how Evaluate scores?\u003C\u002Fstrong>\nBy design. Evaluate is a judging task, so Voice, Rules, and Glossary are kept out of it — only Positioning is used. Voice belongs in writing steps like \u003Ca href=\"\u002Fhelp\u002Fthe-generate-stage\">Generate\u003C\u002Fa>.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>An item is stuck on "evaluating" — what now?\u003C\u002Fstrong>\nUse \u003Cstrong>Reset stuck\u003C\u002Fstrong> in the toolbar to reclaim items stranded by a crashed run, then re-run.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Does Evaluate ever publish anything?\u003C\u002Fstrong>\nNo. It only approves or skips. Drafting happens later in \u003Ca href=\"\u002Fhelp\u002Fthe-generate-stage\">Generate\u003C\u002Fa>, and even that produces a draft you review.\u003C\u002Fp>\n\u003Ch2>What's next\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>See where Evaluate sits in the flow in \u003Ca href=\"\u002Fhelp\u002Fthe-news-manager-concept\">The News Manager Concept\u003C\u002Fa>.\u003C\u002Fli>\n\u003Cli>Understand evaluative vs. generative prompts in \u003Ca href=\"\u002Fhelp\u002Fthe-brand-ai-profile\">The Brand AI Profile\u003C\u002Fa> and \u003Ca href=\"\u002Fhelp\u002Fhow-ai-prompts-work\">How AI Prompts Work\u003C\u002Fa>.\u003C\u002Fli>\n\u003Cli>Approved articles move on to \u003Ca href=\"\u002Fhelp\u002Fthe-generate-stage\">The Generate Stage\u003C\u002Fa>.\u003C\u002Fli>\n\u003Cli>To rewrite the evaluation prompt itself, see \u003Ca href=\"\u002Fhelp\u002Fcustomizing-a-prompt\">Customizing a Prompt\u003C\u002Fa>.\u003C\u002Fli>\n\u003C\u002Ful>\n",{"slug":9,"title":10},"news-manager-help","News Manager",{"meta_title":12,"meta_description":6,"keywords":13,"canonical_url":14,"og_title":5,"og_description":15,"og_image":16,"robots":17,"json_ld":18},"The Evaluate Stage | m18t Help","evaluate stage, relevance scoring, evaluation prompt, excluded topics, positioning, selection only, byok, m18t","https:\u002F\u002Fm18t.com\u002Fhelp\u002Fthe-evaluate-stage","The AI relevance check that judges each article for your brand: what it records, the evaluation prompt and excluded topics, and why brand voice is kept out.","https:\u002F\u002Fm18t.com\u002Fog-default.png","index, follow",{"@context":19,"@graph":20},"https:\u002F\u002Fschema.org",[21,34,59],{"@type":22,"headline":5,"description":6,"articleSection":10,"inLanguage":23,"datePublished":24,"dateModified":24,"image":16,"author":25,"publisher":29,"mainEntityOfPage":32},"Article","en","2026-06-25",{"@type":26,"name":27,"url":28},"Organization","m18t","https:\u002F\u002Fm18t.com",{"@type":26,"name":27,"url":28,"logo":30},{"@type":31,"url":16},"ImageObject",{"@type":33,"@id":14},"WebPage",{"@type":35,"mainEntity":36},"FAQPage",[37,43,47,51,55],{"@type":38,"name":39,"acceptedAnswer":40},"Question","Why is the Evaluate button disabled?",{"@type":41,"text":42},"Answer","It needs your own AI key for the brand. Add an OpenAI, Google, or Anthropic key to the vault and it enables.",{"@type":38,"name":44,"acceptedAnswer":45},"Evaluate keeps approving things I'd reject. How do I fix it?",{"@type":41,"text":46},"Sharpen the brand's Positioning and the evaluation prompt so 'relevant' is defined precisely, and add hard cut-offs to the excluded-topics list. Read the skip reasons to see how the model is reading your brand.",{"@type":38,"name":48,"acceptedAnswer":49},"Why doesn't my brand's voice change how Evaluate scores?",{"@type":41,"text":50},"By design. Evaluate is a judging task, so Voice, Rules, and Glossary are kept out of it — only Positioning is used. Voice belongs in writing steps like Generate.",{"@type":38,"name":52,"acceptedAnswer":53},"An item is stuck on \"evaluating\" — what now?",{"@type":41,"text":54},"Use Reset stuck in the toolbar to reclaim items stranded by a crashed run, then re-run.",{"@type":38,"name":56,"acceptedAnswer":57},"Does Evaluate ever publish anything?",{"@type":41,"text":58},"No. It only approves or skips. Drafting happens later in Generate, and even that produces a draft you review.",{"@type":60,"itemListElement":61},"BreadcrumbList",[62,66,70,73],{"@type":63,"position":64,"name":65,"item":28},"ListItem",1,"Home",{"@type":63,"position":67,"name":68,"item":69},2,"Help","https:\u002F\u002Fm18t.com\u002Fhelp",{"@type":63,"position":71,"name":10,"item":72},3,"https:\u002F\u002Fm18t.com\u002Fhelp\u002Fcategories\u002Fnews-manager-help",{"@type":63,"position":74,"name":5,"item":14},4,1786312171377]