# Opening style analyser: how it is scored and how to change it > Opening style is how a piece typically starts: hook, context, direct point, anecdote or question. Canonical: https://scriptgrain.com/attributes/opening-style # Opening style analyser *By Jack Stovell ยท published 2026-09-24* Opening style is how a piece typically starts: hook, context, direct point, anecdote or question. ScriptGrain reads it as one of those five labels from your writing samples, and a model judges the call rather than any script counting words. It sits in the Rhetoric group and carries a weight of 0.75 in the voice match score. ## Analyse your opening style free Your free voice profile measures opening style among all 45 attributes, from a sample of your own writing. No card needed. [Try the six-signal check first](https://scriptgrain.com/#sg-measure) ## At a glance - API field: **opening_style** - How it is measured: **Judged by an AI model** - Group: **Rhetoric** (ScriptGrain voice profile, 45 attributes in 8 groups) - Scale: **hook, context, direct, anecdote or question** (ScriptGrain attribute definitions) - Voice match: **Scored, weight 0.75** (ScriptGrain voice-match engine (published method)) ## What opening style measures Opening style measures the move a writer makes in the first sentence or two, not the topic. Five labels cover it: hook, context-setting, a direct point, an anecdote, or a question. So a piece that opens "Cut the first paragraph of every draft" is direct. One that opens with a scene from last Tuesday is an anecdote. One that opens "Why does every brief start with the same sentence?" is, well, a question. The attribute lives in the [Rhetoric group](https://scriptgrain.com/attributes/rhetoric) of a ScriptGrain voice profile, six attributes that include [argument structure](https://scriptgrain.com/attributes/argument-structure). Opening style pairs naturally with closing pattern, the other bookend. Here's the thing about labels like this: they're judged, not counted. There's no piece of code scanning your first ten words looking for a question mark. Claude Sonnet 5 reads each sample in Pass 1, forms a view per piece, then synthesises one label across all samples in Pass 2. That's a judgement call, and judgement calls drift. ScriptGrain's study SGR-003 found that the 13 enumerated labels in profiles built from fewer pieces matched a 15-piece profile 72% to 87% of the time, depending on piece count (82% at one piece, 87% at eight). Labels are the least stable part of a profile. Worth knowing before you treat any single label as gospel. ## How opening style is scored Opening style moves your voice match score at a weight of 0.75. When a draft gets compared against your profile, Claude Haiku judges the label from the draft itself, and code compares that judged label against what's stored in your profile. Exact match scores 1. If either side says "mixed" or "varied", that's 0.7. If your profile's value contains the draft's label, that scores 0.9. Anything else, 0.15. For how the 28 attributes that can move the score combine into one number, and where the 90%+ Excellent, 75-89% Good, 60-74% Fair, under 60% Low bands come from, see the [voice match](https://scriptgrain.com/glossary/voice-match) glossary entry and the [voice measurement framework](https://scriptgrain.com/reference/voice-measurement-framework). ## How to analyse opening style in your own writing You can do this by hand in about ten minutes with five or six of your own pieces. 1. Pull the first one to two sentences from each piece, nothing more. 2. Sort each into one of five buckets: hook, context, direct point, anecdote, question. 3. Count which bucket wins. That's your default. 4. Check for a second pattern: a fallback you use when the default doesn't fit. 5. Ask whether the pattern matches your intent, or whether it's just habit. That's the manual version. ScriptGrain's free voice profile reads opening style, among all 45 attributes, from a writing sample you upload, no manual sorting required. ## Examples of opening style in real writing Different openings do different jobs, and seeing them side by side makes the difference obvious. Example (hook): "Most people write their opening line last. That's backwards." Example (context): "Sales pages used to lead with the product. Now they lead with the reader's problem." Example (question): "Why do so many articles open with a sentence nobody needed?" None is better in the abstract. The point is knowing which one you actually reach for. ## How to make AI open the way you do Getting an AI to open the way you do takes specific instructions about this one attribute. 1. Name your opening type in the brief: hook, context, direct point, anecdote or question. 2. Paste two of your real first paragraphs as examples of the move, not as words to copy. 3. Ask for three first lines in your style and pick one yourself. A ScriptGrain profile handles this by sending your `opening_style` value inside the voice profile JSON block of the system prompt, under an instruction to treat every attribute as a hard constraint, not a suggestion. That profile isn't locked to one chat window either: it reaches ChatGPT, Claude and other tools through the API and the MCP server, so the same label travels with you. ## How to stop AI opening with a stock line Stopping the stock opener starts with banning it outright, by name. 1. Ban stock openers by name, such as "In today's world" and "In an age where". 2. Ask the model to start with the first sentence that actually carries information. 3. Delete the draft's opening paragraph and check whether anything was lost. Often, nothing was. When ScriptGrain writes from a profile, these openers are banned unless they appear in the profile's own lists. [Transition style](https://scriptgrain.com/attributes/transition-style) covers the same fix for stock connectors between paragraphs. ## Questions ### What is an opening style analyser? It's a way of identifying which of five moves a piece of writing opens with: hook, context, direct point, anecdote, or question. A ScriptGrain voice profile does this by having an extraction model read your samples and judge the dominant pattern, rather than counting anything mechanically. The result gets stored as one label and used later when checking how closely a new draft matches your voice. ### What are the types of openings in writing? Five, by ScriptGrain's scale: hook, context-setting, a direct point, an anecdote, or a question. A hook provokes attention before making a claim. Context orients the reader first. A direct point states the argument immediately. An anecdote opens with a small story. A question invites the reader to think before you answer it yourself. ### How do I write a strong opening line? Pick your move deliberately rather than defaulting to habit: hook, context, direct point, anecdote or question. Then write the sentence that does that job in the fewest words possible. Read it alone, with no paragraph beneath it. If it still means something and still earns attention, keep it. If it's just throat-clearing before the real sentence, cut it. ### How is opening style scored in a voice profile? It's judged, not counted: Claude Haiku reads a draft's opening and assigns a label, code compares that against your stored profile value, and the match feeds into voice match at a weight of 0.75. Exact matches score 1, "mixed" or "varied" scores 0.7, a broader profile value containing the draft's label scores 0.9, and anything else scores 0.15. ### How do I stop AI writing "in today's world"? Ban the phrase explicitly in your brief, along with cousins like "in an age where". Then ask the model to open with the first sentence that carries real information, not throat-clearing. ScriptGrain does this automatically when it writes from a profile: stock openers are banned unless they appear in the profile's own lists. ## Related - [Rhetoric attributes](https://scriptgrain.com/attributes/rhetoric) - [Closing pattern](https://scriptgrain.com/attributes/closing-pattern) - [Metaphor usage](https://scriptgrain.com/attributes/metaphor-usage) - [Repetition as emphasis](https://scriptgrain.com/attributes/repetition-as-emphasis) - [The anatomy of a writing voice: 45 measured attributes across 8 layers](https://scriptgrain.com/reference/writing-voice-attributes) ## Sources - [How ScriptGrain scores voice match](https://scriptgrain.com/reference/voice-measurement-framework) - [How many samples does a voice profile need before it stabilises? (SGR-003)](https://scriptgrain.com/research/how-many-samples-does-a-voice-profile-need)