Templates
The seven built-in article structures, how to adapt one, and how the builder scores a custom template.
On this page
A template is the section list a draft is written to: what each section covers, roughly how long it runs, and what shape its output takes. Trakkr ships seven, and you can build your own.
Choose one in the composer's Setup step. Match the structure to the buyer question: a comparison may need a table, while a how-to needs clear steps. A useful structure makes the article easier to read, but choosing a template does not predict whether an AI engine will cite it.
Templates are listed at Content → Settings → Templates and managed in full on the Templates page.
The seven presets
| Template | Best for | Sections |
|---|---|---|
| Comparison Roundup | "Best X for Y" queries | 8 |
| How-To Guide | "How do I X?" queries | 7 |
| Authority Explainer | "What is X?" queries | 7 |
| FAQ | High-intent question clusters | 3 |
| Problem-Solution | "X not working" queries | 7 |
| Research Brief | Data and statistics queries | 7 |
| Case Study | "How [company] did X" queries | 7 |
Comparison Roundup and How-To Guide carry a Popular tag. Presets are the same for every brand and cannot be edited in place or deleted.
Open one and the detail page shows its section structure, its LLM-optimized features, and a citation rate of Medium, High, or Very High. That rating is a fixed label written into the preset, not a measurement of how your own articles have performed.
Choosing one
Match the template to the shape of the question, not the topic. "Best running shoes for flat feet" is a Comparison Roundup even though it is about shoes, and "What is a carbon plate?" is an Authority Explainer about the same product line. Ideas is the easiest place to see which shape you are chasing.
If you would rather be asked, Help me choose on the Templates page runs a short "What are you creating?" list and drops you on the matching preset.
What the presets share
Six of the seven open with a short summary section. Research Brief is the exception: it leads with Key Findings as bullets, which does the same job. All seven include an FAQ, optional in Research Brief and Case Study. Tables are not universal. Only Comparison Roundup (two) and Research Brief (one) carry one by default.
Those three are exactly what the product checks for. Open Customize on any preset and the LLM optimization panel ticks or crosses TL;DR, FAQ, and Table against your current section list.
Adapting a preset
Customize gives you a working copy of a preset without touching the original. Switch sections off with the checkbox, reorder them with the up and down controls. You cannot add a new section here; that needs the builder. Use configuration then drafts one article with the result, and Save as template keeps it under a name of your choosing.
Building a custom template
New template opens the template builder. It offers a pre-built structure to start from or Start from scratch, then lays the sections out on a canvas you can pan and zoom.
Nine section types are available: Introduction, TL;DR, Text Section, Bullet List, Comparison Table, Expert Quote, FAQ, Sources, and Image Block. Five of them (TL;DR, Bullet List, Comparison Table, Expert Quote, FAQ) count as snippetable, which is where most of the score comes from.
Select a section and a panel opens with four tabs.
| Tab | What it sets |
|---|---|
| Content | The reader-facing heading, the guidelines the model writes to, a word target, and whether the section is required |
| AI Rules | Output format (Prose, Bullets, Numbered, or Table) and a tone override for this section only |
| Conditions | Always include, or optional |
| Examples | Generates a worked example of this section for your brand |
Guidelines can carry variables in single braces, such as {brand_name} or {competitor}. Variables in the toolbar defines them, offers a few presets, and flags any you reference but never defined. Preview renders the structure, Test runs a trial generation, and Save stores the template.
Saved templates appear in the composer's template list beside the presets and under Your templates on the Templates page. They belong to your user account rather than the brand or the team, so a colleague signing in will not see them. Deleting one leaves articles already written from it unchanged.
The template score
The builder scores the structure out of 100 and labels it Needs Work, Good Progress, or LLM Optimized. Five factors, with fixed weights:
| Factor | Weight | What earns it |
|---|---|---|
| Snippetability | 30 | Share of sections that are TL;DR, FAQ, list, table, or quote. About two thirds scores full marks. |
| Structure | 25 | A TL;DR in the first two positions, plus an intro, an FAQ, and a sources section |
| Completeness | 20 | Every section has real guidelines rather than a stub |
| Word balance | 15 | A total target between 1,200 and 4,000 words, with no single section over 40% of it |
| LLM signals | 10 | An FAQ, a sources section, and section titles phrased as questions |
The panel lists specific fixes with the points each is worth. Clicking one either adds the missing section or opens the section that needs guidelines.
Common questions
Does the score change what gets written?
No. It reads the structure you have built and nothing else. It is a checklist for the shape of the page, so a high score on a template with vague guidelines still produces a vague article.
Can I edit a preset permanently?
No. Presets live in the product, identical for everyone. Use Customize for a one-off change, or Save as template to keep your version as a custom template.
What happens to existing articles if I change a template?
Nothing. A template is copied into the article when it is generated, so later edits to the template apply only to drafts made after the change.
Do templates control article length?
Only as a starting point. Each section carries a word target, but the composer has its own Length setting, and that is what the writer is briefed on for the article as a whole.
Why does my template score badly when it looks fine?
Usually snippetability. A structure made mostly of Text Sections reads well and scores low, because none of those sections are shapes a model lifts. Convert one to a table, add an FAQ, and put a TL;DR first.
None of the seven fit my article. What now?
Start the builder from the closest preset rather than from scratch. Every preset already satisfies the structure and signal factors, so you keep those while you change the middle.