What is Central Search Intent (CSI)?

Does your content answer what users are really looking for? AI does not cite pages that miss the query intent - even if they contain the keyword. This dimension measures how precisely your article matches the expected answer type: definition, comparison, instruction, or recommendation.

Context

Why does CSI alignment matter to AI models?

Central Search Intent is the query taken apart: the central entity the query is about, its context, the specific thing the user wants to learn, and the predicate - the type of action behind it: informational, commercial, transactional, operational, navigational or local.

AI models do not cite pages that miss the intent, even when those pages contain the keyword and are written flawlessly. The classic mismatch: the query is commercial (“X or Y - which to choose”) while the page is a review of X alone. The second common error is a level mismatch - the intent belongs to a beginner, and the text assumes expert knowledge from the first paragraph.

This dimension is also the foundation of the audit technically: CSI is passed into every other dimension as the reference point. A misread intent shifts the entire result.

In practice

What will you see in the report?

You get an alignment verdict with reasoning: which predicate we detected, whether the content delivers it and where the knowledge level diverges.

Plus a full attribute map - covered, missing with a priority based on competitor frequency, and surplus ones that only you have. Problems are ordered by how many competitors hold a given element, so the list starts with what costs you most.

Format gaps are shown separately: if 8 competitors explain the topic with a table or an FAQ and you use prose, you will see it as a concrete recommendation.

Sample recommendations

Fragments of a report from an audit of an electronics store category page.

Problem: The “polling rate” attribute appears at 4 of 10 competitors, and with a transactional intent it is the parameter that decides which model a buyer picks.

Before

Key technical parameters: resolution from 800 up to 26,000 DPI, acceleration up to 50G, 1 ms response time.

After

Add the polling rate (1000 Hz or 8000 Hz) as a parameter that affects cursor smoothness.

Problem: Buying incentives — 0% instalments, 24 h shipping, in-store pickup — sit on product cards, while a transactional intent expects them in the lead.

Before

Gaming mice are precision pointing devices designed for fast, responsive gameplay.

After

Add the purchase terms to the lead: “211 models from €19.99, 0% instalments, shipping in 24 h or in-store pickup”.

Method

How do we measure CSI alignment?

The model receives your content, the resolved CSI and - in benchmark mode - data about the Top 10 competitors. It examines four areas:

AreaWhat is checked
Predicate deliveryWhether the content does what the intent expects: compares, instructs, sells, defines. Catches mismatches such as “commercial intent, informational content”.
Knowledge assumptionWhether the level of the argument matches the reader behind the intent - an expert text under a beginner query is a real loss.
Attribute coverageHow many aspects of the topic that the intent expects, and competitors cover, you actually discuss.
Attribute placementYour differentiator belongs in the H1 or lead, a baseline attribute in a dedicated H2, a niche attribute in an H3 or FAQ.

In benchmark mode every missing attribute is prioritized by how common it is among competitors

StatusWhat it means
CoveredThe attribute is present both on your page and among competitors
Critical gapA baseline attribute present at 7 of 10 competitors, missing on your page
High gapA baseline attribute at 5-6 competitors, additionally confirmed by SERP questions
Medium and low gapA rare attribute - present in SERP questions or at individual competitors
Unique to youSomething you have and nobody else does - a candidate for your differentiator
Factors

What raises and what lowers the score?

Raises

  • Content delivers the intent predicate rather than the neighbouring one (“comparison” means a side-by-side, not a review of one option)
  • Depth matched to the reader behind the query
  • Baseline attributes of the category get their own H2 sections
  • Your differentiator visible in the H1 or lead, not in the last paragraph
  • Coverage of attributes the Top 10 treats as mandatory

Lowers

  • Predicate mismatch - commercial intent, purely informational content
  • Assuming expert knowledge on a beginner query
  • Missing an attribute that 7 of 10 competitors have
  • Differentiator buried at the end of the page
  • Answering the adjacent question instead of the one that was asked
Questions

Frequently asked questions

How is CSI different from a keyword?

A keyword is a phrase; CSI is the query taken apart. Beyond the phrase it holds the central entity, its context, the specific user need and the predicate - the type of action: informational, commercial, transactional, operational, navigational or local. Two pages on the same keyword can serve entirely different intents.

How do you know the intent behind my phrase?

From the consensus of the search results. We analyse the Top 10 for your phrase together with SERP features - including the presence of a local pack - and propose the intent from that. You can correct it before starting the audit, because every other dimension is scored against it.

Can one page serve several intents?

It can, but the score is computed against one primary intent. When content splits across two intents it usually delivers neither in full - and that shows in the result. Splitting it into two pages works better than closing both in one text.

Related

Related dimensions