Segments and Statistical Comparison via MCP

Analyse your data with your AI client · Segments and statistical comparison

Natural groups, and is the difference real?

Group rows into the segments the data itself suggests, and test whether the difference between two groups is statistically real.

Business Core and up Runs on your Network Agent Somente leitura, sempre

Teve alguma dúvida? É só perguntar. Aberto Suporte Em Fluxos de Consulta (menu no canto superior direito → Suporte) e envie-nos uma mensagem. Pessoas reais leem as mensagens, e "Não sei o que esta tela está pedindo" é uma pergunta perfeitamente válida.

What your Network Agent does for you

Your AI client can find the natural groups in a result and learn whether a difference is real or noise. Your Network Agent works over results it is holding: it groups rows by their numbers and describes each group, or compares two groups and returns the test it used, the p-value and the size of the effect. The rows never leave your network.

Perguntar	What comes back
“Group our customers.”	The number of groups that fits the data best, each with its size, a typical value per column and a profile, plus a fit score.
“Is the West really better than the East?”	The test used, the p-value, whether that is below your significance level, and the effect size with a label.

Situação	Before	With segments and comparison
Customer segmentation	Hand-made rules.	Groups the data suggests, described from their typical values.
An A/B result	Your AI client compares two averages.	A real test with a plain verdict.
“Is this region different?”	Opinion.	A number and a sentence.

The two groups for a comparison are either two held results on the same agent, or one held result split by two values of a column. The test is picked to suit the column: Welch’s t-test or Mann-Whitney for numbers, chi-squared for categories.

The plan it needs

Plano	This capability	Notas
Gratuito	Not on this plan	Works as it always has, with standard results.
Personal Core, Personal Plus	Not on this plan	Context management and Result analysis are included; Deeper analysis is not.
Business Core, Business Plus, Business Max	Incluído	Part of Deeper analysis.
Enterprise Core, Plus, Max	Incluído	Part of Deeper analysis.

Why it is on paid plans: these are analyses a chat client cannot do by reading rows. Your Network Agent runs them over the result it is holding, inside your network, and gives your AI client the groundwork to reason about: counts, scores, tests and examples.

The Free plan works as it always has. Paid plans add the Analytic toolset. Free is a plan you keep: MCP, the REST API, federated queries and alerts, with the same results as before. Plan names and prices are on the página de preços.

How to ask

You do not call tools yourself. Ask in plain words and your AI client picks the tool. Three asks that bring this capability in:

Claude

ChatGPT

Cursor

Código Claude+10 moreSomente leitura

Hold customers with their order count, average order and days since last order.

qs_run_query · hold

Group them into natural segments.

qs_result_segments

Is the average order in the West really higher than in the East?

qs_result_compare

Groundwork done on your Network Agent, inside your network Acessei a guia Atividade do MCP.

For people who read tool lists: qs_result_segments (a held result, up to 16 number columns, an optional number of groups) and qs_result_compare (two held results, or one with split_col and two values, mais metric_col and an optional test). See segments on the Tools tab, e compare.

What comes back

Claude

ChatGPT

Cursor

Código Claude+10 moreSomente leitura

Group our customers by how they buy.

qs_result_segments

The data splits cleanly into three groups:

group	clientes	orders/yr	avg order	last order
Regulars	1,240	24	310	9 days
Big occasional	180	3	4,900	70 days
Lapsing	2,790	2	220	210 days

The lapsing group is two-thirds of customers. Want to compare its order values with the regulars?

Is the West’s higher average order real?

qs_result_compare

Yes. Welch’s t-test gives p < 0.001, below 0.05, so the groups differ; the effect is medium. The West averages 342 against 298 in the East.

Groundwork done on your Network Agent, inside your network Acessei a guia Atividade do MCP.

Segments: Each group’s size, its typical value per column, and a profile in the same shape a profile uses.
Fit: A score from -1 to 1 for how cleanly the groups separate; higher is cleaner.
Test, p-value, significant: Which test ran, the chance of a difference this large if there were none, and whether that is below your significance level (0.05 unless changed).
Effect size: How big the difference is, with a label such as small or medium, separate from how sure the test is.

What it does not help with

Segments use number columns. Text columns need to be counted or coded first.
A significant difference is not a cause. The test tells your AI client the difference is real, not why.
Very small groups give weak answers, and each group needs at least two rows to compare.

Limites

Segments: from 2 groups up to your organization’s ceiling, 8 unless changed, picked automatically unless your AI client fixes the number. Up to 16 number columns.
Comparisons need both results held on the same agent.
Each analysis stops after 30 seconds; narrow the original query and hold it again.
The results must still be held: 60 minutes on Business plans, up to 8 hours on Enterprise.

Configurações

On the Deeper analysis card on the Context tab: Most groups (default 8), Comparison confidence (default 95%, a significance level of 0.05), and the Analysis tools list, where switching Segments ou Compare off refuses it for every key. A key’s own Analysis tools setting can narrow it further.

When your AI client says no

When the server refuses, your AI client receives a code and one sentence, and usually relays it. Quoted sentences are the server’s own words; the parts in italics stand for the ids a real message names.

Código	What the server says	O que fazer
`MCP_FEATURE_NOT_INCLUDED`	“Your plan does not include advanced analytics; anomalies, segments, comparisons and result diffs are available on Business Core.”	Ask your AI client to work from a profile, a follow-up query or rows instead, or move up to Business Core.
`MCP_FEATURE_NOT_INCLUDED`	“qs_result_segments is switched off for this organization; a workspace admin can turn it on.”	An admin turns it back on under Analysis tools on the Deeper analysis card of the Context tab.
`MCP_FEATURE_NOT_INCLUDED`	“qs_result_compare is switched off for this key; the key’s owner can allow it on the key’s settings.”	The key’s owner edits the key on the MCP page and adds it under Analysis tools.
`MCP_AGENT_ANALYSIS_DISABLED`	“analysis is switched off for the connector behind result {result id}; a workspace admin can allow it on the connector’s AI results settings.”	An admin opens the connector under Dados → Agentes de Dados and allows analysis in its AI results settings.
`MCP_RESULT_EXPIRED`	“result w_3f9a0c12d47be801 is not held (it expired, was released, or is unknown here); call qs_result_windows for the current list, or run the query again with hold.”	Ask your AI client to run the query again. That is a new query on your database.
`MCP_WINDOW_SQL_REJECTED`	“the analysis could not be expressed over the held result (a column named is not in the result, or its type does not fit the analysis).”	Name a column the held result has, of the right kind for the analysis.
`MCP_WINDOW_QUERY_TIMEOUT`	“the analysis over the held result did not finish within 30 seconds and was stopped; narrow the result (a WHERE clause on the original query, fewer columns) and hold it again.”	Hold a narrower result and ask again.
`MCP_AGENT_NOT_CONNECTED`	The Network Agent that owns the connector is not connected right now.	Aberto Dados → Agentes de Dados and start the agent on its computer.

Exemplos

“Who should the win-back campaign target?”

Your AI client segments customers and names the lapsing group: the customers whose last order was months ago, with their typical order size.

“Do our two carriers really differ on delivery time?”

The client splits held shipments by carrier and compares delivery days: a real difference with a medium effect, and no reliable difference on damage rates.

Ainda com dificuldades? Aberto Suporte Em Fluxos de Consulta, informe-nos em qual etapa você está e o que você vê. Nós o conectaremos.

Facts source

Setup trouble?

Agent shows Offline, or Test connection and Browse are greyed out? Check that the Agent is online → Is the Agent online?