NEWClaude + Improvado, connected

Your marketing data,
one question away

Ask Claude about all your marketing data - campaign performance, pipeline attribution, budget pacing - and get governed, source-cited answers in the app your team already uses.

Thanks! Routing you to the right person now…

Enter a valid work email to continue.

improvado-mcp

ClaudeImprovado connected · shared memory · 1,002 sources

Which paid channels drove the most pipeline last week?

Reading your connected data… ready

$148.2K	spend	$1.31M	pipeline · 8.8x
$92.6K	spend	$704K	pipeline · 7.6x
$61.4K	spend	$319K	pipeline · 5.2x

$2.33Mpipeline · last 7 days

8.8xbest ROAS · paid search

3sources · one answer

+Reply to Claude…

Marketing pulselive artifact · refreshes every 2m

Spend

$302K

Pipeline

$2.33M

ROAS

7.7x

Sources live

1,002

command center · ready

Q3 pipeline review · refreshed 2m agoOn track

Google Ads overspend alert · -8% CAC targetFlagged

LinkedIn ABM segment · 34 accounts engaged+12%

Claude · Improvado connector1,002 sources synced · 3 anomalies flagged

Session wiring · improvado-mcp

$ claude mcp add --transport http improvado \
    https://mcp.improvado.io/v1
✓ connected · OAuth workspace token · 214ms handshake

{ "mcpServers": { "improvado": {
    "url": "https://mcp.improvado.io/v1",
    "scopes": ["read:spend", "read:crm", "write:memory"],
    "memory": "shared", "audit": "per-query" } } }

toolsmcp.query · mcp.schema · mcp.memory.recall · mcp.memory.write · mcp.audit

schema1,002 sources → one governed semantic layer · versioned · typed

identityrole-based scopes per seat · SSO · SCIM

complianceClaude sees only what the person asking is allowed to see - every query runs under their own SSO role, is logged read-only, and no raw credential ever reaches the chat.

latencyp50 588ms · p95 1.9s · cold cache on first call only

lineageevery answer cites source table + field + query id · traceable to the raw row

piifield-level masking at the MCP layer · no raw PII enters model context

rate limit120 req/min per seat · burst 20 · 429 backs off with Retry-After

One MCP endpoint · no per-platform keys · median answer 588ms

How it works

From question to answer in three steps

Connect once, ask in plain English, and Claude remembers what your team taught it - no setup project and no query language to learn.

Connect

Connect Improvado to Claude once - your admin sets it up, nothing for you to install.

PII masked · SOC 2 Type II · EU residency

Set up once at onboarding - nothing for your team to install or maintain.

Ask anything

Ask about spend, pipeline, CAC, attribution - Claude queries live, normalized data across every connected source.

Which channels deserve more budget?

led with $1.31M in pipeline at 8.8x return ·

brought $704K at 7.6x.

And vs the week before?

Pipeline up +9.4% WoW - paid search led the gain; Meta slipped -5.8% and needs a creative refresh.

412 ms median answer · every number source-cited · saved to shared memory

It remembers

Your data model, naming conventions, and team context persist in your team's shared memory across every session and every teammate.

Shared memory3 recalled

Your namingset in onboarding
Q3 targetsfrom the plan review
Spend anomalieslearned last week

recalled every session, for every teammate

Saved to your team's shared memory.

See the technical setupThe exact commands, queries & scheduling - for your engineers and AI agents

1Show setup command2See the underlying query3See what’s stored

# 1 - Show setup command
claude mcp add --transport http improvado \
  https://mcp.improvado.io/v1
# 2 - See the underlying query
> which paid channels drove pipeline?
google_ads $1.31M · 8.8x  linkedin $704K · 7.6x
# 3 - See what’s stored
memory: saved to shared memory
naming + context recalled · every session

What teams run in Claude

Show me what Claude does for reporting

Real workflows marketing teams run every day - each on your live, connected data.

$302K7.7x ROASweekly ad spend by channel

$148.2K

8.8x

ABM

$92.6K

7.6x

Prospecting

$61.4K

5.2x

Blended · weekly

$302K

7.7x

Google Ads · Search carries the highest efficiency at 8.8x ROAS - Claude recommends shifting weekly budget there first.

$2.33M+12% wk/wkmulti-touch pipeline · 30 days

Paid search

$1.18M

51%

Paid social

$769K

33%

Organic + direct

$383K

16%

Total modeled · 30d

$2.33M

100%

Paid social is compounding fastest at +12% week over week - Claude flags the budget headroom before quarter close.

94/100+4 ptsdata quality score · 8 weeks

Freshness

96%

+2 pts

Completeness

93%

+6 pts

Consistency

94%

+3 pts

Overall · 8 weeks

94/100

+4 pts

Completeness improved the most this cycle at +6 pts - the remaining gap traces to three legacy connectors.

1,002100% auditedgovernance coverage · all workspaces

Naming conventions

100%

pass

Connector health

98%

2 alerts

Access audit log

100%

complete

Sources governed

1,002

all ws

Two connector alerts remain open - Claude already drafted the fix ticket, keeping connector health at 98%.

Why Claude + Improvado

Marketing data superpowers, inside every conversation

Everything Claude needs to be genuinely useful on marketing questions - clean data, real context, safe access.

You: pull spend across every channel we run

1,000+ sources, one connection

Google Ads, Meta, HubSpot, Salesforce, TikTok and the rest of your stack, in one connection.

1,047 connectors live

Claude: using your naming conventions from May

Persistent memory

Claude remembers your data model, naming and team context between sessions.

avg 214 fields recalled

read-only · workspace scope · audit row #48,206

Secure by design

No raw credentials, scoped role-based access, and a full audit trail on every query.

0 raw credentials stored

same thread, continued on desktop and mobile

Works where you work

Claude on web, desktop, and mobile - the same single connection everywhere.

4 surfaces, 1 connection

spend means spend - on every platform

One clean format

Every channel in one clean format - no messy per-platform differences to manage.

1 format, every channel

as of 41 seconds ago, not last night's sync

Real-time and historical

Live metrics and years of history through the same interface, always current.

3.2 yrs history, <60s lag

You: how did paid social do this week?

Ask in plain English

Ask the way you'd ask a teammate. Claude reads your connected data and answers.

YouWhat's our blended ROAS across Google, Meta, and TikTok this week?

Claude2.94x blended, up from 2.61x last week - Meta carried the lift after Tuesday's creative refresh.

For engineers & AI agentsThe schema, field mappings, and the query behind each answer

1The blended-ROAS answer, as a query2Unified schema (excerpt)3Access

# 1 - The blended-ROAS answer, as a query
improvado.query(
    metric="roas",
    channels=["google_ads","meta_ads","tiktok_ads"],
    window="7d",
)
# 2 - Unified schema (excerpt)
# spend:number
# revenue:number
# roas:number
# channel:enum
# date:date
# updated_at:timestamp
# 3 - Access
# read-only
# workspace-scoped
# every query in the audit log
# zero raw credentials

From question to answer

Minutes to your first answer, not a data project

Connect once and Claude reads your live marketing numbers the same day - no integration backlog, no analyst queue, no keys pasted into a chat.

Weeks become minutes

Ask on day one. The data-engineering project that used to sit between you and an answer is already done.

One connection, every channel

Every ad platform, your CRM, and your warehouse arrive as one clean set of numbers - not 500 exports to reconcile.

Numbers you can trust

Governed access and one clean, consistent data source mean Claude's answer matches what your BI team would hand you.

See the full technical comparisonHow this compares to wiring up each platform yourself

Dimension	Raw platform APIs	With Improvado MCP
Authentication	Separate OAuth per platform	One OAuth connection
Schema	Drifts per platform update	Unified schema, versioned
Memory	None between sessions	Persistent knowledge graph
Credentials	Keys pasted into prompts	Zero credentials in context
Rate limits	Per-API, hit blind	Governed by one layer
Time to first queryThe gap that compounds	Weeks of integration workweeks	Minutes, one MCP configminutes
Dependency	N direct integrations you maintain and control	One governed endpoint - a single vendor dependency, by design

Based on Improvado onboarding data across 500+ connected ad, CRM, and analytics platforms

Architecture

One governed layer between any agent and your entire stack

Switch AI tools whenever you want. Your data, your access rules, and everything your team has taught the system stay exactly where they are - nothing to reconnect, nothing relearned.

Switch agents. Keep your data. Keep your memory

ConnectedImprovado connection1,002 sources · one clean format

connected once - secure, audited, remembered

Shared memoryQ3 CAC targetpipeline reviewspend anomaliespersists across every switch

Claude reaches your whole marketing stack through Improvado - same data, same memory, your credentials never touch the chat.

For engineers & AI agentsThe MCP contract, endpoints, and how memory persists across agents

1One connection, declared once2Scopes - role-based, read-only by default3Shared memory - survives an agent switch

# 1 - One connection, declared once
{
  "mcpServers": {
    "improvado": {
      "url": "https://mcp.improvado.io/v1",
      "auth": "oauth"
    }
  }
}
# 2 - Scopes - role-based, read-only by default
# read:spend
# read:crm
# read:attribution
# write:memory
# 3 - Shared memory - survives an agent switch
# persisted in the governed knowledge graph
# versioned
# every agent reads the same nodes

Enterprise security

Agent access your security team will actually approve

Improvado was built for enterprise marketing data from day one - agents get answers, never credentials.

SOC 2 Type II · HIPAA · GDPR/CCPA · SSO/SAML · every agent query logged read-only, scoped to the human's SSO role, zero raw credentials.

For engineers & AI agentsThe live audit trail and role-based scopes

Live audit trailread-only by default

09:41:03acme-industrialClaude · Paid Media AnalystMeta Ads spendread-only

09:44:52northwind-retailClaude · Finance AgentHubSpot pipelineread-only

09:47:08acme-industrialClaude · RevOps AgentSnowflake revenue martread-only

09:52:31vantage-healthClaude · Compliance AgentSalesforce opportunitiesread-only

4,812 scoped queries logged today across 3 workspaces · 0 raw credentials issued

Role-based scopes

Workspace role	Data access
Analyst	Read: campaigns, spend
Manager	Read: campaigns, spend, budgets
Admin	Read/write: full workspace

Every scope maps to an SSO role. No agent gets more access than the human it acts for.

SOC 2 Type II

HIPAA

GDPR & CCPA

SSO & SAML

FAQ

Common questions

Straight answers on setup, security, and how Claude fits alongside other agents.

Does Claude work with Improvado today?

Yes - Claude supports MCP servers natively. Add the Improvado MCP endpoint to your config, authenticate, and every connected source becomes queryable.

Source: MCP config reference

Which Claude models does this support?

Whatever your Claude plan runs - the MCP layer is model-agnostic and works identically across models.

Is my data safe?

Yes. Agents connect through Improvado's MCP layer with scoped, role-based access and a full audit trail. No raw database credentials are ever shared with an agent, and every query is logged. The platform is SOC 2 Type II audited and GDPR-ready.

What is MCP?

MCP (Model Context Protocol) is the open standard AI agents use to reach external tools and data. Improvado exposes your entire connected marketing stack - 1,000+ sources, unified schema, knowledge-graph memory - through one MCP endpoint.

How long does setup take?

Minutes. If your sources are already connected to Improvado, you add one MCP config to your agent and authenticate with OAuth. No pipelines to build, no schemas to map.

Does it work alongside other agents (Codex, custom agents)?

Yes - the MCP layer is agent-agnostic. Point Claude, OpenAI Codex, or a custom agent at the same connection and every one of them sees the same schema and shares the same knowledge-graph memory.

See it on your own data

Let's talk about your measurement problem

Bring your own data - we'll show your pipeline, attribution, and the agent, live, in twenty minutes.

Book a working session

A solutions engineer maps your stack and shows your own pipeline - no slideware.

Thanks! Routing you to the right person now…

Enter a valid work email to continue.

20 minLive working session

Your dataNo slideware, real pipeline

SOC 2 IIEnterprise-grade security

More agents, same connection

ClaudeYou’re here Codex→ Your agent→

One shared Improvado connection1 connection · 1 memory · 1,000+ sources

Your marketing data,one question away

From question to answer in three steps

Connect

Ask anything

It remembers

Show me what Claude does for reporting

Marketing data superpowers, inside every conversation

1,000+ sources, one connection

Persistent memory

Secure by design

Works where you work

One clean format

Real-time and historical

Ask in plain English

Minutes to your first answer, not a data project

One governed layer between any agent and your entire stack

Switch agents. Keep your data. Keep your memory

Agent access your security team will actually approve

Common questions

Let's talk about your measurement problem

Your marketing data,
one question away