Dedicated MCP · included in every engagement

Your paid media, in Claude.

A read-only MCP connector gives Claude your Google Ads, LinkedIn Ads and GA4 data, with my reading rules. Every answer starts from the same corrected figures and the same calculations.

Let’s talk See the comparison

MCP (Model Context Protocol): the standard that lets Claude connect to your tools.

Comparison

Six ways to connect Claude to your data.

My dedicated MCP against the five other possible setups, criterion by criterion.

Yes
Partly
No
In development

Criterion	My dedicated MCP	Third-party aggregatorWindsor.ai, Supermetrics…	Google MCP + LinkedIn aggregator	BigQuery + generic MCP	HubSpot connector	PostHog connectorIts data warehouse sources
Sources covered
Ad platformsGoogle Ads, LinkedIn Ads, Meta Ads, Microsoft Ads	Partly: Google Ads and LinkedIn Ads; Meta Ads and Microsoft Ads in development	Yes	Partly: Meta Ads and Microsoft Ads depending on the aggregator	Partly: Google Ads and Meta Ads only	Partly: Spend and attributed contacts only	Yes
Analytics toolsGA4, PostHog, Mixpanel, Microsoft Clarity	Partly: GA4 for acquisition; PostHog, Mixpanel and Clarity connected directly to your Claude, with my reading rules	Partly: GA4; the others depending on the tool	Partly: GA4 only	Partly: GA4, plus PostHog or Mixpanel via export; not Clarity	No	Partly: PostHog itself and daily GA4 reports; neither Mixpanel nor Clarity
Sales pipelineMQLs, opportunities, customers	Partly: Stages your CRM sends to Google Ads and LinkedIn	Partly: Stages imported into Google Ads; LinkedIn depending on the tool	Partly: Stages imported into Google Ads; LinkedIn depending on the tool	Partly: Stages imported into Google Ads only	Yes	Yes
History of decisions and changes	In development	No	No	No	No	No
Goals and running tests	In development	No	No	No	No	No
Setup
Usable by the whole team on claude.ai	Partly: Once hosted	Yes	No	Partly: If hosted	Yes	Yes
No technical setup on your side	Yes	Yes	No	No	Yes	No
Data
Real time	Partly: GA4 and LinkedIn conversions only	Yes	Yes	No	Yes	Partly: Scheduled syncs; GA4 once a day
All platform fields	Partly: Main fields	Yes	Yes	Yes	No	Partly: Predefined reports; LinkedIn at campaign level
Reading reliability
Reading rules built into every answer	Yes	No	No	No	No	No
Figures already correctedYour customers’ logins excluded	Yes	No	No	Partly: If the views handle it	No	No
Same calculations for every question	Yes	No	No	No	No	No
Security and privacy
Read-only	Yes	Yes	Yes	Yes	Partly: Can write, depending on permissions	Partly: Can write to PostHog
No extra third party on your data	Yes	No	No	Yes	Yes	Yes
Cost
Monthly cost	Included in every engagement valued at €200 per month, excl. tax	$23 to $499 depending on the tool and plan	$0 to $23 depending on the LinkedIn aggregator	Close to $0 excluding LinkedIn pipeline	$0 extra existing HubSpot subscription	$0 extra up to 1M synced rows per month

In development. Decision history, goals and running tests, Meta Ads and Microsoft Ads are planned in the connector, but not available yet.
Analytics tools. GA4 goes through my connector for acquisition: conversions, cost per conversion, corrected figures. PostHog, Mixpanel and Clarity connect directly to your Claude through their own MCP, and I set up that connection as part of the engagement.
Sales pipeline. The connector reads the stages your CRM sends to Google Ads and LinkedIn.
Third-party tools. Public features and pricing as of October 2, 2026, subject to change.

Use cases

What your team can ask Claude with my MCP, and not with the others.

The same question, asked of Claude with a third-party aggregator, then with my MCP. Without context, Claude misreads the numbers and makes recommendations from that misreading. With my MCP, it reads them right, and recommendations stay human: mine, grounded in experience, with AI assistance.

Google Ads · LinkedIn Ads · CRMHistory in development

Answers that take your context into account.

My MCP knows your conversion types, the pipeline stages your CRM sends back and, soon, the history of our discussions and decisions. Claude explains a number instead of just reporting it.

Sample questionWhy did our LinkedIn cost per MQL go up in September?

With a third-party aggregatorWindsor.ai, Supermetrics…

LinkedIn cost per MQL rose from €540 to €690 (+28%). MQLs fell from 12 to 10 while spend went up: your campaigns are losing efficiency.

I recommend cutting the LinkedIn budget and refreshing the creatives, which are probably fatigued.

Misreadings

It doesn’t know that webinar sign-ups no longer count as MQLs since September 1. On the same definition, August was at €720.
It cuts a budget that is producing more opportunities than in August: it doesn’t link them to MQLs.
It fills the missing context with a generic assumption.

My dedicated MCP

LinkedIn cost per MQL rose from €540 in August to €690 in September.

Conversion typesSince September 1, webinar sign-ups no longer count as MQLs. Without them, August was at €720: September is slightly better.

Pipeline4 opportunities from LinkedIn in September, versus 3 in August.

HistoryChange decided at the August 28 check-in, to align MQLs with what your sales team works on.

The increase comes from the MQL definition, not from the campaigns.

Context used

Conversion typesCRM pipelineDecision history

Example: fictional figures.

LinkedIn Ads · ABMIn development

The target accounts your campaigns reach.

For an ABM campaign, Claude lists the companies on your list that saw your ads, those that engaged, and the most engaged. A signal for your sales team, available on the company record in HubSpot.

Sample questionWhich accounts on our ABM list engaged the most with our ads this quarter?

With a third-party aggregatorWindsor.ai, Supermetrics…

Your most engaged accounts this quarter:

Proseware, 41 engagements
Woodgrove Bank, 2,300 impressions
Adatum, 22 engagements

I recommend prioritizing these three accounts with your sales team.

Misreadings

Proseware is you: your employees’ reactions inflate engagement.
It ranks by impressions, which measure delivery, not interest.
Adatum isn’t on your ABM list: it doesn’t know your target accounts.

My dedicated MCP

Of the 300 accounts on your list, 214 saw your ads and 46 engaged. The most engaged:

Northwind Traders, 27 engagements
Contoso, 19 engagements
Litware, 11 engagements

Woodgrove Bank is highly exposed (2,300 impressions) but barely engaged (6 engagements): the message isn’t landing.

Your employees and off-list accounts are excluded.

Context used

Target account listYour employees excludedHubSpot company record

Example: fictional companies and figures.

Let’s talk about your data.

The dedicated MCP is included in every engagement (valued at €200 per month, excl. tax). 45 minutes to review your data sources and what your team wants to ask Claude.