One question to AI, the right companies in your CRM: using Innovadata MCP
Most prospecting work starts with a long list and ends with a shorter one. Innovadata MCP lets you begin with the shorter list. Describe the company you want in an AI chat, check the count, then decide whether that segment is worth pulling into your sales process.
For example: "Find Estonian software companies with at least three employees, email available, and quarterly turnover growth above 20%. Count them first."
The AI translates that request into the right filters. You do not have to memorize API parameters or hunt through a dozen screens.
What MCP does here
MCP is the connection that lets an AI client use Innovadata as a tool. The server can list the filters available in a country, find the right industry code, count a segment, search companies, and open a detailed company record.
The sensible sequence is simple:
- Ask which filters the selected country supports.
- Count the segment before pulling rows.
- Pull a small sample and review it.
- Send approved companies to the CRM through a separate CRM connection.
Innovadata MCP is a read and search connection. It does not write directly into a CRM. The writing step uses a CRM API, a CRM MCP connector, or an automation tool such as Make, n8n, a webhook, or a spreadsheet export. That separation is useful. It gives the sales team a chance to stop bad matches and duplicates before they enter the pipeline.
The MCP endpoint is https://app.innovadata.eu/mcp. Set it up in an MCP-capable AI client with Streamable HTTP and an Authorization Bearer key. Keep that key in the client’s secure settings, never in a web page, a prompt library, or a shared document.
Three live segment checks
We ran the following free count_companies checks on 23 July 2026. These are live registry counts, so they will move over time.
| Use case | Checked segment | Result |
|---|---|---|
| Growth-led outreach | Estonian software firms, 3+ employees, email, quarterly growth 20%+ | 129 |
| Account-based marketing | Estonian companies, 20+ employees, annual revenue €5m+, email, no public tax debt | 1,385 |
| Digital gap campaign | Estonian companies, 5+ employees, quarterly turnover €50k+, email, no website, no public tax debt | 7,881 |
The last result is a good reminder that a count is only the start. A missing website in registry data is a useful lead for a manual check, not proof that the company has no website at all.
Three ways to use the results
1. Use growth as a reason to start a conversation
Industry alone is a weak reason to contact someone. A growth signal gives the message a little more context. A company adding turnover or headcount may be reviewing software, recruiting, finance operations, office capacity, or marketing support.
Try this:
Find Estonian software companies with at least three employees, quarterly turnover growth above 20%, and an email address. Show the count first. Then return five examples with the turnover and employee fields. Do not create CRM records yet.
The first count was 129. That is small enough for a focused campaign, and large enough to split by location, company size, or service fit.
2. Build an ABM list from facts you can explain
For a high-value sale, a compact account list is more useful than a huge export. The second check found 1,385 Estonian companies with 20 or more employees, annual revenue above €5m, an available email address, and no public tax debt.
That does not tell you who will buy. It does give a sales team a defensible starting point. Add the company registration code, industry, revenue band, headcount, owner, and source date to the CRM. Use the registration code as the primary de-duplication field.
3. Start with a visible digital gap, then verify it
Web agencies and digital consultants often want a practical reason to approach a business. We counted 7,881 Estonian companies that met the turnover, headcount, email, and tax-debt conditions above while their registry record had no website.
Treat this as a review queue. Ask the AI to group the list by industry, pull a small sample, and then check the actual web presence before anyone sends a message. The campaign becomes more credible when the offer refers to a real situation rather than a generic pitch.
Prompts worth keeping
Which filters are available for Poland? Build a segment of VAT-registered companies with a website and phone number. Count it before returning any rows.
Find companies similar to these three customers by industry, employee count, and location. Exclude the registration codes already present in our CRM.
Look up this company by registration code. Summarise its status, activity, financial fields, contact channels, and board information in five short points.
Return only five records. Show the fields that make the segment relevant and explain which filters you applied.
The last prompt matters. It keeps the first review manageable and makes the AI show its work.
Move approved results into the CRM
Map the registration code to an external unique ID in the CRM. Map the company name, industry, address, domain, contact channels, turnover, employee count, and source date to the fields your team actually uses. Then ask the AI to show a preview before writing anything.
HubSpot, Pipedrive, Salesforce, Microsoft Dataverse, and many other systems have APIs for creating or updating company records. The right implementation depends on your CRM and who should own the records. The important part is less glamorous: avoid duplicates, retain the source date, keep a clear campaign label, and require approval for bulk actions.
Use the data with care
Company data and public business contacts can support B2B work, but a list is not a blanket permission to contact everyone. Assess the lawful basis for each campaign, use only data you need, respect local direct-marketing rules, and make opting out straightforward. If a campaign uses personal data, document the purpose and the balancing assessment.
Start with one narrow segment and five sample rows. If the segment makes sense after a human review, you have a repeatable way to turn a plain-language question into a useful sales list.