Find a direct number in four steps
Connect Leadbase MCP to Claude Code or Codex, identify the person you want to research, submit a phone question with enrichment_start, then retrieve the result with enrichment_get. For a list already in Leadbase, enrich the selected rows with sheets_enrich instead. Your AI client coordinates the research; it does not place calls.
You need a Leadbase workspace with research access and sufficient credits. A name, company and professional profile help identify the right person. A direct business number may be a mobile number or a work line; a phone result does not guarantee a mobile number for every contact.
For the HTTP version of this workflow, use the complete email and phone API example. It includes submission, retrieval and the difference between a missing number and a failed job.
1. Connect Claude Code or Codex to Leadbase MCP
Follow the MCP setup instructions for your client and authorize the intended workspace. Check the connected tool list and verify access with a read call such as billing_get before starting paid research.
Use the public tools exposed by that connection:
HTTP integrations use dotted operation names such as enrichment.start. Your MCP client uses the underscore tool names above. It does not need to discover an underlying provider operation such as prospects.find_phone.
2. Give the agent a person and a precise phone question
Start with one contact or a small list you intend to research. Include the professional profile URL when known, plus the person's name and current company. If you only have a common name, first identify the person and company rather than accepting a number for a different person.
This fictional example shows the arguments for enrichment_start. Replace the profile placeholders and company details with your own contacts before running it. Choose one stable idempotency key for this exact batch; execution can consume credits.
Each item has its own ID so the agent can join results back to the original person. One phone question per person fits up to 100 people in a batch. If you also need work emails, submit a separate email item per person; a 100-item batch then fits up to 50 people with both fields. See bulk enrichment.
You can give Claude Code or Codex this instruction instead of assembling the JSON yourself:
3. Retrieve the same batch and interpret the answer
The start response contains batchId and jobs. It confirms acceptance, not completion. Pass the returned UUID to enrichment_get:
While a job is queued or running, read the same batch at a reasonable interval with backoff. Save its IDs if you close the client and resume retrieval later. A lost start response is recovered with the original payload and idempotency key; a fresh key could create new paid work.
For example, if the first item completes with a phone value and the second completes with [], you have one contact with a returned number and one without. That is not two successful phone finds. If the second job is still running, its outcome remains unknown. The contact-data documentation shows a fictional response excerpt and the complete HTTP retrieval request.
credits: null means unknown usage, not zero. Read settled costs in the result or Activity. A missing phone can still have research costs even when no contact-result charge applies; check the current credit policy before increasing volume.
4. Enrich a phone column in an existing Sheet
If your people are already in a Sheet, keep research beside their company and profile context. Use sheets_enrich with the configured phone column and the selected row UUIDs:
To create the column, replace column with {"name":"Direct business phone","question":"Find this person's direct business phone number.","type":"phone"}. Existing values and active cells are skipped. Keep rowIds for a selected subset: omitting it covers all existing rows, up to 10,000. Leadbase manages execution admission and dispatch.
Read sheets_get_enrichment_status, then sheets_query for saved values and evidence. Phone and email enrichment use separate columns and requests. See Sheet enrichment for the full workflow.
Mobile numbers, direct lines and DACH coverage
For leads in Germany, Austria and Switzerland, preserve the person's current company and country in your input and evaluate the actual returned results by market. This guide does not establish a coverage rate or promise a mobile number for every DACH contact.
Ask for the number type your task needs. A returned phone string alone does not establish whether it is a mobile number, a direct extension or a switchboard. Only retain type or verification labels when the result actually provides evidence for them. Leave unknown attributes unknown rather than inferring them from a prefix.
Check the identity, current company and evidence before using a result. Retrieving a number does not establish permission to call it. Apply your organization's country, purpose and suppression checks before outreach; those decisions are separate from this research workflow.
Plan and evaluate a larger list
Once your first results are useful, expand with the same record IDs and explicit batch boundaries. A direct workload of 5,000 people with one phone question each requires at least 50 batches. A Sheet lets you manage that list without dispatching a separate client tool call for every cell. Neither batch size nor row count guarantees simultaneous execution or a completion time.
Use the planner to compare the workload with and without first qualifying your target people. Its rates and credit inputs are assumptions, not measured Leadbase performance. Add observed AI, web and identity research costs separately.
Spend phone credits on the right prospects
See how qualifying companies and roles first can focus phone lookups on the prospects your team wants to call.
Use your own results for the most useful estimate. The default values are examples; phone availability can differ between segments.
Under these assumptions, 2,000 people reach phone lookup and approximately 1,200 results are found. The scenario avoids 9,000 credits compared with submitting every candidate.
| Metric | Lookup every candidate | Qualify before lookup |
|---|---|---|
| Expected phone lookups submitted | 5,000 | 2,000 |
| Estimated found results | 3,000 | 1,200 |
| Estimated credits | 15,000 | 6,000 |
Both scenarios use the same assumed found rate and credits per found result. Only the number of people submitted changes.
The planner multiplies your candidate count by the account-and-role fit rate, then applies the same assumed phone-availability rate and contact-action credit price to both scenarios. It retains one decimal throughout and excludes AI, web research, and other usage. The five-credit default models one found phone action. Public enrichment_start accepts 1–100 independent items: up to 100 people for one phone item each, or 50 when email and phone are separate items. A 5,000-person phone list requires at least 50 batches; retrieve results with the returned batch IDs. Sheet enrichment handles the selected rows for you. These estimates are not measured coverage, accuracy, savings, or permission to call.
Illustrative estimate based on your inputs. Actual lookup results and total costs vary; research usage is calculated separately.
Preserve the denominator: every submitted person, including missing results, failures and unfinished jobs. Review returned numbers as confirmed correct, confirmed incorrect, inconclusive or not reviewed. Returned coverage and confirmed accuracy answer different questions; do not count an unreviewed result as correct.
Score a direct-phone lookup pilot
Enter observed pilot counts to separate returned coverage, right-person accuracy, unresolved results, call eligibility, connection outcomes, and lookup-and-review cost.
The accounting value per credit is your input. Included-plan credits, marginal overage credits, reviewer time, and downstream calling cost are different economic categories.
Returned coverage is 62.5%; confirmed right-person accuracy is 85.7%. 15 records passed the separate call-eligibility review, and 7 attempts reached the right person.
| Metric | Result | Definition |
|---|---|---|
| Returned coverage | 62.5% | Found results ÷ submitted lookups |
| Confirmed right-person accuracy | 85.7% | Confirmed correct ÷ (confirmed correct + confirmed incorrect) |
| Inconclusive found share | 16% | Inconclusive found results ÷ found results |
| Call-ready yield | 37.5% | Call-eligible ÷ submitted lookups |
| Right-person connection rate | 58.3% | Right-person connections ÷ call attempts |
| Lookup-and-review pilot cost | €98.75 | Credits × credit value + review minutes ÷ 60 × hourly cost |
| Lookup-and-review cost per call-eligible record | €6.58 | Lookup-and-review pilot cost ÷ call-eligible records |
| Lookup-and-review cost per right-person connection | €14.11 | Lookup-and-review pilot cost ÷ right-person connections |
Every submitted lookup remains visible as found and reviewed, found but not reviewed, not found, pending, or failed. A found result is not treated as correct until the identity review decides it.
Coverage uses submitted lookups. Confirmed accuracy excludes inconclusive and not-reviewed results but keeps both visible separately. Call-ready yield uses every submitted lookup. Connection rate uses actual call attempts. Lookup-and-review pilot cost equals credits multiplied by the entered credit value plus reviewer minutes multiplied by the hourly cost divided by 60; it excludes calling operations and other downstream costs.
The scorecard does not establish statistical significance, data freshness, legal permission, consent, presumed consent, or vendor superiority. Retain row-level evidence and report the segment, dates, definitions, and unresolved cases beside these metrics.
Download the fictional phone-lookup ledger to retain IDs, review decisions and costs. For a deeper evaluation, use the European B2B data-provider test and sales-ready account cost guide.
Start with your first contacts
Connect your client using the MCP setup, run a small phone batch and retrieve the actual results. Use the email and phone API documentation when integrating your own application, or create your workspace to keep the list in Leadbase.
Examples are fictional and explain the workflow; they are not measured coverage, accuracy or cost results. The operation contract was checked against the product source on 20 October 2026.
