
A lead-gen agency needs four tools, not eight: a data and enrichment layer, a CRM that handles multiple clients, a parallel dialer with per-client number pools, and a light sequencing tool for the touches around the call. The dialer is where the margin is made, because an agency sells meetings and meetings come from conversations. Here is the stack we see working for French and European agencies, and the integrations that make or break it.
An in-house SDR team has one ICP, one CRM and one brand on the caller ID. An agency has 8 clients, 8 ICPs, 8 ways of reporting, and the same 5 SDRs switching between them several times a day.
That changes the stack in three ways. Every tool must be multi-client by design, not by workaround. Reporting per client must come out of the tools, not out of a Friday afternoon spreadsheet. And cost per meeting is the agency's only real metric, so every tool must be judged on meetings per SDR, not on features.
Each client brings a different target list, and the quality of that list caps everything downstream. A list with 40% wrong or missing mobiles wastes 40% of the dialer. For French targets, mobile coverage varies widely by provider; in our experience, waterfall enrichment (several providers in sequence) is the only way to hold match rates above 50%. Build the enrichment step into client onboarding, before the first call, and document the data source for GDPR.
Two models work. Either the agency works inside each client's CRM (cleaner for the client, more switching for the SDR), or the agency runs one CRM partitioned by client and syncs meetings out. What does not work is a shared CRM with no partitions, where a SDR dials a contact that another client already closed.
This is the layer that pays for the others. Our benchmark of 1,000,000 B2B cold calls in France puts the average connect rate at 11.3%. On one line, a SDR gets 5 or 6 conversations a day. A parallel dialer gives 5x more conversations for the same hours, and for an agency that means 5x more raw material for meetings without a single hire.
Agency-specific requirements: a separate number pool per client (so client A's reputation does not burn client B's connect rate), caller ID that matches the client being represented, call recording per client for proof of work, and outcome logging that routes to the right CRM.
Email and LinkedIn touches warm an account before the call and follow up after. Agencies tend to over-invest here because it looks like scale. Keep it light, make sure the sequence pauses when a call connects, and judge it on meetings, not on reply rates.
Prefer native integrations. Connectors built on third-party automation platforms break the week a client renames a field, and the agency is the one explaining the missing data.
Sonik's parallel dialer is used by lead-gen agencies in France with per-client number pools, per-client caller ID and recording, and outcome logging routed to each client's CRM. AI call prep builds the brief per client and per contact before each call, so a SDR switching accounts mid-afternoon gets the context in the dialer rather than in a shared document.


The signals layer can run per client, sorting each list by job posts, website visits and other triggers, and reporting is per client by default. We are French, so number provisioning, caller ID authentication and GDPR-compliant recording are defaults.
Four layers, native integrations, per-client everything, and the dialer at the centre because that is where the meetings are. Judge each tool on meetings per SDR per month and nothing else. This week, count the tools your agency pays for and the number of clients each one can report on separately; any tool that cannot report per client is costing you a Friday afternoon.