Common Room’s pitch is simple: every buying signal in one place, resolved to real people and companies. The pitch holds up — but an enterprise implementation succeeds or fails on sequencing, prerequisites, and the engineering around the edges. This is how we implemented it for an enterprise hybrid-cloud storage company, as part of the broader engagement described in the overview.
Connect signals before you build anything else
The first pillar of the rollout — before segments, before scoring, before rep workflows — was connecting every data source the platform would resolve contacts across. Segments built on partial data just get rebuilt later. Our connection list, roughly in order:
- Salesforce (bidirectional). The backbone sync. Scheduled deliberately after the basics because the company’s sandbox was contended by other projects — see the intent-data architecture page for why sandbox-first matters.
- Salesloft. Connected in minutes via native integration, then verified end-to-end by pushing a single test contact into a cadence before trusting it with batches.
- Website visits. Three-part setup: a tracking snippet in the site HTML or Google Tag Manager (Tag Manager gives you trigger control), an enhanced US visitor-identification account through the bundled Vector partnership, and manual identity calls — an extra snippet on form-fill pages like demo requests and logins that dramatically improves de-anonymization from company-level to person-level.
- LinkedIn company page. The simplest connection technically — paste the page URL while logged in as an admin — but it requires a content admin or super admin of the company page, which in an enterprise means finding that person and getting on their calendar.
- Bombora intent topics. Bundled with the Common Room contract at five topics. We used three to four for solution-relevant topics and reserved the rest for competitor names, layered with technographics for a competitor-displacement motion.
- Job changes. Out of the box; nothing to configure. A champion moving to a new account is one of the highest-converting signals in the system.
- CSV imports. Event lists, webinar registrations, conference badge scans, content-syndication leads. Trivial to upload, but see below on making it repeatable.
Sources without a native connector — G2 and TechTarget in our case — show up as “get set up” in the integrations page. You have two routes: push them in through the API (best when you want net-new contacts and activity records) or land them in Salesforce first and map the fields through the CRM sync. We chose the API route for fresher data and fewer CRM schema changes.
Identity resolution and enrichment: the actual product
What justifies the platform is the contact graph. An anonymous website visit, a LinkedIn comment, an intent spike, and a CRM contact record resolve into one person at one company, with title, seniority, and firmographics enriched perpetually — no per-record credits, which changes how liberally you can enrich compared to credit-metered tools like Clay. In practice we still see Clay-style tooling as a complement for the last 10–15% of custom enrichment; Common Room covers the standing motion.
Two things move match rates more than anything else: the manual identity calls on form pages, and connecting more sources — every additional identity edge (an email here, a LinkedIn profile there) helps the resolver. Expect a meaningful minority of US web traffic to resolve to actual people once Vector is live; the rest resolves to company level, which still feeds account-level segments.
Segments and alerts: living audiences, not lists
Dynamic segments are the working surface. A segment like “target-account contacts with a pricing-page visit or active intent topic in the last 14 days” adds and removes people automatically as behavior changes. From a segment, workflows run on a schedule — the platform re-evaluates several times daily — and can push contacts into a Salesloft cadence, write a tag back to Salesforce (which is how the AE team, working out of Outlook and the CRM, sees the same signals the SDRs act on), or fire an AI-summarized alert telling the owning rep who moved, why, and what to do about it.
Design segments around the play, not the data source. “Visited the site” is a source; “second-visit from a target account plus an active competitor intent topic” is a play. Scoring came after two weeks of watching real signal volume — calibrating thresholds against live data beats guessing weights up front.
What Common Room does well — and where you need engineering
What it does well: signal aggregation across native sources, identity resolution, perpetual enrichment, dynamic segments, and the action layer into sales-engagement tools. The MCP connector is a genuine differentiator — reps and ops query the contact graph in natural language from Claude instead of learning another dashboard.
Where you need engineering around it:
- Non-native sources. G2, TechTarget, and niche intent vendors need API pipelines built and maintained. Budget real engineering time.
- CSV normalization. Every list source formats headers differently. We standardized field mappings and built Claude skills to merge, validate, and reformat files to the import template — covered in the marketing AI page. Without this, imports stay manual and error-prone.
- MCP write-back at volume. Reading via MCP worked immediately; writing 150-contact batches hit rate limits and permission edges that took troubleshooting with the vendor. Verify write-back permissions per connector instance and keep bulk loads on the native importer until the path is proven.
- Prerequisites and approvals. LinkedIn admin rights, tag-manager access, SOC 2 review of the visitor-identification vendor, security sign-off on each connection. None are hard; all stall without an owner. This was the single biggest source of calendar time.
- Vendor risk watch. Common Room was acquired by Zoom mid-engagement. No functional impact for us, but consolidating your signal layer onto one platform means tracking its corporate trajectory like any other critical dependency.
Sequencing summary
Week one: licenses provisioned, CRM and sales-engagement connected, website tracking handed to the web owner with documentation. Weeks two to three: LinkedIn, intent topics, first CSV loads, admin settings, draft segments. Week four: Salesforce sync completed, scoring calibrated, rep workflows live. Signal connection is front-loaded coordination work; everything downstream — the cadence automation and the alerting — moves fast once the graph is populated.
Work with Node8
Node8 implements Common Room as part of a full signal-driven GTM build — connections, identity strategy, segments, and the engineering around the edges the platform doesn’t cover. Get in touch if you want the signal layer running instead of planned.