Skip to main content
CRM is the single spine through which every motion — outbound, content, paid, events — connects to people. A lead enters once, gets enriched, scored, contacted, and tracked in one place, regardless of which channel sourced it. This page covers what you can do in the CRM, split into what Emma handles in conversation and what lives only in the portal UI.

​
What Emma can do in conversation

Volume caps apply to every run, including scheduled ones. Per run: max 5 LinkedIn invites, 10 LinkedIn messages, 25 emails, 0 deletions in unattended mode. When the cap is reached, remaining items are queued for human review — Emma never exceeds the limit. These are ceilings, not targets.

​
What lives only in the portal UI

These are CRM operations that Emma can’t do in chat — you do them in the portal at CRM → [your list].

​
Lead lists

A lead list is a group of leads with its own pipeline stages, custom fields, and point of contact. When you create one, configure:
  • Name (required) and an optional description.
  • Pipeline stages — customizable per list. Defaults: New → Contacted → Qualified → Proposal Sent → Negotiating → Won / Lost / Churned. Each stage has a label, color, and type (active, won, lost, churned). Add, rename, or reorder to match your motion.
  • Point of contact — the team member or email sender assigned to the list. AgentWeb uses this when sending outreach.
Landing pages Emma generates create a lead list automatically and enable public form submission, so ad and content traffic flows in without manual setup. See measurement for how to carry source attribution through.

​
Lead fields

Each lead has these fields out of the box:

​
Custom fields

Add unlimited custom fields with the Add Field button. Custom fields are searchable, filterable, and valid targets for enrichment output. Use snake_case names for fields you’ll target with enrichment: company_size, industry, funding_stage. These become the fields Emma writes to when she enriches.

​
Three custom fields every list should have

If you’re using measurement, add these three to every list so cross-list reporting works:
  • lead_source — the category: founder-outbound, organic-content, paid-search, paid-social, event, referral, direct.
  • lead_source_detail — the specific thing: campaign name, post URL, event name, referrer.
  • first_touch_date — when the lead first entered, not when you last talked.
Landing pages inject lead_source and lead_source_detail from UTM parameters automatically.

​
Enrichment

Emma researches leads and fills missing fields from public data. Common uses: fill company size, industry, or funding stage on imported leads; find LinkedIn URLs; add context Emma uses later to personalize outreach.

​
How it works

1

Ask Emma

In Agent Mode: “Enrich my leads in [List Name] with company size and industry.”
2

Preview on 2 sample leads

Emma shows what the enrichment would produce on two leads so you can check quality before the batch runs.
3

Confirm

You approve before anything writes. In an unattended run, the batch is capped and the remainder is queued rather than draining credits.
4

Execute

Emma processes leads in parallel and updates fields in real time. Large batches (500+ leads) can take several minutes.

​
Data sources

Combinable per enrichment job:

​
Provenance labels

Every enriched value carries a label so you know what you’re building on: Funding figures especially tend to carry web-unverified or ai-knowledge. Don’t build a segment on top of an inference without spot-checking.

​
Best practices

  • Always preview first. The 2-lead sample shows quality before credits are spent.
  • Use specific output field names. Vague names like info make filtering harder later.
  • Combine enrichment with filters — enrich only leads matching a condition like Company is not empty.
  • By default, Emma skips leads that already have the target field populated. To re-enrich, tell her to force overwrite.
  • Enrichment depends on public data availability. Leads with only a name and no company often return little.

​
ICP scoring

Separate from enrichment. Scoring reads your Market Strategy ICPs and evaluates each lead against them — it doesn’t discover new data, it evaluates what’s there. Ask Emma: “Score each lead in my [List Name] list against my ICP.” Each lead gets a 0–100 score and a one-line reason. Emma tiers them:
  • 🏆 Tier 1 — core firmographic fit + priority role + priority vertical.
  • 🥈 Tier 2 — right shape but below threshold, or a secondary role.
  • ❌ Exclude — out of ICP (B2C, students, agencies, free-mail with no company) unless your strategy says otherwise.
The score is only as good as the data on the record. If leads are missing fields that affect scoring (title, company, industry), enrich first, then score.
ICP scoring loads your ICP definition at runtime from Market Strategy — it never uses a hardcoded profile. If your Market Strategy is empty, Emma asks rather than guessing. A stale or broad ICP degrades every score at once.

​
Outreach

​
Drafts

Ask Emma to draft outreach for a lead or a batch: “Draft a personalized email to each lead in my Q3 Outreach list.” Emma reads Brand Guidelines, the lead’s record and engagement history, and the Knowledge Base, then generates an email draft per lead. A social message draft — for LinkedIn or whatever platform fits — is generated alongside it automatically. Show both to the person before sending.

​
Sending

Email sends go through your connected Gmail or Outlook address. The message looks like it came from you. Emma confirms before sending and tracks the send as an engagement event on the lead. For LinkedIn, the social message draft needs to be sent manually by you on the platform itself. After you confirm it’s sent, tell Emma and she logs it as an engagement.

​
Multi-touch sequences

The Cross-Channel Sequence Orchestrator skill runs multi-step outreach across LinkedIn, email, and personalized video:
  • Picks the right channel and step per lead based on where they are in the sequence.
  • Sends in your brand voice.
  • Stops on reply — a reply breaks the sequence and surfaces the lead for your attention.
  • Schedules the next touch.
It runs unattended with the same hard caps (5 LinkedIn invites, 10 LinkedIn messages, 25 emails per run) and produces a completion ledger at the end.

​
Compliance

  • Marketing emails include unsubscribe handling and a physical mailing address (CAN-SPAM).
  • Opt-outs recorded in lead status are honored automatically.
  • LinkedIn volume is capped below platform limits to avoid account flags. The caps are conservative by design — LinkedIn’s enforcement is unpredictable and the cost of an account restriction is weeks, not hours.
  • Emma never enriches EU leads with fields your data-processing terms don’t cover. When unsure, she flags rather than proceeds.

​
Engagement tracking

Every interaction with a lead is logged as an engagement event — email sent, email opened, LinkedIn message, call, meeting, custom event. Emma checks engagement history before drafting new outreach so she doesn’t repeat herself or cold-pitch someone she’s already spoken to. Log interactions manually: “Log that I had a call with [lead name] today. We discussed pricing and they asked for a proposal.” Emma adds it as a custom engagement event with the details you provide.

​
Deal promotion

When a lead qualifies, promote them to a deal: “Promote [lead name] to a deal at $X.” The lead record stays linked so engagement history carries forward.

​
Search and filter

Chat-based search returns up to 50 leads per call. There’s no “next page” — refine the query if you need a narrower set. For operations across an entire large list (mass updates, full exports, bulk ICP scoring), use the portal UI.

​
How CRM connects to the rest of the system

CRM is not a standalone feature. It’s the spine that makes the pipeline system work by holding the leads, scores, outreach drafts, and engagement history that every other motion reads from and writes to.

​
Next steps

Last modified on September 1, 2026