Because the ad platform, the tracking library and the chat widget all write to the same record, this CRM knows things a standalone one cannot: what someone clicked, what they browsed, what they asked, and how likely they are to leave.
Define “at risk” — usually a guess about days since last activity. Export a list from analytics. Export another from the email tool. Match them in a spreadsheet on email address. Find the addresses do not match. Clean it. Upload a static list. Send. Repeat next month, because the list is already stale.
Build a segment where churn risk is high and the contact is opted in. Preview how many people match. Use it. It updates itself as behaviour changes, and the risk score is a model output rather than a guess.
Alongside the usual fields sit three most CRMs do not have: an AI score, a health status, and a recommended action.
Every CRM lets you build an audience on what already happened. Being able to say “churn risk is high” as a condition — in the same builder, with no export and no data-science step — is the part that is unusual.
A dynamic segment describes a rule and updates itself as people qualify or stop qualifying. A static list is a frozen set. You can snapshot a dynamic segment into a static list when you need the audience held still.
Most businesses have a lifecycle in their heads — visitor, signup, active, at risk, lapsed. It exists nowhere in their tools, so every campaign starts by rebuilding it as a filter, and different people rebuild it differently.
The reason marketers do not automate more is fear: a journey is a program that sends real messages to real people, and a mistake is public and irreversible. So you can run a real contact through a real journey and watch exactly what would happen, without anything being sent.
“This simulation does not send messages, mutate contacts, call webhooks, or create jobs.” — stated in the product, not paraphrased here.
Start from blank, a welcome sequence, cart recovery or an event-based template. Choose entry by segment or event, then the enrolment policy — only new members, or everyone already in the segment, with an explicit confirmation before back-filling.
Not a generic error. “This journey needs an entry point.” “This step is not connected to anything.” “This decision is missing a branch.” “Two branches have the same label.”
Select a real contact and see every condition, branch, delay and exit they would encounter. Messages, profile changes and external actions remain suppressed throughout the simulation.
A one-time message to a segment. Pick the segment, choose the channels, assign an active template to each, optionally schedule it. The system refuses to proceed if a selected channel has no active template. Afterwards you see recipients, queued, sent, failed and skipped.
The always-on reflex: when this event happens, wait this long, then send this template on this channel — optionally limited to a segment. Where a journey is a program, an automation is a single reliable response.
Reusable content per channel — email, SMS, WhatsApp, web chat and web popup — with a visual editor for email, a live character counter for short-form, personalisation fields, CTA links, link parameters for attribution, and a preview in a device frame. The list shows which broadcasts and automations depend on each template.
WhatsApp requires business-initiated messages to use templates approved by Meta, and getting one rejected is a common and opaque frustration. Nommos gives you a three-step builder with a live phone preview and a pre-submission checklist that catches what Meta actually rejects.
Social messages are the most commonly dropped channel in marketing, because they arrive in five different apps and belong to nobody.
When the agent decides a conversation needs a person, it marks it, records why, with what severity and confidence, and it appears in Needs Human. Someone takes it, replies, and closes or reopens it.
If a reply cannot go out — the conversation is closed, or the platform's 24-hour reply window has expired — it says so with the reason rather than failing silently.
Analytics report total conversations, how many needed a human, how many the AI resolved, average human first response, average resolution time, and the split between AI and human handling.
Most platforms that added predictive scoring stopped at the score. The distance between “this contact has a 78% churn risk” and “I have sent the right message to the right 240 people” is where such features usually die. Nommos closes that distance in the same screen.
Opens the contact list with the insight's audience already applied as a filter.
Not just from the insight — from one specific cause within it, so “contacts matching this particular driver” becomes a real audience.
It plans an action against that audience and waits for approval before creating anything.
Insights are work items, not a permanent wall of charts.
It compresses the biggest time cost in CRM work — the building — while keeping the decision exactly where it belongs. It also lowers the skill floor: someone who doesn't know how to build a nested segment can ask for one, then inspect what was built.
“Create a win-back campaign for customers who haven't bought in 60 days”
Returns a segment, a template and a broadcast, as drafts.
“Which of my segments are underperforming?”
Analysis rather than creation.
“Build a journey for people who abandon checkout”
Returns a draft journey to inspect on the canvas.
It takes about three minutes, and the match count updates as we add each condition — which is the fastest way to tell whether any of this is real.
