The complete customer handbook

NeuraLeads is an outbound revenue platform. It finds companies that are actively buying, identifies the decision-makers inside them, verifies their email addresses, runs personalised sequences from your own mailboxes, and tracks every reply through to a closed deal — with guardrails that stop your sending domain from being burned along the way.

Who it is built for

What makes it different

• It sources its own leads. You are not uploading a purchased list. The platform pulls live buying signals — job postings, hiring activity, funding news, technology changes, provider registries — and turns them into leads.
• Every send passes a safety gate. Ten ordered checks run before any email leaves. If one fails, the email is not sent and the reason is recorded.
• Mailboxes are warmed, monitored and rested. A built-in warmup engine builds sending reputation, checks your DNS records and watches blacklists.
• AI is bounded, not autonomous. AI writes, rewrites, classifies and suggests — but a deterministic policy engine, not the model, decides whether an email may send.
• Flat pricing, no per-seat fee. Add your whole team on any plan.

Nine terms carry most of the meaning in the product. Learn these and every screen becomes readable.

The left-hand navigation inside the app is ordered as a working sequence, not an alphabet. Follow it top to bottom in your first week and the platform sets itself up in the right order.

Sending reputation is the one asset you cannot buy back. Everything in this stage exists to protect it.

Mailboxes

Connect any SMTP-capable email account — Google Workspace, Microsoft 365, or a custom domain host. For each mailbox you control:

• Daily send limit — how many emails this mailbox may send in a day (30 by default).
• Health score — a live rating combining bounce rate, complaint rate, warmup age and recent deliverability.
• Warmup status — not started, warming, or ready.
• Signature — appended automatically and available as a merge field in every template.
• Outreach role — which decides how the mailbox behaves, below.

Automated mailboxes vs. personal mailboxes

When you create a personal mailbox, the platform can create or link the matching login user for you and map the outreach role to the right permission level automatically.

How a mailbox is chosen for a send

When a campaign has no mailboxes assigned, the platform picks the healthiest available automated mailbox on a weighted score: health 40% · remaining daily quota 30% · warmup age 15% · deliverability 15%. When a campaign does have mailboxes assigned, it rotates round-robin through exactly those.

Warmup Engine

The warmup engine builds and defends reputation on every connected mailbox. It has seven tabs:

Behind the tabs, the engine sends and auto-replies to peer warmup emails on a schedule, emulates human reading behaviour, resets daily counters, and runs a recovery check each morning for any mailbox whose reputation slipped.

The sourcing pipeline queries your enabled sources, normalises everything into one shape, removes duplicates three different ways, and applies your exclusion rules before a single lead reaches your database.

What a sourcing run actually does

1. Query — every enabled source is queried in parallel with your job titles, locations and filters. Where a source supports it, your exclusion keywords are pushed into the query itself, so unwanted postings are filtered before they are ever downloaded.
2. Normalise — results from different sources are mapped into one common lead shape.
3. De-duplicate — three layers in order: exact posting identifier, then employer profile, then company + title + state + city.
4. Company gate — confidential or blank employers are dropped, companies above your size ceiling are dropped, and excluded industries are dropped.
5. Enrich the gaps — where a source did not report a company’s industry or size, an AI step fills it in, with an optional paid firmographic lookup for size specifically. An unknown value is never used as a reason to drop a company.
6. Record — leads are saved with their source and sub-source, and the run’s counters show exactly how many were dropped and why.

Run counters — how to read a sourcing result

Sourcing controls you own

• Job titles and locations you target, and the keywords that exclude a posting outright.
• Company size ceiling — default 200 employees; set it to 0 to disable the limit entirely.
• Excluded industries — IT, staffing agencies and government by default.
• Minimum salary — default $40,000.
• Excluded companies — a permanent do-not-source list you manage on its own screen.
• Per-source tuning — batch size, page depth and result caps for each individual source, plus a global cap per source and how many sources are queried in parallel.

Enrichment turns a company into named people you can write to, ranked so the best contact is emailed first.

Priority tiers

Rules applied during enrichment

• A hard cap on how many contacts are pulled per company per opportunity — 2 by default — so you never carpet-bomb one account.
• Contacts already in your workspace are matched and re-used rather than duplicated.
• Cost is recorded per provider on every run and appears in your cost analytics.
• Company records are created or updated alongside the contact, including the timezone derived from the company’s location — which is what makes timezone-aware sending possible later.

Validation is not optional and it is not advisory. Only contacts whose address is confirmed Valid can receive outreach — the send gate enforces it regardless of what any campaign says.

Validation runs in bulk from the Validation screen or as part of a pipeline run. Results are written back to the contact, so a contact only needs validating once unless you re-check it. Per-contact validation cost is tracked in your cost analytics.

A campaign is a sequence of steps, a schedule, a set of mailboxes and a list of enrolled contacts. Everything else on this screen exists to make that sequence land in an inbox and sound like a person wrote it.

Building a sequence

Steps run in order and can be of six types:

Personalisation tools inside a step

• Merge fields — recipient first name, sender first name, job title from the lead, job location, company name, mailbox signature and more.
• Spintax — write {Hi|Hello|Hey} and each recipient gets a different variant. Nested patterns are supported, so no two emails read identically.
• A/B variants — weighted variants per step, with automatic statistical selection of the winner once there is enough data.
• Templates — pull any active template from your library straight into a step, grouped into Outreach and Follow-up.
• AI assistance — generate a whole sequence from a brief, suggest five subject-line variants, or improve the campaign’s own name and description.

Campaign statuses

Scheduling and send windows

• Send window — the hours of the day the campaign may send, with a timezone.
• Send days — which weekdays are allowed.
• Multiple dated schedules — different windows for different date ranges on the same campaign, in a defined order.
• Per-contact timezone — where a contact’s timezone is known it overrides the campaign timezone, so an East-coast prospect is emailed on East-coast time.
• Optimal-hour scheduling — follow-up steps are placed into local business hours, with 9–11am treated as the strongest window.
• Slow ramp — increase daily volume by a set amount per day instead of starting at full throttle.

Enrolment

• From leads — select leads and the platform names the campaign, builds a three-step sequence, assigns your active mailboxes and enrols the contacts in one action.
• Available leads view — shows only leads not already in an active campaign, with their contact counts, so you never double-enrol.
• Auto-enrolment — a scheduled job adds newly qualified contacts into eligible campaigns.
• Assignment modes — manual, round-robin or weighted distribution of enrolled contacts across your reps.

Automatic pausing

You set bounce-rate and spam-complaint thresholds per campaign. An hourly health check pauses any campaign that breaches them and tells you why. Complaint-rate monitoring pauses sending at a 0.3% complaint rate independently of that.

Preview mode and the Email Preview screen

Switch a campaign into preview mode and it generates drafts instead of sending. On the Email Preview screen every draft can be:

• Read exactly as the recipient will see it, with all merge fields resolved.
• Scored for spam risk — trigger words, suspicious patterns, link-to-image ratio — and for overall deliverability.
• Auto-fixed for the specific spam issues found, or rewritten by AI.
• Approved, rejected, or approved as a batch and released to send together.

Drafts carry their source — campaign, pipeline or broadcast — so you can filter a review queue down to one origin.

Every outbound email, from every path in the product, passes the same ten checks in the same order. The first failure stops the send and is recorded with its reason. This is the single most important safety mechanism in the platform.

1. Contact statusUnsubscribed, bounced or inactive contacts are never mailed.
2. Suppression listBlocks any address, or a whole domain, that you have suppressed.
3. Email validationThe address must be confirmed Valid, or an allowed catch-all.
4. CooldownNo second email to the same person inside the cooldown period — 10 days by default.
5. Per-lead limitCaps how many emails a single opportunity can generate.
6. Company capCaps concurrent conversations inside one company, so one account never receives a flood.
7. Sequence fatigueStops a contact who has already received many emails from receiving more.
8. Domain throttleLimits how many emails go to one recipient domain per day — 30 to major consumer providers, 50 generally.
9. Cross-campaign de-duplicationPrevents two campaigns from independently emailing the same person.
10. Policy decisionA final go / no-go combining policy rules, content quality and lead score, with a written reason code.

What happens after a send

• Opens are tracked by a tracking pixel and clicks by a redirect on each link. You can point both at your own custom tracking domain, which the platform verifies for you.
• Hard bounces automatically suppress the address and mark the contact inactive — no manual clean-up.
• Complaints feed the complaint-rate monitor and can auto-pause the campaign.
• Replies stop the sequence for that contact immediately and appear in the unified inbox.
• Engagement scoring weights signals by strength — a reply counts far more than a click, a click far more than an open.

Content quality safeguards

• Spam checker — over a hundred trigger words plus pattern and link-ratio checks, run before a send and on demand.
• Similarity guard — measures how close an email is to ones already sent, so near-identical mass mail is caught.
• Humaniser — varies sentence length and rhythm so copy does not read as machine-generated.
• Rendering check — warns about HTML that breaks in common email clients.
• Optional DKIM signing for custom SMTP connections.

Replies from every mailbox land in one place, already threaded, categorised and scored.

Unified inbox

• Threading — messages are grouped into real conversations across mailboxes and campaigns.
• Sentiment — each inbound message is scored positive, negative or neutral.
• Categories — Interested, Not interested, Out of office, Question, Referral, Do not contact, Other.
• AI reply suggestions — drafted from the whole conversation, not just the last message.
• Reply macros — your own saved quick replies with variable substitution and usage tracking.
• Objection library — tested responses to common objections, seeded for you and extendable, each with an effectiveness score.

AI Reply Agent — two modes

Replies categorised as interested can be forwarded into your CRM automatically.

Deals

A Kanban board across seven stages — New Lead → Contacted → Qualified → Proposal → Negotiation → Won or Lost. Stages are configurable per workspace.

The claim queue

New deals arrive unclaimed and are broadcast to every business developer and recruiter, in-app and by email. Any of them can claim a deal to take ownership; the claimer or an admin can release it back. Admins can also assign a deal directly to a named person, who is notified immediately. Both notification types respect each user’s own notification preferences.

Deal detail — the 360 view

• The originating job or signal that created the opportunity.
• The full mail chain — outbound sends and inbound replies merged into one chronological history.
• Candidate submissions with their own lifecycle: Submitted → Reviewed → Sent to client → Placed or Rejected.
• Tasks with assignee, due date, priority and status, plus a personal “my tasks” view.
• A complete activity timeline.

Filtering the board

Filter by stage, by value or probability (equals, not equals, less than, greater than, between), by created-date range, by claim state (anyone, unclaimed, me, or a named person), by free-text search, or narrow to just your own deals. Each card shows who claimed it, its owner and its age in days.

Forecasting

An AI forecast engine projects pipeline outcomes from stage, value, probability and engagement history.

AI is used in fourteen places in the product. In none of them does a model have the final say over whether an email is sent.

Where AI helps

The guardrails around it

• A deterministic policy engine decides sends, not the model. Its rules are fixed, readable and overridable per workspace.
• Every AI decision is logged with its inputs, its output and its reason code, so any send or auto-reply can be explained after the fact.
• Inbound content is sanitised before it reaches a model, so text inside a prospect’s reply cannot issue instructions to your AI.
• Low confidence escalates to a human, and so does anything classified as a destructive action.
• Structured outputs are validated against a schema; malformed model output is rejected rather than acted on.
• Failures retry and fall back across providers, so an outage at one AI vendor does not stop your campaigns.
• Token cost is metered per call and appears in your cost analytics alongside every other provider.

Every screen in the product, what it is for, and who reaches it by default.

Reports in detail

Every report exports to file, capped at 10,000 rows per export.

Settings tabs

Every provider tab has a test connection button, so you can confirm a credential works before a pipeline depends on it. Stored credentials are encrypted and never displayed back to you — a saved field shows only that a value is set.

You bring your own provider accounts and connect them in Settings. Every category has several options, so you are never locked to one vendor — and every category has a mock mode for training and testing without spending credits.

Your own API access

• API keys — create scoped keys with expiry dates for your own integrations, and revoke them at any time.
• Webhooks — subscribe to events and receive signed payloads you can verify. Failed deliveries retry three times with increasing delays.

Webhook events available today: email sent, opened, clicked, replied and bounced; contact unsubscribed; campaign completed; lead created.

Seventeen scheduled jobs keep the platform working while nobody is logged in. Every one can be switched off individually from the Automation screen — useful during a migration, a quiet period, or a provider outage.

Additional monitors run alongside these: hourly campaign health checks that auto-pause on a threshold breach, and continuous complaint-rate monitoring. A locking mechanism guarantees a job never runs twice at the same moment, even across multiple servers.

Four built-in roles cover most teams. Where they do not, you can build your own on top of them.

The permission matrix

Permissions are set per module at four levels: Full access · Read & write · Read only · No access. Three modules go further and are controlled tab by tab rather than as a whole:

• Warmup Engine — each of its seven tabs independently.
• Pipelines — each of the four stages independently, so a user can run sourcing but not outreach.
• Settings — each of the ten configuration tabs independently, so a user can tune sources without seeing everything else.

Custom roles

Create a role, give it a display name, base it on one of the built-in roles, then adjust its matrix. Built-in roles cannot be deleted, and neither can a role that is currently in use by a user.

Account security controls

• Email verification is required before an account becomes active.
• Accounts lock automatically after repeated failed logins; a super admin can unlock them.
• Every login attempt is recorded with its outcome and reason — invalid credentials, inactive, unverified or locked.
• Sessions expire and refresh silently in the background, so you are not logged out mid-task.
• Each user belongs to exactly one workspace and cannot see another’s data.
• Unverified accounts older than 72 hours are cleaned up automatically.

Notification preferences

Every user has two master switches — in-app notifications and email notifications — editable by an admin or by the user on their own profile. They gate every notification the platform would otherwise send that person: deal broadcasts, assignments and system alerts.

Flat monthly pricing with no per-seat charge, a 14-day trial on every plan, and 20% off when billed annually.

Plan limits are enforced at the moment you create something, not silently in the background — you are told immediately if an action would exceed your plan, so nothing is ever created and then quietly discarded. Alongside those caps run two monthly meters: credits, which pay for data enrichment, validation and AI (roughly six credits move one contact from sourced to ready-to-send), and a separate send allowance. They are independent — running out of one does not affect the other — and extra credits can be bought on paid plans at $20 per 1,000, valid for 12 months.

Invoicing

• Invoices are generated monthly and numbered sequentially, with duplicate protection.
• Each invoice carries line items, tax at your configured rate, and a downloadable PDF.
• Status moves Draft → Sent → Paid, or → Overdue if the due date passes.
• Automated email at each point: a new invoice with the PDF attached, an overdue reminder, and a payment acknowledgement.
• Pay by card through hosted checkout, or have payments recorded manually — bank transfer, cheque or card.
• Subscriptions can be started, reviewed and cancelled at period end from the Billing screen.

Cost tracking & ROI

Separate from what you pay for NeuraLeads, the platform meters what your own provider accounts cost you, in five categories: lead sourcing, contact discovery, validation, sending and AI. Costs are captured per call — accurate to fractions of a cent for AI tokens — and surface as cost per source, cost per lead and return on investment in Analytics. Credit usage is metered per workspace with a running balance.

You remain the controller of the contact data in your workspace. These are the controls the platform gives you to honour that responsibility.

Isolation

Every record belongs to exactly one workspace, and every query is scoped to the workspace of the person making it. There is no shared pool of contacts, leads or campaigns between customers.

Data-subject rights

Suppression & unsubscribe

• Every outbound email carries an unsubscribe path, and an unsubscribe is honoured immediately and permanently.
• You can suppress a single address or an entire domain by hand at any time.
• Hard bounces suppress automatically.
• A reply categorised “do not contact” removes that person from further outreach.
• Suppression is checked at position two of the send gate — before validation, before any campaign logic.

Backups

A backup of your workspace data is taken automatically every day, and old backups are cleaned up on a schedule. From the Data Backups screen you can create a backup on demand, download one, restore from one, or delete one. Restoring is a deliberate, confirmed action — you type the word RESTORE to proceed — because it replaces current data.

Audit trail

Login attempts, permission changes, AI decisions, scheduled job runs, campaign sends and data-erasure actions are all recorded. Administrators can review authentication history, currently active users and system activity at any time.

Platform security

• Passwords are stored using a modern memory-hard hashing algorithm — they are never recoverable, only reset.
• Provider credentials you enter are encrypted at rest and never shown back to you.
• All traffic is served over HTTPS with strict transport security and a content security policy.
• Login, pipeline execution and billing actions are rate-limited to resist abuse.
• Webhook payloads are cryptographically signed, so your receiving system can verify they genuinely came from us.
• API keys are stored only as hashes, carry scopes, and can be given an expiry date.

These are the shipping defaults. All of them are configurable per workspace.

A realistic path from an empty workspace to a first reply.