How StellarReach works.

It runs cold outreach for you, with proof attached. Proto (the prospect finder) finds small and medium-sized businesses, an AI scores how weak each one's website is, a personalized demo site is built for the best fits, and an email sequence follows up — all sending from your own inbox and domain. You step in only when a lead engages.

Three things, all in Settings: a sending inbox (Gmail, Outlook / Microsoft 365, or SendGrid), a way to pay for AI usage (a prepaid balance on platform keys, or your own keys on Scale & Agency), and your targeting + scoring rules. Your setup page walks you through these one step at a time, and disappears once you're set up.

Largely, yes — that's what Funnel Blueprint is for. Answer about twenty short questions about your business and it writes your scoring rules, build prompts, and email copy from them, for you to review before anything is saved. Worth doing before your first scoring run: see Funnel Blueprint.

1. Find — Proto scrapes candidate businesses from your target market. 2. Score — a vision AI rates each website against your rubric. 3. Build — high-scoring leads get an auto-generated demo site. 4. Reach out — the email sequence sends from your inbox. 5. Engage — when a lead opens, clicks, or replies, they move to Engaged Leads for you to close. 6. Sync — engaged leads can push into your CRM.

A lead's status tracks the outreach mechanics: In progress (queued/building), Sent, Email opened, Clicked (opened the demo), Engaged (replied or submitted a brief), Follow up, and Won. Status advances automatically from tracking signals; you can also set it by hand.

Leads are grouped by temperature — how warm the interest is: Frozen (manually set aside), Cold (email opened, demo not clicked), Lukewarm (demo clicked), Warm (project brief opened), and Hot (brief submitted). Sort and filter by it to spend your time on the warmest leads first. Note: a bare email-open is the weakest signal — Apple Mail and spam scanners fire it with no human — so the Engaged Leads view hides Cold leads (and demo clicks that were never actually scrolled/read), showing only genuine interest. Use Lead Pipeline for everyone.

Today is your prioritized worklist — the leads that need action right now, in the order to do them. It pulls four buckets together: replies waiting, follow-ups due, hot leads to contact, and leads going cold. Work top to bottom; when it's empty, you're genuinely caught up. It's the one page that answers "what should I do next" without you deciding first.

Three views of the same leads, answering different questions. Today — what needs me right now. Engaged Leads — everyone showing genuine interest, grouped by temperature. Lead Pipeline — the full board, every lead at every stage, with filters and search for when you're looking for something specific.

My Performance is your own numbers, not the workspace's — the leads assigned to you or emailed from your inbox, how they're converting, and your progress against your monthly goal. It carries your own worklist too, so you can start the day here instead of Today if you prefer.

Turn on Notify me when follow-ups are due under Settings → Notifications and you get one reminder each morning listing that day's follow-ups. Your browser has to allow notifications too — use the "Enable notifications" banner on any page here.

Two halves, on purpose. Whether you get them is your own preference on the device you're using, so a teammate isn't opted in by your choice. When they may arrive is workspace-wide (days and an hour, in your workspace's timezone) and is deliberately separate from your sending days — you can email on a narrow rhythm and still be at work the rest of the week.

Each morning every seat gets a short plain-text digest of the same worklist — replies waiting, follow-ups due, hot leads, leads going cold — plus goal progress. It exists so you don't have to remember to open the app to find out whether anything is waiting. The numbers match Today because both are built from the same signals.

By design. Each card surfaces a single next action as its main button, and everything else — Call, Emails, Brief, Demo, Deal info, Follow-up, Research, Lead details — sits in collapsible pills you open when you need them. A board of fully expanded cards is a wall of fields; a board of single actions is a to-do list. Clicking the main action opens the pill it relates to.

That's the lead's temperature (see How the pipeline works) — the card's one heat signal, so the board can be scanned at a glance. Colour is deliberately rationed: apart from the stripe, only the Follow-up pill (amber = due today, red = overdue) and an unread reply on the Emails pill are coloured. Demo and Brief show their state as plain text instead, so the board doesn't read as a rainbow.

Use the picker in the card head. A manual choice overrides the automatic reading and holds until the prospect does something new — a reply, a chat, a brief or offer visit — at which point the signals take over again. Freezing a lead you've decided not to chase also keeps it out of the "going cold" bucket on Today.

Searches the web for one to three checkable facts about the business — a second location, an award, a local write-up — each with the page it came from, so you can open with something specific and true. It proposes; you pick. Nothing is saved until you tap a fact, and a fact with no source link is discarded rather than kept, because an unverifiable claim emailed to the business it's about is worse than no claim at all. Saved facts become a dated note on the lead, so whoever picks it up in three months can see where it came from. Part of the AI Assistant add-on; saving a fact you've already found is not.

If your workspace has connected a telephony provider (see CRM & calling integrations), Call now rings your phone first and then bridges the prospect — no softphone, no app. The outcome is logged onto the lead automatically. Without a provider connected the card still offers a plain tap-to-call link and the manual "save call" form.

Notes are the lead's running history — call outcomes, what the owner said, saved research, bounce records, Proto's score. They're worth keeping accurate because the demo build reads them: a sourced 🔎 Research line can earn a mention on the page you send. Everything else in the notes is explicitly off-limits to the build, so a scribbled reminder never ends up on a prospect's demo.

For each candidate Proto loads the live website, captures it, and asks a vision AI to judge it against your rubric. The model returns a structured score plus specific findings — outdated design, broken mobile, missing essentials (no contact/booking/SSL), and business size fit — and a short rationale. Prospects at or above your threshold proceed; the rest are filed away.

In Settings you control the scoring rubric (the system prompt the model follows), the score threshold, the target categories and market, and your pricing guidance. These are per-workspace — change them anytime; they apply to future runs.

Scoring runs on a dedicated worker outside the app (it needs a real browser and longer AI calls). Pressing Start scoring dispatches a run; results stream back into the prospect list as each site is judged. A run can take a while for large batches — you can leave the page and come back.

The most common cause is the usage balance or the monthly spend cap: when AI is paused for billing reasons, new sites are skipped rather than charged. Check Settings → Billing for your balance and cap. Sites that fail to load (dead domains, blockers) are also skipped and noted.

A one-off, branded mockup of what the lead's site could look like, generated automatically from their business details. It's reachable at a short link you can drop into outreach, and it tracks clicks so you know when a lead looked.

Outreach always sends from your own inbox and domain, so the emails are entirely yours. Full white-label demo sites — no "powered by" badge — are included on the Growth plan and above.

By default demo and brief links live on the StellarReach domain. On a qualifying plan you can serve them from your own hostname instead, so a prospect sees your brand end to end. Set it up under Settings → Custom domain; you add a DNS record, we verify it and issue the certificate. The reply address is a separate setting on the same panel — one doesn't require the other.

A preview link stays live for 90 days from the outreach send — long enough to convert, short enough that an unsold demo doesn't sit on the internet forever. Demos can also carry a preview password; emailed links include it automatically, so the prospect never types anything, and anyone who finds the URL another way is asked for it.

Two things move independently. The build evolves the merchant's own brand — their palette, logo, and voice — so different prospects diverge on their own. On top of that your workspace has a design signature: an art direction, style notes, and reference URLs under Settings that govern layout and mood without touching the merchant's colours. Every workspace is assigned a distinct art direction out of the box, so two agencies pitching the same street don't send look-alike pages.

If you've turned on the demo quality check, each build is compared against the prospect's real website before anything sends: first for structural faults, then a before/after audit, then an AI review scoring the redesign against your minimum. A demo that isn't clearly better is held rather than emailed — the outreach only works if the page beats what they already have, and there's no second first email. The card shows the reasons and two choices: Rebuild, or Send anyway if you've looked and disagree.

The demo uses the lead's own phone, email, and address where we have them, and simply leaves out anything we don't — it never invents a plausible-looking number. A page that slips an invented detail through is held rather than sent. If a demo has no address to show, the build puts a plain contact form there instead.

A multi-step sequence, each step a fully editable template in Email Automation: Outreach (the first touch — A/B testable so you can compare two variants), Brief (invites the lead to fill in a short project brief), Nudge (a gentle reminder), and Farewell (a graceful last touch). The AI personalizes each from the lead's details and demo. Several optional stages sit alongside these — see the full list below.

Eight stages can send to a prospect without you. Four are the core sequence: Outreach (the first touch), Brief follow-up (after they click the demo), Nudge (a reminder — and, where you've priced a flat offer, a warm re-offer variant instead), and Farewell (the last touch).

Four more are optional and off unless you switch them on in Settings: Cold-open nudge (they opened the email but never clicked — one cheap "still interested?" pointing back at the demo, instead of building them a full brief), Frozen retry (they never opened at all — one "in case this got buried" note, once per lead, ever), Re-engagement (a lead you'd already said farewell to comes back and visits), Rebuilt demo (you rebuilt a demo for someone who looked once and went quiet, and want to send the new one), and the price follow-up with its price follow-up reminder (they read their scorecard, clicked through to your pricing page and then asked for nothing — one email offering to hold a build slot for them until a stated date, and one reminder before it closes). Every one of them is a real email from your domain, so all ten count against your daily send cap and your inboxes' own limits.

The price follow-up is the only email here that names a deadline, so it is worth knowing exactly what that deadline is. What is held is a build slot — a promise about your own calendar, which you set in Settings. Nothing is taken away from the prospect when it passes: their demo link and their report stay live for the usual 90 days, and a late reply is answered normally. It exists because someone who goes looking for your prices before accepting anything free is telling you they do not want to be handled, and a countdown over something you would withdraw is exactly what they are watching for.

Under Settings → Email Campaign → Demo builder → When to build, in the control called How you approach a cold lead. Send them a finished redesign (the default) generates a demo for every scored lead and sends it — the strongest first touch, and the most expensive, since most of those pages are never opened. Send their scorecard, and build only if they ask emails a report about their existing site instead and only builds the redesign for the ones who press the button, so your build budget follows actual interest.

The third setting is usually the one you want: … except your best leads, at or above score. Your best-fit leads get the full treatment up front; everyone below gets the scorecard and costs a build only if they say yes. Leave it blank to ask everyone — but note that 0 is not "off": a threshold of 0 pre-builds every scored lead, which turns the scorecard path off entirely.

A short page about their site — what's working, what isn't, and what it's costing them — with a button to ask for the redesign. It's a lighter first touch than a finished demo and it costs almost nothing to produce, which is what makes the reach-out-first path work. Reading it counts as engagement on the lead, and clicking through to ask for the demo is recorded as a scorecard-sourced click so you can see the two-step converting in Reports.

No, and deliberately. A lead who filled in your own "free demo" form already knows who you are, so the cold-outreach copy — introducing yourself, explaining why you're in touch — reads as amnesia. Inbound requests get their own prompt and land on the board flagged as a request, paused and marked hot so you see them. A repeat submission warms the existing lead instead of opening a second card.

Your own. Connect Gmail or Outlook / Microsoft 365 (OAuth — you authorize your mailbox) or SendGrid on the Demo Sites / Settings pages. StellarReach drafts and schedules; the mail goes out as you, from your domain and signature. Your team can also connect their own real inboxes, which then take over automated lead email from this one rather than adding to it — read Sending inboxes & mailbox rotation before you do.

Outreach is paced: it's gated to business hours and a daily send cap, so the sequence spreads out instead of blasting. The cap is one workspace-wide allowance counted across every automated stage together — not per inbox and not per queue — so outreach, nudges and farewells all draw on the same number. This protects deliverability and your domain reputation.

Once you connect a seat sending inbox there is a second, separate limit: each connected inbox has its own safe daily rate, and the day stops at whichever of the two is lower (unless you switch on overflow sending, below, which carries on through SendGrid up to the daily cap). That's usually the inboxes, and when it is, raising the daily cap changes nothing. Upcoming Emails shows what's queued, when it will go, and which of the two limits is binding — read Sending inboxes & mailbox rotation before you connect one.

Yes. You choose who to contact and what to send, so consent, suppression, unsubscribe handling, and laws like CAN-SPAM, CASL, and GDPR/PECR are your responsibility. See the Terms.

Each seat on your team can connect its own real Gmail or Outlook / Microsoft 365 inbox on the Team & Inboxes page. From then on every automated lead email — outreach, brief follow-up, nudge, farewell, re-engagement, frozen retry, price follow-up — goes out through the connected inboxes. That's mailbox rotation.

There is no on/off switch for it. Rotation is switched on by the inboxes existing: a workspace has it the moment one inbox is connected with sending enabled, not once you've connected several. Read the next answer before you connect your first one — it changes your daily volume, sometimes downwards.

This is the part that surprises people. Your workspace email provider (the Gmail, Outlook or SendGrid account you connected in Settings) stops carrying automated lead email while any seat inbox is connected. It isn't disconnected or switched off — it still sends your admin notifications, it's still the fallback for manual one-off sends, and it's what overflow sending uses — but the sequence goes through the inboxes instead.

So seat inboxes are not extra capacity on top of what you had. They are your capacity. If your workspace provider is SendGrid, connected precisely because it can carry volume, one teammate connecting their Gmail takes the whole workspace down to that one inbox's safe rate of 25 a day. The upside is real — mail sent from actual human mailboxes lands better than bulk mail, which is the whole reason to do this — but plan the volume first. There is one way to run both together, overflow sending, and it is off by default and only works on SendGrid; see "Overflow sending" under Troubleshooting.

The sum of every connected, sending-enabled inbox's own safe rate: 25 a day for Gmail, 50 for Outlook / Microsoft 365, 25 for anything else. Two Gmail inboxes is 50 a day, two Outlook inboxes is 100. An inbox that's warming up contributes only its reduced ramp figure until it finishes. Team & Inboxes shows the running total.

That total isn't automatically your ceiling, because your daily send cap and your subscription plan's own limit still apply. The number that actually governs the day is the lower of the two, and the app reports which one it is — so if the inboxes are binding, raising the daily cap does nothing and the fix is another inbox. Connecting inboxes never lets you exceed the cap you saved in Settings, or your plan.

Anything above the day's ceiling is not dropped or failed. The lead goes back in the queue untouched and is picked up on a later run — nothing is even drafted for it, so no AI is spent on a send that can't go out. (With overflow sending on, the surplus goes out through SendGrid the same day instead of waiting.) Each inbox's daily count resets at midnight in your workspace's timezone; short of that it waits for capacity to free up, which connecting another inbox does immediately. The manual Send outreach… button on the Lead Pipeline is bounded by the same limits and says so in its dialog ("13 can go out today"), which is why it sometimes offers fewer than the number of approved leads.

No. With no seat inboxes connected, everything behaves exactly as before and sends from your single workspace mailbox — but that's only true while zero are connected. The first connection is the switch.

Connect inboxes when you want mail coming from real human mailboxes rather than one shared account, and when the inboxes you can add add up to the volume you need. If your workspace provider already carries more per day than the inboxes you'd connect, connecting them lowers your volume rather than raising it — turn on overflow sending (SendGrid only) or wait until you have enough inboxes to cover it.

Everyone with a login is on Team & Inboxes, where an admin invites colleagues, changes a role, resets a password, or removes someone. There are two roles. A Team Leader sees every lead in the workspace and owns the settings, billing and integrations. A Team Member works leads and sees only their own — the ones assigned to them, plus (while the auto-assign setting is on) any sent from their own connected inbox. Admins can also switch individual AI helpers on or off per member seat.

A send-only inbox is the exception to all of that: a real mailbox added purely so rotation can send from it, with no login and nobody to assign leads to. Use it for a shared address like hello@ that should carry volume but isn't a person. Your plan sets how many seats you can have.

Each email goes from whichever connected inbox is least used relative to its own safe rate, so inboxes of different sizes fill proportionally and reach their limits together rather than one being drained first. A lead then sticks with the inbox that first emailed it for the rest of its sequence, so the prospect keeps hearing from one person. If that inbox is spent for the day the lead waits for it rather than switching to someone else; if the inbox is paused or removed, the lead moves to another connected inbox and stays with that one from then on.

With overflow sending switched on, stickiness also runs the other way: a lead whose earlier email went out through the workspace provider stays on the workspace provider, even after seat inboxes free up. (With overflow off — the default — such a lead does move onto a seat inbox.)

Not automatically back into StellarReach — that depends on the provider, exactly as it does for the workspace mailbox. For an Outlook / Microsoft 365 seat inbox we read the mailbox that sent the mail, so replies are captured, classified and moved onto the lead. Gmail seat inboxes are not read (that needs a restricted Google scope), so a reply simply lands in that rep's own Gmail.

Turn on reply capture (see Inbox, bounces & engaged leads) and it changes that routing: each email carries a per-lead reply address on our relay whichever inbox sent it, so the reply reaches us first and is then forwarded on — to the seat inbox that sent the original where there is one, and to your workspace inbox otherwise. Reply capture is for Gmail / Workspace and SendGrid workspaces; an Outlook / Microsoft 365 workspace already reads replies from its connected mailbox, so the switch does nothing there.

An admin can add a real Gmail or Outlook mailbox to the rotation without creating a login — nobody signs in as it, and its leads stay in the admin pool. It's the way to raise your daily ceiling without hiring people: each one counts toward the total exactly like a seat's inbox. Each uses a seat, and it's a Growth-plan-and-up feature. Add one from Team & Inboxes.

One difference worth knowing: send-only inboxes never receive overflow leads. Overflow hands each lead to a person so it shows up on someone's board, and a send-only inbox has no person behind it.

Pause sending on (or remove) every connected inbox, on Team & Inboxes. With none left, all automated email goes back to your workspace provider — including leads that were previously stuck to an inbox — and that provider's own limits apply again. Pausing just some of them doesn't do this: while any inbox is still active, its leads move onto the remaining inboxes rather than back to the provider.

Fresh inboxes that suddenly send at full volume get flagged. To avoid that, you can turn on warm-up for any seat inbox on the Team & Inboxes page: a warming inbox starts at a tenth of its normal rate — about 3 a day for Gmail, 5 for Outlook — and ramps up to the full rate over 21 days.

Warm-up is opt-in per inbox and connecting one never starts it: an established daily-driver mailbox sends at its full rate from day one. While an inbox is warming it contributes only that reduced figure to your daily ceiling, so a workspace whose only inbox is warming is sending around 3 a day, not 25.

The ramp above paces your own sending; a warm-up network does the other half — seeding positive engagement by exchanging mail between inboxes. StellarReach doesn't run that itself, but it can surface each inbox's warm-up reputation in one place. Connect a provider (Instantly today) with an API key on the Sending Inboxes page, and we read each seat mailbox's warm-up score via the provider's API. You connect the mailboxes inside the provider's own secure flow — StellarReach only reads the scores and never sees your mailbox credentials. The API key is encrypted at rest.

Reputation is mostly about your domain, and every inbox on the same domain shares it. For cold outreach the best practice is a dedicated sending domain (e.g. outreach-yourbrand.com) kept separate from your primary, with SPF, DKIM, and DMARC set up before you run real volume. The Sending Inboxes page summarizes the authentication status of your connected inboxes' domains so you can get this right first.

For connected Outlook / Microsoft 365 mailboxes, StellarReach watches the inbox and classifies new mail: a bounce clears the bad address and requeues an email lookup, and a genuine reply moves the lead to Engaged and notifies you (and pushes it to your CRM). It also quietly handles the noise — an unsubscribe / opt-out is suppressed so the lead is never contacted again, and an out-of-office autoresponder is ignored, so neither pollutes your engagement numbers.

Reading the mailbox directly is Outlook-only. Gmail and SendGrid workspaces get the same result a different way: turn on reply capture under Settings → Reply capture and each outreach email carries a per-lead reply address on our relay domain. A prospect's reply reaches us, gets classified exactly like an Outlook one, and is also forwarded into your real inbox with the prospect as the reply-to — so you keep the conversation where you normally have it. Keep your own reply-to address as your own; the relay address is added behind the scenes. Bounces are unaffected either way — they go back to your sending provider, never the relay. The switch does nothing for Outlook workspaces, which already capture replies through the connected mailbox.

Inbox is a back-office inbox that collects every genuine reply and opt-out in one place — read or unread, deduplicated per message — so you can triage inbound without leaving StellarReach. Lead brief-chat conversations show up here too (see AI assistants below), each as a single running transcript per lead that re-flags as unread on new activity.

Engaged Leads lists everyone who actually read your demo, replied, called, chatted, or submitted a brief — grouped by temperature (bare email-opens and un-read demo clicks are hidden). From there you can read a lead's submitted project brief, draft the follow-up brief email with AI, edit lead details, and mark progress toward Won.

Lead Pipeline is the broader board of where every lead sits across the stages — a good place to spot leads that have stalled and need a manual touch.

Set a price under Settings and every demo you serve gains a "Get your $X redesign" button leading to a one-page offer with a short acceptance form. A five-page site for a plumber doesn't need a discovery call — it needs a number and a way to say yes, at 11pm on a Sunday. Your booking link drops to a small "or book a call" line beneath it. The price is the on switch: leave it blank and nothing changes anywhere.

Because it's a starting price. The demo's button carries the first line of your disclaimer and the offer page carries all of it. Don't remove it — the whole point of the three scope questions below is that some jobs won't land at the flat price.

The form asks how many pages, whether they have a logo, and whether they need anything transactional (booking, ordering, logins, payments). Those are the three things that decide whether the price still holds. Every acceptance arrives already sorted: at the flat price (send the invoice) or needs a conversation (call them) — more pages than your offer includes, a logo that has to be designed from nothing, or anything with moving parts. A plain contact form is deliberately in scope and isn't asked about.

Connect your own Stripe account under Settings, and the lead card can raise the invoice in two halves — a deposit up front (50% by default) and the balance on delivery. The money goes to your account and the invoice comes from your business; StellarReach never touches it. Sales tax is handled by Stripe Tax on your account, not a rate we type in. Once a half has been invoiced the button won't raise it a second time.

Optional: with pay now switched on, an acceptance that lands at the flat price raises the deposit invoice straight away and puts a Pay button on the prospect's confirmation page, without waiting for you. It only ever fires for an acceptance at the agreed price, only once per lead, and if anything goes wrong the acceptance is still saved and you invoice from the card as usual.

If you connected the payment webhook, paid invoices tick themselves. Two real payments can't be seen that way — one settled outside Stripe, and one that landed before the webhook existed — so an admin can mark a half paid manually. The card distinguishes them ("paid" vs "marked paid"), and only a manual stamp can be undone; a payment Stripe reported has to be voided or refunded in Stripe. Marking the deposit paid moves the lead to Won, exactly as an automatic payment would.

Three states show up in the 💼 Deal info pill. They opened the offer and never answered — worth a direct follow-up while the price is fresh. Balance still to collect — the deposit settled and the final half was never raised; shown in green, because only you know when the work shipped. Overdue — invoiced, past terms, no payment recorded. Without the payment webhook connected, "no payment recorded" means we don't know, so nothing is ever called overdue. The board's Offer filter cuts the same cohorts.

Gives you a folder that deploys as-is: the page, its assets, a config file, and a README checklist. It does the three things a demo has no reason to get right — downloads the images that were hotlinked from the client's old site (those URLs die the moment the old host goes away), wires the forms up so submissions actually reach someone, and names every form field, with the resulting column headings listed in the README. Anything too large to bundle keeps its original link and is named in the README rather than silently dropped.

The automated version: puts the site in a private repo in your GitHub account and publishes it to your Netlify account, using tokens you connect under Settings. Both are always your accounts — there's no shared fallback. It runs in the background, so the lead card is where the result (or the reason it stopped) appears.

It's the one lead action that creates real infrastructure outside StellarReach — a repository and a public website in your agency's own accounts, under your name. The seat working the lead doesn't get to make that call.

Only if the repo hasn't moved on. A redelivery replaces the files wholesale, so if anyone has committed to the repo since we last delivered, the deploy stops rather than erasing their work, and says so on the card. Export the new version and merge it by hand instead.

Open any lead and each section carries a contextual AI helper grounded in that lead's signals: draft the next email, prep a call or its voicemail, draft a reply, suggest the next step, plan follow-ups, anticipate objections, find the decision-maker, summarize the brief, and draft a proposal or pricing. Up top, Leads to work scans your whole pipeline and surfaces just the leads that need a touch right now. These ship with the AI Assistant add-on (add it in Settings → Billing); admins can also toggle which helpers each member seat gets.

Stella is the floating chat widget (bottom corner, collapsed by default) on the marketing pages and on every back-office page. She answers questions about how StellarReach works, helps with setup, and hands off to support@stellarreach.app when a human is needed. She runs on the platform's AI — using her never touches your usage balance.

On the project-brief page you send to a lead (the /f/… link), an optional "Questions about your brief?" chat answers their questions in your agency's voice, grounded in their own brief. It deliberately never quotes a price, promises a timeline, or commits you — those hand off to you. Because the transcript is shared with you, the widget shows the lead a short privacy note.

It's BYOK-only (Scale & Agency): it runs on your own AI key so an unauthenticated lead chat can never draw down the platform balance, and it bills to your provider directly. It's on by default for eligible workspaces; toggle it under Settings → Brief assistant. In platform-key mode the toggle has no effect.

A lead starting a brief chat counts as interest: it bumps the lead's chat counters on the lead card, the first chat fires a one-time notification to you, and the full transcript lands in your Inbox so you can see exactly what they asked.

Zoho CRM, HubSpot, Pipedrive, Close, Salesforce, GoHighLevel, EspoCRM, and a generic webhook for Zapier, Make, or your own endpoint. You connect one per workspace from Settings → CRM integration; leads land in your account, never ours.

Two ways: a one-click Send to CRM button on a lead, or best-effort auto-sync the first time a lead engages. Each lead is de-duplicated, so it's pushed once, not on every signal.

No — that's the point of the split. StellarReach is where you work a lead toward a deal; your CRM is the system of record the rest of your business reads. Leads push out to it so nobody has to re-key anything, but a rep never has to open it.

Connect Twilio or Telnyx under Settings → Calling integration and the card's Call now button places the call: your phone rings first, then the prospect is bridged in — no softphone to install. The prospect leg's outcome is logged onto the lead automatically, so the call history builds itself.

Nothing here. It's bring-your-own-provider: you connect your own Twilio or Telnyx account and they bill you for the calls directly. Calling is never metered against your usage balance. With no provider connected, the card keeps its ordinary tap-to-call link and the manual call-logging form.

On platform keys (Starter and Growth) the AI runs on our provider accounts and you prepay a usage balance. On Scale & Agency you can switch to BYOK: connect your own Anthropic and Google keys, pay those providers directly, and we meter only the scoring worker's infrastructure. Your keys are encrypted at rest.

On platform keys, all AI and Maps usage is metered against your balance. On BYOK, only the scoring worker is metered — your own keys cover everything else, billed to you by the providers directly.

Yes. Every workspace has a hard monthly spend cap, separate from the prepaid balance. When the cap is hit, AI pauses automatically — no surprise bills. Set it in Settings.

Your monthly plan pays for platform access and capabilities and includes no usage. AI usage draws down a prepaid balance that auto-recharges your saved card when it runs low. On Scale & Agency with your own keys, your providers bill you directly instead. Manage it all in Settings → Billing via the Stripe portal.

If auto-recharge can't top up, AI pauses (the same safe-stop as the monthly cap) until the balance is restored. Your data and settings stay exactly as they were.

Change plans or cancel from Settings → Billing through the Stripe customer portal; cancellation takes effect at the end of the current period. If a subscription payment fails, we retry and email you; if it can't be recovered the workspace is paused until billing is resolved, then reactivates automatically.

No. Payments run through Stripe; full card numbers are handled and stored by Stripe, never by StellarReach.

The demo works because it's a surprise; the second one to land in the same inbox is a commodity. When you reach a business first, you can hold it — while the hold lasts, no other StellarReach workspace's outreach to that business goes out. The card shows 🔒 Claimed · yours until <date>, and the hold rolls forward each time the prospect engages.

Another StellarReach workspace got there first, so your outreach to that business is on hold until their claim lapses — you'll see the date. Nothing is lost: come back after it lapses. You'll also see 📡 Contacted elsewhere on a business someone else has already emailed through StellarReach; there, an automatic demo build is skipped to save the spend, but you can still build and pursue it by hand.

Holds start free and lapse on a date. Within about a week of lapsing, an admin gets a Keep locked · $5/mo option on the card to keep holding it while you work it, and a Release button to stop the charge at any time. If you're paying to hold quiet leads, a weekly email lists exactly which ones and what they cost — a forgotten hold is money leaking, so you get told rather than quietly billed.

A hold inside StellarReach stops StellarReach users. It can't stop the freelancer down the road who has never heard of us. It's also anonymous in both directions — another workspace can see that a business is taken, never by whom, and you can't see whose a claim is either. And enforcement only ever acts on a positive, confirmed, paid-up hold: if anything is uncertain the email goes out, because wrongly blocking your send is worse than missing a block.

Reports summarizes your funnel — how many prospects were found and scored, how many emails went out, and open / click / engagement rates — so you can see what's converting and where leads drop off.

That's what the Funnel Tuner is for. It shows the same funnel as a chain of steps with the widest drop called out, then walks you through the settings behind each step — screening scores, sending caps and hours, subject-line tests, demo style, follow-up automations — one stage at a time. With the AI Assistant add-on it also reads your numbers and pre-fills what it would change, with the reason next to each suggestion. Nothing is saved until you apply it, and you can tune every setting by hand whether or not you have the add-on.

Your own spend against the cap is shown in Settings → Billing. It updates as AI usage is metered, so you always know where the balance stands.

Send Tracker follows a cohort rather than a total: each batch is one day's wave of outreach, tracked from first send through frozen → cold → lukewarm → warm → hot, with the running cost alongside. It answers the question the funnel percentages can't — "did the emails we sent on Tuesday actually turn into anything, and what did that cost?"

Set a start date so the analysis begins where your real selling did. There are two, on purpose. The Funnel Tuner's start date narrows what gets counted, so a leak you fixed in the spring stops dragging on the averages — move it around freely to diagnose a period. The Lead Pipeline's start date (in the board's Filters panel) narrows what reps see, to bury old history. That one deliberately keeps any older lead that still has a follow-up set, reached engaged or won, or has a reply waiting — burying history should never hide a live deal.

Funnel Blueprint asks about your trade, your market, what makes a lead worth chasing, what should get built for each prospect, and when a person should take over — then writes the configuration: how leads are scored, the scoring rubric and target categories, the demo-build and project-brief prompts, an email prompt for each stage you switched on, and the cadence and hand-off settings. It's the same shape of configuration a starter preset gives you, except written for your business instead of picked off a shelf.

No. It's about twenty small questions, mostly pick-one or pick-many, asked one at a time with a progress bar — nothing asks you to know how the product works. You can go back, skip, and change any answer from the summary screen at the end. Progress saves as you go, and generation runs in the background, so you can close the page and come back.

Partly. Paste your URL and it pre-fills the questions about your business — but only pre-fills. A marketing homepage is written to impress visitors, not to configure a prospecting tool, so the answers are a starting point you check. It never overwrites something you've already answered, and it won't touch the decisions no website has an opinion about: how picky to be, when to send, and when a person takes over.

Never. It can write a first prompt, but a prompt you've edited is yours and it refuses to replace it — there's no override. It also leaves money, credentials, and your brand identity alone entirely, and tells you on screen which settings it deliberately didn't touch. You review everything, and can edit any of it, before a single value is saved; applying it is all-or-nothing, and it can be undone.

Blueprint is for a workspace that hasn't sold yet — it authors a configuration from scratch. Funnel Tuner is for one that has — it reads your measured funnel and adjusts what's already there. They can't collide: the Tuner needs real sending history before it will say anything, which is exactly what a Blueprint workspace doesn't have yet.

Because scoring looks at a screenshot of a prospect's website and little else. A criterion it can't observe there doesn't get skipped — it gets guessed, on every prospect, and the guess is indistinguishable from an observation once it's scored. So criteria that can't be seen are dropped and named on the review screen, and the Blueprint prefers an observable proxy instead. This is the honest boundary of the whole product: it qualifies prospects by looking at their website.

In a managed PostgreSQL database, isolated per workspace. Connected API keys and OAuth tokens are encrypted at rest (AES-256-GCM); passwords are hashed with scrypt. More on the Security page.

No. StellarReach is multi-tenant with row-level workspace scoping — each workspace's leads, settings, keys, and inbox stay isolated from every other.

Yes. Prospect contact details and your reps' notes are encrypted at rest under a key belonging to your workspace alone, so a compromised database — a leaked connection string, a restored backup, a read replica — yields ciphertext rather than your list. To be straight about the limit: this protects against someone reaching the database, not against the platform. The senders and the pages your prospects open have to read this data with nobody present, so the running application can always decrypt it. What it does buy outright is deletion: removing a workspace destroys its key, which puts its data beyond reach in backups too.

Your data is yours. Email support@stellarreach.app to request an export or deletion; see the Privacy Notice for details.

Check, in order: is a sending inbox connected and still authorized (Settings)? Are you inside business hours — the sequence waits for them? Has the daily send cap already been hit? Or have your connected seat inboxes sent all they safely can? That's a different state with a different fix — it clears overnight, and raising the daily cap won't touch it (see Sending inboxes & mailbox rotation). And is billing healthy — a paused workspace won't send. Upcoming Emails shows what's queued, why, and which limit is binding. If nothing at all is going out, see the next answer first.

Two things set a cap of 0, and the dashboard's sending check says which. Either someone typed it, or the safety brake tripped: if too many of a day's emails bounce or get blocked, sending stops automatically to protect your domain's reputation — a reputation that takes weeks of careful sending to rebuild, which is why it stops the same day rather than waiting for a weekly report. It does not re-arm itself, and waiting won't help. Look at why first: hard bounces mean a bad list (usually guessed addresses), blocks mean receiving servers are refusing your domain. Then raise the cap deliberately in Settings.

Connecting any seat inbox switches the whole workspace to mailbox rotation, and rotation replaces your workspace provider for automated email rather than adding to it. So a workspace sending through SendGrid, where one teammate connects their Gmail, stops using SendGrid for the sequence entirely and inherits that one inbox's safe rate: 25 a day. One Outlook inbox instead gives 50. If the inbox is warming up, it starts nearer 3 a day and climbs over three weeks. Everything above the ceiling isn't lost — it waits and goes out on a later day.

Your real ceiling is the sum of your connected inboxes' rates, shown on Team & Inboxes, and Upcoming Emails names whichever limit is binding — the daily cap or your inboxes. If it's the inboxes, raising the daily cap does nothing.

Three ways forward. Connect more inboxes — each one adds its own rate, and spreading real mailboxes is what protects deliverability in the first place. Turn on overflow sending (below) so SendGrid sends whatever the inboxes can't. Or pause sending on every seat inbox (Team & Inboxes → Pause sending); with none active, the workspace provider takes over again exactly as it did before.

By default the two don't run side by side: connect one seat inbox and rotation owns all automated sending (see Sending inboxes & mailbox rotation). Overflow sending (Settings → Team & inboxes → Overflow sending, off by default) changes that. Your connected inboxes still send first, up to their own safe rates; once they're spent for the day, the rest of the queue goes out through SendGrid instead of waiting — still bounded by your daily send cap.

Every overflow lead is handed to a person: it goes to whichever seat has the fewest leads among those with a connected sending inbox, so it shows up on that rep's board and the email carries their name, phone and booking link. Send-only inboxes are skipped — nobody works that queue — and a lead that already has an owner keeps it. Assignment is best-effort: if it fails the email still goes out and you have a lead to assign by hand, rather than an email that never sent. This matters because a lead with no owner and no sending inbox behind it belongs to nobody and is hidden from every rep's board.

The email sends from your workspace address rather than the rep's, because that's the address your domain is set up to authenticate; sending as someone's personal mailbox through SendGrid is what gets mail filtered. Replies come back to the workspace address too — only the signature is theirs.

Three things it deliberately won't do. A lead that has already had an email from a rep's inbox never switches to the workspace address mid-sequence — it waits for that rep's inbox instead, so the prospect keeps hearing from one person, which also means overflow won't clear a backlog stuck behind a full inbox. The mirror of that rule applies too: a lead first sent through the workspace provider stays there for the rest of its sequence rather than migrating onto rotation later. And it only applies on SendGrid: overflowing onto a Gmail or Outlook workspace mailbox would push that single account past the limits the per-inbox rates exist to respect.

The manual Send outreach… button overflows too, exactly like the scheduled runs. Its pre-flight dialog is bounded by both limits — your daily cap and what the inboxes can still take — and names whichever one is binding, so it can offer fewer than the number of approved leads for either reason.

Almost always the Lead Pipeline start date in the board's Filters panel — it's workspace-wide, not per-browser, so "Clear filters" deliberately doesn't reset it. Leads created before that date are hidden, unless they have a follow-up set, reached engaged or won, or have a reply waiting. Also check the Archived chip and whether a filter or search is still applied.

The mailbox grant has died — someone changed the account password, revoked access, an admin removed the app, or it simply went unused for months. Nothing sends from that inbox until it's reconnected, so we check daily and tell you rather than letting it fail quietly. Reconnect it on Team & Inboxes (or Settings for the workspace mailbox); it's the same OAuth click as the first time.

Usually billing: an exhausted balance or a reached monthly cap pauses AI, so the worker has no work to claim. Confirm your balance and cap in Settings → Billing. If billing is fine and a run still won't start, try Start scoring again — a previous run may still be finishing.

Confirm a CRM is connected for this workspace in Settings → CRM integration and that its credentials are still valid. Auto-sync is best-effort and fires once on first engagement; if it missed, use the manual Send to CRM button.

That's almost always the usage balance auto-recharge topping up after AI usage drew it down. Settings → Billing shows the balance and recharge history; you can adjust the cap to control how much can be spent.