If you are doing outbound right now, you probably have a list, a sequence and a nagging sense that your messages land in the same pile as everyone else's. Every company gets the same email on the same schedule, whether it just opened the exact role you could help with or has not changed in a year. Sending more does not fix that. Better timing does.
Signal-based outbound means contacting a company because something observable has just changed, such as a new hire, a job post, a funding round or a tool switch, instead of because it appears on a list. The signal gives you a real reason to write today, so you send fewer, better-timed messages.
Disclosure: some tool links in this guide are referral links from our AI Stack page. If you sign up through them we may earn a commission at no extra cost to you. We list tools we use or have vetted, and you can run every step here with other tools.
What you will have at the end
- Three chosen signals, a signal log with proof for every row, a verification checklist and a send, queue or drop rubric.
- Three message templates, one drafting prompt (the only prompt here), a touch plan, a reply rule table, a KPI table and a seven-day plan.
- For the wider system, start with our guide to AI agents for lead generation, and read what an AI agent is if the term is new.
Start here: the tech stack and where the orchestration runs
Signal-based outbound is a chain of jobs: detect, verify, score, write, send. The orchestrator runs the chain, hands each result to the next step and stops for your approval. Choose it first, because it decides where your files live and how often the chain runs. All four choices use the same worksheets and templates later in this guide.
Asset 1: Stack map, pick your orchestrator| Path | What it is | Pick it when | Trade-off |
|---|---|---|---|
| A. Spreadsheet plus an AI assistant | The log lives in a sheet and you work through the checklists by hand. An assistant such as Claude is optional, for the one drafting prompt in Step 5. | You are proving the method, or signals arrive at a trickle. | Manual work, but no build time and you see every row. |
| B. Claude Code | Claude Code reads your files, runs commands and connects to other tools through MCP. Per Anthropic's docs it runs in the terminal, IDE extensions, a desktop app and the browser. We run our own GTM workflows through it. | You are comfortable with files and a terminal and want verify, score and draft in one repeatable chain. | More setup than path A. Keep every send behind a human approval step. |
| C. A workflow engine | Make is the no-code automation layer we connect most client stacks with. n8n is an open-source workflow engine we run alongside Make to orchestrate outbound. | You want triggers (a new row, a schedule) and app-to-app connections without code. | Many small steps to maintain. Check each vendor's documentation for limits and pricing. |
| D. Ready-made | CampaignFlow Pro is our own campaign orchestration platform. Deepline provides enrichment, lead and account discovery and signal detection, run through a Claude Code skill. | You would rather not build the chain yourself. | Less control over each step. Review the vendor's documentation first. |
Decision rules. If you are unsure, start with path A for the first week. Move to path B or C when a manual step becomes the bottleneck in your KPI table. Many setups mix them: an orchestrator for verify, score and draft, and a sending platform for delivery. Whatever you pick, a person approves drafts before anything is sent. Anthropic's guidance on building agents is to use the simplest solution that works and add complexity only when it clearly improves results, which is why path A comes first. (Source: Anthropic, "Building effective agents", accessed 9 November 2026.)
Tools that can run these steps
The descriptions follow our AI Stack page. Check each vendor's documentation for current features, limits and pricing before you buy. The guide works without any of them: the last column is enough for a pilot.
Asset 2: Tools by step| Step | Tool | How it fits | No tool needed |
|---|---|---|---|
| Step 5: draft the message | Claude, Perplexity | Claude: the assistant we use daily to write, build and run our stack. Perplexity: our default AI search engine for cited research. | Write from the templates |
| Step 2: detect signals | TheirStack, Apify, Phantom Buster, Deepline | TheirStack: tech stack and job posting data. Apify and Phantom Buster: scrapers (check each source's terms first). Deepline: signal detection. | Check careers and news pages by hand |
| Contacts for rows that pass | Findymail, Bouncify, ZeroBounce | Find and verify decision-maker emails so touches do not bounce | Look up 20 contacts by hand |
| Step 6: email touches | Instantly, Lemlist, Zapmail | Instantly: cold email sending with built-in deliverability. Lemlist: email, LinkedIn and calling in one sequence builder. Zapmail: done-for-you inbox infrastructure. | Your normal mailbox for a pilot |
| Step 6: LinkedIn touch | HeyReach, GetSales | LinkedIn outreach tools; use them within the platform's rules | Manual notes |
| Step 7: replies, CRM, booking | Breakcold, Pipedrive, Cal.com | Breakcold: social-selling CRM. Pipedrive: simple visual CRM pipeline. Cal.com: our preferred booking tool. | Shared inbox and a calendar link |
| Log and approvals | Google Workspace | Email, docs and file storage; the log and approval list can live in a sheet | Any spreadsheet |
Path B in practice: a Claude Code project
We have not run your accounts, so treat this as a template to adapt.
- Create a project folder with
icp.md(ideal customer and exclusions),signals.md(three signals, freshness windows, hypotheses and the rubric),signal_log.csv, atemplatesfolder andsuppression.csv. - Optionally add a
CLAUDE.mdconfig file in the project root. Anthropic's docs say Claude Code reads it at the start of every session. The starter below is a configuration file, not a prompt, and you can skip it on paths A, C and D. - Package the weekly chain as a skill (per the docs, skills package repeatable workflows): check new rows against the verification checklist, score them, draft the "send now" rows.
- Connect tools through MCP where a connector exists, and use CSV files in and out where one does not. Check each vendor's documentation for connector support.
- Schedule the weekly pass only after you trust a manual run (the docs describe cloud scheduled tasks). Keep sending separate: the agent writes drafts with the status "needs approval", a person approves and your sending platform delivers.
# Signal-based outbound project rules
Goal: find companies matching icp.md with a fresh, verified signal.
Draft one short message each into drafts.csv ("needs approval").
Never send anything.
Rules
- Use only facts in the source text. Unknown if missing. No guessed dates.
- Drop signals older than their window.
- Score fit first, then signal, with the rubric in signals.md.
- Check suppression.csv and the icp.md exclusions before drafting.
- Drafts: 90 words max, one fact, the inference as a question,
one ask, an opt-out line. No private information.
"Run the weekly pass": check rows where verified is empty, score the
verified rows, draft the "send now" rows, print counts per decision.
Path C in practice: Make or n8n
Build one flow per signal type, or one that routes by signal type: a new log row or daily schedule triggers it, the source text is stored in the row, a person ticks the Step 3 checklist (or a filter checks the date against the freshness window), a sheet formula applies the Step 4 rubric, the right template is filled, and drafts land in a "needs approval" sheet. Approved rows join a campaign set to stop on reply or unsubscribe. We have not published exact module or field names, so check the vendor's documentation for each.
What you need before you start
- Your offer and ideal customer in one paragraph each, plus your best three past customers.
- Fifty to 100 companies that already fit your ideal customer, in a spreadsheet.
- A sending mailbox on a domain that is not your main domain, with SPF, DKIM and DMARC set up (see the compliance section), a suppression list, and a named person who approves drafts and owns replies.
Does it work? It works when the signal is real and recent, the company is a genuine fit, and your message connects the two in one honest sentence. We do not quote a reply-rate benchmark, because we have not found one we can source and verify. Measure your own results with the KPI table below.
Signal-based outbound versus list-based outbound
List-based outbound starts with who: you message every company that matches a profile on a schedule. Signal-based outbound starts with when: you message only companies with evidence of a reason to care now. Your ideal customer profile still matters, because a signal never rescues a bad fit; the signal decides the timing and the first sentence.
List-based outbound sends high, steady volume with a generic opening line, and its main risk is looking like spam. Signal-based outbound sends lower, uneven volume with a real, checkable first sentence, and its main risks are a stale or wrong signal and sounding like surveillance. It suits considered B2B services where timing drives the decision.
The signal menu: what to watch and what each one means
A signal is only useful if you can say what it implies about the buyer's situation. Use this table to choose three signals to start with. Do not try to track all of them. Pick the ones that map most directly to the problem you solve.
Asset 4: Signal menu| Signal | Where you can see it | What it usually implies | Freshness window | Message angle |
|---|---|---|---|---|
| Hiring for a role you could replace or support | Company careers page, job boards | They have the problem and are paying to solve it a slow way | While the post is live | Offer a faster or cheaper alternative, or a bridge until the hire lands |
| New senior hire in the buyer's function | Public profile job changes, company news | A new leader reviews tools and vendors early in the role | First 90 days (we split it: 0 to 30 and 30 to 90 days) | Help with the first-quarter problems that role usually inherits |
| Funding announcement | Company press page, news coverage | New budget and pressure to show growth | Roughly the first few weeks after the announcement | Connect your offer to the growth plan they stated publicly |
| Expansion: new office, market or product | Company news, product pages | New process needs and capacity gaps | Weeks after the announcement | Offer to remove one specific bottleneck of the expansion |
| Tool or stack change | Product pages, public tech-stack data you are licensed to use | A migration or consolidation is in progress, usually with messy process | While the change is under way | Help them avoid the known failure mode of that migration |
| Past client or past conversation goes quiet | Your own CRM | Old context and a reason to check in | Twelve months of silence is a common cut-off | What changed since you last spoke |
Two rules from our own setup: write each signal as its own mini playbook (how it is detected, subject options, an email body and a shorter LinkedIn version), and split time-based signals into windows, because what a new hire needs in week two differs from month three.
Step 1: Choose three signals and write the hypothesis
Input: your offer and your best three past customers. Do: for each candidate signal in the menu (Asset 4), fill one row of the worksheet below. If you cannot complete the sentence, drop the signal. Output: three signals, each with a hypothesis, a false-positive screen and a two-minute check. Check: a colleague can verify any one of them from public pages in under two minutes.
Asset 5: Signal hypothesis worksheet| Column | What to write | Rule |
|---|---|---|
| Signal | One row per candidate from Asset 4 (three to five) | Use the menu wording |
| Hypothesis sentence | "When a company [signal], it probably has [problem], which my offer solves by [outcome]." | If you cannot finish it, drop the signal |
| False-positive explanation | One other reason the signal could appear that would make it useless to you | Turn it into a screen, such as "the job description must mention reporting" |
| Two-minute check | How a person confirms it from public pages only | If it takes longer, the signal is too expensive to run |
| Link strength | Strong, medium or weak | Strong: the signal names the problem you solve. Medium: it suggests it. Weak: it only correlates. Drop weak signals for now |
Rank the survivors and keep the strongest three. Track the fourth only after four weeks of data.
Step 2: Build the watch list and the signal log
Input: 50 to 100 companies that already fit your ideal customer profile. Do: put them in a table with the columns below. Output: a log that records each signal with proof. Check: every row has a source link and a date before it moves on.
- Company, website, contact name, job title, profile link, fit score (from your ideal customer rubric, see the lead generation guide)
- Signal type (from your three), signal summary (one sentence, plain facts), signal date, source link
- Verified (yes, no) and by whom; decision (send now, queue, drop); next touch date; status
Source the signals from places you are allowed to use: the company's own site and press page, public job posts, news coverage, and your own CRM. Check the terms of any platform or data tool before you automate collection from it. Many platforms restrict scraping and automated activity in their terms, and breaking them can cost you the account.
In our setup we do not scale anything before a pilot. Run the whole table on 10 to 20 companies first, read every row, and estimate the cost of running the next 100 before you do. Re-run only the rows that failed, never the whole table.
Step 3: Verify the signal before you use it
Input: a log row with a source link. Do: open the source and fill the checklist below, one line per signal. Output: a keep or drop verdict for every row. Check: a second person re-checks five rows and gets the same verdicts.
The most expensive error in signal-based outbound is writing "congrats on the new role" to someone who changed jobs a year ago, or referencing a job post that was filled last month. Verification is a separate step so that nothing unproven reaches a draft.
Asset 6: Signal verification checklist| Signal text (copy the exact words) | Source link | Date in window? | Company match? | Verdict |
|---|---|---|---|---|
| "build and maintain weekly operations performance reports in spreadsheets" | Careers page link | Yes: 2026-11-03, window is while the post is live | Yes: same legal name and site | Keep |
| "Head of Operations joined last year" | Profile link | No: older than 90 days | Yes | Drop |
| "expanding to a new market" (no date, no quote) | Blog repost | Unknown | Parent company, not the target | Drop |
The three rows are illustrative. The rules behind the verdict column:
- Signal text: you can copy an exact quote from the source. If you can only paraphrase it, the signal is unsupported.
- Source link: opens the page that states the event, not a search result or an aggregator summary.
- Date in window: the event date is written in the source and falls inside the freshness window for that signal (Asset 4). Never guess a missing date; unknown means drop.
- Company match: the page is about this exact company, not a similarly named company, a subsidiary or a parent.
- Verdict: keep only when every column passes. Any "no" or "unknown" means drop, with the reason written in the log.
Step 4: Score fit and signal, then decide send, queue or drop
Input: verified rows. Do: score each row with the rubric. Output: a decision per row. Check: the rubric ranks a strong-signal, poor-fit company as drop.
Fit comes first. Signal strength second. A strong signal at a poor-fit company is a drop, not a send.
Asset 7: Send-queue-drop rubric| Criterion | Points | Evidence required |
|---|---|---|
| Fits the ideal customer (industry, size, region) | 0 to 3 | Wording from the company's own site |
| Contact owns the problem | 0 to 2 | Job title and scope |
| Signal is verified and inside its freshness window | 0 to 3 | Source link, date, quote |
| Signal links directly to your offer (strong, not medium) | 0 to 2 | Your hypothesis sentence from Step 1 |
- 8 to 10 points: send now.
- 5 to 7 points: queue. Re-check in a week. Either a second signal arrives or the first one expires.
- 0 to 4 points, or any exclusion hit: drop. Exclusions are existing clients, competitors, partners and anyone who has asked not to be contacted.
Two signals at one company are worth more than one strong signal alone, but only count a second signal if it is independent (a job post and a funding round, not two job posts for the same team).
Step 5: Write the message, one signal and one ask
Input: a "send now" row with its verified fact. Do: fill the template for that signal type, or draft with the prompt below. Output: an email of about 90 words and a short LinkedIn note. Check: read it aloud and test it against the rules.
Rules that keep a signal message from sounding like surveillance: reference public, professional information only and only the one fact that matters; state the fact plainly, then the inference as a question ("often means... is that the case for you?"); give one sentence on why you are relevant, with no company history; make one low-effort ask, such as a short call or permission to send a two-line idea; and stay under about 90 words.
Start with the templates. Edit the wording until it sounds like you. These are fictional and use placeholders.
Asset 8: Three signal templatesTEMPLATE A: Hiring signal
Subject: <ROLE> role at <COMPANY>
<FIRST_NAME>,
Saw <COMPANY> is hiring a <ROLE>. Teams hiring for that role are often trying to fix <PROBLEM>, and a hire takes a while to pay off.
We help with <PROBLEM> by <ONE_LINE_METHOD>, either instead of the hire or until the person starts.
Is <PROBLEM> the reason for the role? Happy to share how we would approach it in 15 minutes.
<SENDER_NAME>
If this is not relevant, tell me and I will not follow up.
TEMPLATE B: New leader (days 0 to 30)
Subject: first weeks as <TITLE>
<FIRST_NAME>,
Noticed you joined <COMPANY> as <TITLE> recently. New leaders in that seat often inherit <COMMON_INHERITED_PROBLEM>.
We help with that by <ONE_LINE_METHOD>. No pitch deck, just a short outline of what we would look at first.
Want me to send the outline?
<SENDER_NAME>
If this is not relevant, tell me and I will not follow up.
TEMPLATE C: Funding or expansion
Subject: after the <EVENT>
<FIRST_NAME>,
Congratulations on <EVENT_IN_PLAIN_WORDS>. Growth like that usually puts pressure on <AREA>.
We help companies at that stage with <ONE_LINE_METHOD>. Is <AREA> on your list for the next quarter?
<SENDER_NAME>
If this is not relevant, tell me and I will not follow up.
If you would rather have an assistant fill the template, this is the only prompt in this guide. Everything else here is a table, checklist or template. Review every draft before it goes anywhere.
Asset 9: Signal-to-message prompt (the only prompt)Write a first-touch B2B email of at most 90 words from the template below. Use ONLY the facts supplied; invent nothing about the company or person. Mention the fact once, phrase the inference as a question, make one ask, make no claims about results or customers, and end with the opt-out line.
Sender: <SENDER_NAME>, <SENDER_COMPANY>
Offer: <OFFER_ONE_SENTENCE>
Verified fact: <FACT> (date: <DATE>)
Recipient: <FIRST_NAME>, <TITLE> at <COMPANY>
Template: <PASTE_TEMPLATE_FOR_THIS_SIGNAL>
Return 3 subject lines (under 6 words, lowercase) and the email body.
Step 6: Set the timing and the stop rules
Input: approved drafts. Do: send on the touch plan, and stop on any stop condition. Output: a sent log with a next-touch date for every row. Check: no row has more than three touches or a touch scheduled after a reply.
Signals decay. Send quickly, keep the cadence short and stop early.
Asset 10: Touch plan| Touch | When | Channel | Content | Stop if |
|---|---|---|---|---|
| 1 | Within 3 working days of verifying the signal | Signal template (Asset 8) | Reply, unsubscribe, bounce | |
| 2 | Day 3 to 4 after touch 1 | LinkedIn note or connection with a short message | Same idea, shorter | Any reply, or the signal is no longer current |
| 3 | Day 7 to 10 | One new, useful thing: a short checklist or an observation relevant to the signal | Any reply | |
| End | Day 14 | None | Mark "no response", set a re-check date of 60 to 90 days | Always |
If a second, independent signal appears after the sequence ends, you have a new reason to write. Start a fresh, shorter sequence that mentions only the new fact. Do not restart the old one.
Step 7: Handle replies and keep a suppression list
Input: every reply. Do: label it with the table below within a working day. Output: a label, an action and an owner for every reply. Check: test the table on ten invented replies, including a rude one that says stop.
Asset 11: Reply rule table| If the reply... | Label | Action | Owner |
|---|---|---|---|
| Asks to stop, remove or not be contacted, in any tone | Unsubscribe | Add to the suppression list for every campaign today and confirm in one line | Whoever sees it first |
| Says it is not relevant or not a fit | Not relevant | Suppress and close the row | Reply owner |
| Asks a question, wants detail or agrees to talk | Interested | Hand to the named person with the signal fact and the thread | Named person, same working day |
| Says the timing is wrong | Not now | Set a re-check date and stop touches | Reply owner |
| Says someone else owns it | Wrong person | Send one polite reply asking who owns it, then close the row whatever the answer | Reply owner |
| Is an automatic absence message | Out of office | Pause and resume after the return date | Sender tool or reply owner |
| Is unclear, mixed or emotional | Unsure | A person reads it and decides; never auto-reply | Named person |
Unsubscribe and "not relevant" go straight to the suppression list that every campaign checks before drafting anything. If a reply fits two rows, take the more cautious one. A person owns anything labelled interested or unsure.
Go live: pilot, review mode, then widen
- Pilot (week 1): run the full chain on 10 to 20 companies. Read every row and approve every draft by hand.
- Review mode (weeks 2 to 3): run a fresh batch, check only the drops, the queue and a sample of the drafts, and compare the KPI table with your first baseline.
- Widen (week 4 onward): raise volume slowly, keep three signals for four weeks before adding a fourth, and keep approvals in place until the complaint and bounce numbers stay low.
A worked example (illustrative)
This is a fictional example with invented names, shown to make the flow concrete. No real company or result is described.
Setup. Priya runs Northwind Analytics, a small consultancy that builds automated reporting for logistics firms. Her offer: "We replace manual weekly performance reports with automated dashboards in six weeks." Her ideal customer: logistics companies with 50 to 300 staff in the Gulf. Her three chosen signals: a job post for a reporting or data analyst, a new Head of Operations, a new depot or route announcement.
Steps 1 to 3. Hypothesis: "When a logistics company posts a job for a reporting analyst, it probably spends analyst hours building reports by hand, which we solve by automating the weekly reports." False-positive screen: the job description must mention reporting, spreadsheets or dashboards, because the role may be a pure backfill. Log row: Harbourline Freight (fictional), careers page link, signal date 2026-11-03. Verification on 2026-11-09: the exact quote "build and maintain weekly operations performance reports in spreadsheets" is on the page, the date is six days old and the post is live, and the company matches. Verdict: keep.
Step 4 output (score). Fit 3 (logistics, 120 staff, in region), owner of problem 2 (Head of Operations is the contact), signal 3 (verified, six days old), signal-to-offer link 2 (strong). Total 10, so: send now.
Step 5 output (draft). Template A with ROLE = reporting analyst and PROBLEM = weekly reports that eat analyst time, sent to Karim at Harbourline under Priya's name. Notice what is missing: no flattery, no "I came across your company", no statistic and no claim of past results. It names one fact, asks one question and makes it easy to say no.
Steps 6 and 7 output. Touch 1 goes on 9 November. The reply, if any, is labelled with the reply rule table and routed to Priya. If the job post disappears before touch 2, touch 2 is cancelled because the signal is no longer current. That is the point of the freshness check.
Not spamming: the rules we hold ourselves to
- One verified reason per message. If you cannot state the reason in one sentence, do not send.
- Three touches maximum per signal, then stop.
- Cap daily volume well below what your tool allows, and grow it slowly.
- Honour opt-outs immediately and permanently across every campaign.
- No private information. Use only professional, public facts. If a fact would feel invasive if the reader saw how you found it, leave it out.
On email deliverability, Google's sender guidelines require all senders to meet the sender guidelines when emailing personal Gmail accounts, and set stricter requirements for bulk senders (those sending close to 5,000 or more messages a day to personal Gmail accounts): authenticated domains with SPF, DKIM and DMARC, spam rates kept below 0.30 percent in Postmaster Tools (with a recommendation to stay below 0.10 percent), and one-click unsubscribe for marketing messages. A signal-based programme sends far below that volume, but the principles are the right ones: authenticate your domain, keep complaints low and make opting out easy. (Sources: Google, "Email sender guidelines" and the sender guidelines FAQ, accessed 9 November 2026.)
Laws on cold email and outreach differ by country, and by whether the recipient is a person or a company. This is not legal advice. Check the rules for every market you contact, and ask a qualified adviser if you are unsure.
Decision rules
If the signal is older than its window, drop it. If fit is low, drop it. If fit is high but the signal is weak, queue it. If two independent signals appear, move it to the front. If the reader replies "not relevant", suppress and close. Track three signals for four weeks before adding more, and use one short playbook per signal rather than one template for all.
Troubleshooting
| Problem | Likely cause | Fix |
|---|---|---|
| Most rows fail verification | Detection is noisy, or the hypothesis has no screen for false positives | Add the screen from Step 1; collect from the company's own pages first |
| Almost nothing reaches "send now" | Fit rubric too tight, or the watch list is wrong | Review 10 dropped rows by hand and fix the list before the rubric |
| Replies say "how did you know that?" | The fact was not obviously public or professional | Use only facts the company itself published; cut anything that feels like surveillance |
| Good signals, but few replies | The ask is heavy, or one template is used for every signal | Ask for 15 minutes or permission to send two lines; write one short playbook per signal |
| Bounces or complaints rise | Unverified emails, too much volume, or no opt-out line | Verify emails, cut volume, add the opt-out line and check domain authentication |
| A person got a touch after saying stop | Suppression list not shared across tools | One suppression list checked before every send, in every campaign |
| Messages read as automation ("saw you raised money") | The signal is the whole message | Link the fact to one problem and ask one question |
| Interested replies go cold | No owner or deadline | Name a person and answer the same working day |
What to measure and how to read it
We deliberately give no target numbers here, because benchmarks vary widely by market, deal size and list quality. Set your own baseline in the first two weeks and compare each change against it.
Asset 12: KPI table| Metric | How to calculate | How to read it | If it looks wrong |
|---|---|---|---|
| Verified signal rate | Verified signals divided by logged signals | Low means your detection is noisy | Tighten how you collect signals; add the screen from Step 1 |
| Send rate | "Send now" rows divided by verified rows | Very low means your fit rubric is too tight or your watch list is wrong | Review 10 dropped rows by hand |
| Reply rate by signal type | Replies divided by delivered first touches, per signal | Compare signals with each other, not with outside benchmarks | Stop the weakest signal and rewrite its template |
| Positive reply rate | Interested replies divided by delivered first touches | The number that matters most | Review the angle and the ask |
| Time from signal to send | Hours between signal date and first touch | Shorter is better; signals decay | Automate the queue; reduce approval lag |
| Opt-out and complaint rate | Unsubscribes and complaints divided by delivered | Your early warning for spam problems | Cut volume, retarget, review the copy |
| Meetings booked per 100 sends | Calls divided by sends, times 100 | Your actual outcome measure | Check reply handling speed first |
Pre-launch QA checklist
- Each of my three signals has a worksheet row with a two-minute verification method.
- Every pilot row has a source link, a date and a verdict, and 10 source links I opened by hand matched the log.
- The freshness window is enforced and a stale test row was dropped correctly.
- The rubric ranks a strong-signal poor-fit company as "drop".
- I read 10 drafts aloud: none mention private information, and none contain a statistic, result or customer claim I cannot prove.
- Sending domain is authenticated and is not my main domain; every email has an opt-out line.
- The suppression list is connected to every campaign, and a named person owns replies with a response-time target.
Your first week, day by day
- Day 1: Write your offer and ideal customer paragraphs. Fill the Asset 5 worksheet and choose three signals.
- Day 2: Build the watch list of 50 companies and the log columns. Collect signals for 20 of them.
- Day 3: Run the Asset 6 checklist on all 20 and open 10 source links by hand.
- Day 4: Score with Asset 7, then draft the "send now" rows from the Asset 8 templates (or the Asset 9 prompt) and edit by hand.
- Day 5: Authenticate the sending domain, connect the suppression list and send touch 1 to 10 to 20 approved rows.
- Day 6: Set up the Asset 11 reply table, the handover rule and the response-time target.
- Day 7: Fill in the KPI table, decide which signal to keep, rewrite or drop, and size next week's batch from the cost you saw.
Frequently asked questions
What is signal-based outbound?
Signal-based outbound is a way of prospecting where you contact a company because something observable has changed, such as a new hire, a job post, a funding round or a tool switch. The signal supplies the timing and the opening line, while your ideal customer profile still decides who is worth contacting.
Does signal-based outbound work better than list-based outbound?
It depends on your market and execution, and we do not quote a benchmark because we have not found one we can source. The logic is sound: a timely, relevant message gives the reader a reason to answer. Measure reply and meeting rates per signal type against your own baseline before drawing conclusions.
Which buying signals should I start with?
Start with three that map directly to the problem you solve. Common first choices are a job post for a role your offer could replace or support, a new senior hire in the buyer's function, and an expansion or funding announcement. Write a one-sentence hypothesis for each and drop any you cannot complete.
How fresh does a signal need to be?
Fresh enough that the reader still cares. Hiring signals are only useful while the post is live, new-leader signals work best in the first 90 days of the role, and funding or expansion signals fade within weeks. Set a freshness window per signal type and drop anything outside it.
How do I use signals without sounding like surveillance?
Reference one public, professional fact, state it plainly, and phrase your inference as a question. Never mention personal information or anything the reader would find invasive if they saw how you found it. Keep the first email short, make one low-effort ask and include a clear way to opt out.
Which tool should I use to run signal-based outbound, Claude Code or something else?
Start with a spreadsheet and an AI assistant to prove the process. Move to Claude Code if you are comfortable with files and a terminal and want verify, score and draft in one repeatable chain, or to a workflow engine such as Make or n8n for visual app-to-app flows. Keep a person approving every send, whichever you choose.
Sources and further reading
- Anthropic, "Building effective agents" (workflows versus agents, routing, starting with the simplest solution). Accessed 9 November 2026.
- Anthropic, "Claude Code overview" (surfaces, MCP, CLAUDE.md, skills and scheduled tasks). Accessed 9 November 2026.
- Google, "Email sender guidelines" (authentication, spam rate and one-click unsubscribe for bulk senders). Accessed 9 November 2026.
- Google, "Email sender guidelines FAQ" (definition of a bulk sender; all senders should follow the guidelines). Accessed 9 November 2026.
- SFC AI Stack page (tool descriptions; some links are referral links).
- SFC internal enrichment SOP and signal playbook structure (methods adapted and anonymised).
What to do next
Do Day 1 today: pick your three signals and write the hypothesis sentences. If you would rather have the system built and running in your own accounts, we build AI outbound agents that watch signals, verify them, score fit and draft outreach for your approval, and you can book a short call to talk through your pipeline. When replies start arriving, our guide to the AI appointment setter covers turning them into booked calls.
