Native Teams cold outbound pilot: strategy pack

A. Executive summary

Goal. Book demos (MQLs in HubSpot) from cold outbound. Our work ends when the calculator form creates a HubSpot deal and sales takes over.

Positioning. Know the full cost of your next hire abroad before you hire, then hire them through us in days. Every cold email starts with something useful: a cost estimate already filled in for the exact role and country the prospect is hiring for.

Who we target. Software, IT, consulting, marketing, fintech, staffing and e-learning companies. They have 11–250 employees (11–50 is the sweet spot), were founded in 2012 or later, and can be headquartered anywhere. They must be hiring in a country where Native Teams has an entity and they don't. We build campaigns around the hiring country, not the HQ, because hiring country predicts wins better REAL. There is no hiring into the US, and India is not a target segment.

Why this approach.

• Generic cold email is weak. Apollo cold sequences get a 1.05% reply rate and 0.68 meetings per 1k delivered REAL.
• Copy built around a trigger roughly triples replies. "Hiring in Spain" reached 3.2% REAL.
• Warm email gets about 16× more meetings than cold REAL.
• Once a deal exists, US/UK companies hiring into Germany win about half the time REAL.

Where we are today [REAL, live Instantly audit]:

• The Instantly instance is fresh: 0 campaigns, an empty blocklist, no webhooks and no lead labels.
• All 40 inboxes are warmed (score 100, warmup running since 20 Jan 2026) but all 40 have status = 2, which means "Paused". The enum is confirmed against developer.instantly.ai.
• No error codes are present, so it looks like a manual pause. Who paused them and why is OPEN.

The plan.

• Wave 0 is Germany and Portugal, with US/UK-HQ companies first, at 8–10 sends per inbox per day.
• We then expand by hiring country over 6 weeks.
• The Balkans and Philippines follow through Ella.
• We build each sequence once, with variables. Instantly sends on its own cadence, and people and agents watch the replies.

Base forecast [PLAN, ASSUMPTION-driven]:

• 6 weeks: about 12 demos, 9 deals and 1.4 closed-won from about 4,200 contacts.
• 90 days: about 40 demos and 4.7 closed-won. This includes lost-deal re-engagement, which still needs approval.

Honest framing. Six-week revenue will be small. The prize is a repeatable, measured motion by hiring country.

Asks for tomorrow: the decisions in section H, above all the inbox-pause cause, Wave 0 sign-off and the outbound booking link.

B. Initial plan: weeks 0–2

B1. Hardening the Instantly setup (week 0; every write needs Shyam's OK)

B2. Wave 0: Germany and Portugal (weeks 1–2)

• Lists (Prospector):
• Apollo search by job location = DE or PT, with ICP filters.
• Drop companies that already have a local office or more than 3 roles there.
• Run the HubSpot exclusion check.
• Pull job postings, then reveal 1–2 decision-makers per company.
• Verify every email, because bounces have historically been high.
• HQ order: US and UK/IE first (UK/IE→DE wins 51%, US→DE 47% once qualified REAL), then DACH and Benelux into Portugal.
• Volume PLAN: about 300 companies, 450 contacts, about 1,800 sends and about 1,050 Apollo credits. Spain moves to Wave 1 to match the Wave 0 = DE+PT brief.
• Capacity: 40 inboxes × 8–10 per day = 320–400 sends/day. Warmup stays on.

B3. Build the sequence once; Instantly runs the cadence

• There is one campaign per hiring country (no US hiring), plus one warm campaign.
• We write each sequence once, with variables: first_name, company, country, role, salary_band, calc_url (pre-built with UTMs) and personal_line.
• personal_line is the only per-contact text. Prospector drafts it in the CSV, and a human reviews it before upload.
• Cold sequence:
• D0: a question about the cost of the hire, with a pre-filled link.
• D3: a local employment fact in the same thread, no link.
• D7: a new thread on EOR vs contractor, with an Ask Ella link.
• D12: breakup email.
• A/B test on D0 only, first 300 contacts:
• A = pre-filled calculator link.
• B = no link, ask whether they'd like the estimate.
• The winner is picked on reply rate.
• Claims: only the allowed ones (entities in 95+ countries, 4.9 on G2 / #1 EOR, free 30-minute call). No €199 flat fee and no "€300 saving" claims.

B4. Approval gates (week 0–2)

Shyam approves each of these, one by one:

• resuming the inboxes
• each campaign's copy
• each day's upload list
• each launch
• every reply draft
• the blocklist load
• labels
• any webhook
• any HubSpot write

Coordinator keeps the queue in /workspace/pilot/approvals.md and summarises it in the 08:30 digest.

B5. Reply visibility from day 1

• The Instantly Unibox plus interest statuses are the system of record.
• The Replies agent triages every hour from 08:00 to 19:00 on weekdays.
• The daily digest at 08:30 Paris lists every reply, its tag and a draft response.
• Slack or HubSpot webhooks are optional and need Shyam's approval.

C. Midterm plan: weeks 3–6

• What we take from the A/B test: the D0 readout comes after the first 300 contacts. The winning variant becomes the default for every country, and new tests change one thing at a time (subject, then the personal_line style).
• Warm track (calculator downloaders who consented at the gate):
• D3: "How did the numbers look?"
• D10: what the calculator leaves out, plus the booking link.
• After that, the contact goes to the BDR list.
• Inbound drafts these emails; the warm campaign in Instantly sends them once approved.
• Inbox ramp rules: go from 8–10 to 12 to 15 per inbox only while bounce stays < 2%, there are no spam flags and warmup health holds. Any breach means Coordinator proposes a pause, and Shyam decides.
• Analyst feedback loop:
• Monday: the funnel by hiring country × HQ × step (delivered, replied, positive, link click, calculator download, demo, deal).
• Tuesday: Shyam approves the reallocation and the next wave's lists.
• Thursday: the copy readout.
• Friday: the compliance check.
• We re-forecast after week 2 using the pilot's own rates.
• Apollo budget: track pilot spend as deltas in budget.json, separate from the shared account. The pilot needs about 9.8k credits; the October cap is 100k.

D. Later plan: days 60–90+

• Balkans/Philippines track (Ella):
• Covers N. Macedonia, Serbia, Croatia, Bosnia, Montenegro, Albania and the Philippines.
• The CTA is Ella plus a free 30-minute call.
• Balkans win rates are 38–41% (Bosnia 51% on a small sample) REAL, but the median deal is small (€99–185).
• Needs Balkans BDR call capacity.
• Lost-deal re-engagement (optional; needs Shyam's approval as an exception to the past-deal exclusion):
• Scope: about 1,737 qualified lost deals from 2025–26 REAL.
• Send from a named AE mailbox, not the cold inboxes.
• At most 4 emails; exclude unsubscribes and recent competitor losses.
• Precedent: "Lost Deals 2025" got 17.3% reply and 15 meetings per 1k REAL.
• Standing approval thresholds (proposal):
• A campaign qualifies after 2 weeks with bounce < 2%, unsubscribe < 1%, 0 spam complaints and a positive-reply rate at or above plan.
• Shyam can then give standing approval for new prospects into that campaign with unchanged copy, up to a daily cap.
• Approved reply templates for standard cases (OOO, wrong person, unsubscribe acknowledgement) could also be pre-approved.
• Any breach revokes the standing approval automatically. The digest still lists everything.
• Scale rules:
• Move volume to the hiring countries and HQ pairs that produce demos, not just replies.
• Add inboxes only after the 40 run at 15/day cleanly.
• Never reuse unverified data.
• Keep hiring into the US off until the calculator and calendar support it.

E. How Instantly is run

Question: templates or bulk? Templates. We build the full 4-step sequence once per hiring country, with variables. Instantly enrolls the contacts and sends D0, D3, D7 and D12 on its own schedule, in each HQ's time-zone window. No one writes D3 or D7 by hand.

Watching replies does not pause the sequence. Only an approved pause action does.

Full detail is in instantly-phases.md.

F. Reply handling and visibility

Tags. Interest statuses (Instantly lt_interest_status, enum confirmed in the docs):

Custom labels to add: Meeting booked, Referral given, Unsubscribe.

Visibility:

1. Instantly Unibox (live).
2. Daily digest at 08:30 Paris: every reply, its tag, a draft response and the counts.
3. Optional: a Slack channel post and/or a Google Sheet log via webhook. Needs Shyam's approval.

Approvals. The Replies agent drafts. Coordinator packages the drafts. Shyam approves, until standing rules exist. Unsubscribes go to the blocklist and suppression list; whether that step can run automatically is a decision for Shyam (section H).

G. Agent roster and how they work together

Handoff chain: Prospector (verified lists) → Campaigns (variables CSV in the country campaign) → Shyam approves → Instantly sends on cadence → Replies triages → Coordinator packages the drafts → Shyam approves → interested prospect → calculator / booking link → Inbound: form creates a HubSpot deal → sales takes over. Each week the Analyst reports to Coordinator and Shyam, which closes the loop.

Daily rhythm (Paris):

Involvement by phase:

H. Decisions needed from Shyam before launch

1. Inbox pause: OK for Campaigns to read the Instantly audit log (GET only) to find out who paused the inboxes and why? After that, approve resuming them, or not.
2. Sender identity: keep the Cyprian Sarabosing signature with the London address on all 40 inboxes? Confirm that Cyprian agrees and that replies get monitored.
3. Wave 0: approve Germany and Portugal, about 300 companies, US/UK HQs first, at 8–10 per inbox per day (320–400/day), warmup on.
4. D0 A/B test: approve link vs no-link on the first 300 contacts.
5. Outbound booking link OPEN: a dedicated HubSpot meetings link (built by Indi via the SPECIALIST_LINKS override), or the shared inbound calendar plus a utm_source=instantly report.
6. Link tracking: only on if the UTM-redirect test passes.
7. Blocklist load: approve loading HubSpot customer and deal domains plus unsubscribes into the Instantly blocklist.
8. Automatic suppression: may unsubscribes and hard bounces go to the blocklist without a per-item approval? This is legally required, so the recommendation is yes.
9. Webhooks: replies to Slack and/or HubSpot, yes or no?
10. Compliance: legal sign-off on the GDPR legitimate-interest assessment, the German wording, and holding CA/AU HQs.
11. Lost-deal re-engagement: approve the exception, or defer it.
12. Standing approval thresholds: agree them now, or approve case by case.

I. Forecast

Inputs:

Outputs:

Volumes: ~2,800 companies, ~4,200 contacts, ~16,800 sends, ~9.8k Apollo credits PLAN. We re-forecast after week 2 on the pilot's own data.

J. Confidentiality

This pack is for the pilot team only: Shyam, the CEO and the pilot agents.

• Do not forward it or share links to it outside the team.
• If it is hosted on Dokploy, put it behind authentication or an unlisted, access-controlled URL, and only after Shyam approves the deploy.
• The data shown is aggregated. There is no personal prospect data in this pack.