01 The modelNo retainerPaid on delivery

Performance fee per SQL.

An agency invoices for work done. We invoice per SQL, after it reaches your CRM, and the criteria are agreed in writing before the first message goes out.

An SQL isa company in an agreed segment, a buyer in an agreed role, our first approach, and a conversation that happened.

What you pay for

The SQL, not the activity.

A fixed fee per SQL, invoiced monthly against a statement that lists each one with its account, its buyer and the signal we opened on. You set the monthly cap and can move it for a future month. Nothing above it is charged without your written approval.

What starts it

One setup instalment.

It covers infrastructure, data and operator time: domains, mailboxes, the contact base, the segment work and the first sequences. It is capped before launch and it includes the first SQL. There is nothing monthly behind it.

What if nothing lands

No SQL, no invoice.

An empty month costs you the setup and nothing else, which puts the risk of a wrong segment on us rather than on your budget. That is why we check the size of the market before we sign and say so when it is too thin to work.

Both figures come out of your own deal economics and are fixed in writing on the first call, next to the definition of what we get paid for.

See how we start

02 Who it's for

Built for the pipeline problems you actually have.

Pain: not enough leads

Your pipeline runs on referrals and luck.

Outbound never worked for you: bought lists, blasted templates, silence. The problem was never outbound, it was writing to companies with no reason to write. We write only to accounts carrying a signal and the message names it. Meetings stop depending on who you happen to know.

Pain: leads, but wrong ones

The calendar fills. Nothing closes.

Wrong size, wrong stage, no budget: your closers burn days on meetings that were never going anywhere. Volume is not your problem, fit is. The segment and the buying role are written down before launch, and a conversation outside them is not an SQL and is not invoiced.

Pain: sales lives in two heads

You are the sales system.

You or one strong closer carry everything: prospecting at midnight, notes nowhere, replies triaged between calls. We take the top of the funnel off your desk: sourcing, the reason to write, the sequences and the reply handling, up to the point where somebody agrees to talk.

See yourself in one of these? The first call sets the segment, the SQL definition and the monthly cap.

Book the 30-minute call

03 How it worksLinkedIn, email, TelegramAI-native

From a market to an SQL.

Five steps, run as a cycle. What comes back decides the next cohort, so the fifth step is what rewrites the first.

The cycleBetter cohort → better reason to write → better conversation → more SQLs. Every cycle.

Market.

We draw the boundary of who can be sold to before we source anybody: the segments, the revenue and headcount floor, the geography, and the buying roles inside each account. The count is measured against real databases, not estimated, so you know how many accounts the model has to work with.

segments and thresholds, written down
buying roles per segment
reachable accounts, counted
exclusion list applied first

INyour marketOUTa first cohort, sized

Signal.

An account enters the queue when something public and dated happened to it: a raise, a hiring pattern, a change of platform, a new market, a person moving. No signal, no message, which is what keeps the volume low and the reply rate high.

funding, hiring and tech-change events
role changes and new mandates
public filings and registries
the date the signal appeared

INthe cohortOUTaccounts worth writing to this week

Message.

The context is read before anyone writes, and the message names the signal it was sent on. Micro-segments per signal, one sequence per prospect, across LinkedIn, email and Telegram. Sending runs on our own warmed domains, registered in your name, so your corporate domain is never used for cold outreach.

a reason per account, not a merge field
micro-segments per signal
LinkedIn, email and Telegram
deliverability engineered from day one

INan account and its signalOUTa message that names why

Conversation.

We work the replies ourselves: classify them, answer the questions, handle the objections and get to a time in the calendar. Every thread and every objection is recorded in your CRM as it happens, so your closers walk into a conversation that already has a history.

replies classified and answered
objections logged against the account
scheduling to a confirmed time
handover brief before the call

INrepliesOUTan agreed conversation

SQL.

The conversation happened, the company and the role match what was agreed, and the record lands in your CRM with its evidence attached. That record is what we invoice for, and what came back on the way to it decides which accounts open the next cohort.

the account, the buyer and the signal
the correspondence that produced it
the check against the written criteria
what the answers change for next week

INan agreed conversationOUTan SQL in your CRM

What lands in your CRM · and stays there

Account evidenceEvery signal, reply and objection pinned to the account.

Buyer mapWho actually sits in the buying committee, learned from real replies.

Message-market fitWhich reason to write converts, per segment and per objection.

Proof metricsReplies, conversations and SQLs, counted off your CRM.

Why the reason has to exist first The research problem behind personalization that gets answered.

05 Who we are

Engineers with a business background.

We are not another GTM agency and not lead generation. We have built commercial teams and launched B2B and B2C startups ourselves, so we count revenue and the business goal rather than leads.

Pavel Papin

Co-founder · GTM architect

"I ran this as a CCO whose own team's number depended on it, before I ran it for anybody else."

ex-CCO, Nebius Academy (NASDAQ: NBIS), closing large deals
built and ran the AI-assisted sales system his team used daily
builds the technical side himself, sourcing to CRM sync

Sofia Altman

Co-founder · GTM strategist

"Most AI GTM is theatre. I write sequences a buyer answers because the reason is real."

$1M+ closed in B2B revenue as a closer, ex-Head of Sales
grew an SDR team from 5 to 25 in Dubai
builds the signal-led sequences and works the replies

These two run your outreach themselves.

06 Against a retainer

An agency invoices for work done.

A retainer is agreed before anyone knows what the market answers, and it pays for volume of work. Ours is the same work with the invoice attached to the other end of it.

	Activity-based agency	Spice GTM
What is invoiced	A fixed retainer, monthly	A fee per SQL, after it reaches your CRM
When it is agreed	Before the market has answered anything	Before launch, against a written definition
What it pays for	Volume of work: lists built, messages sent, hours logged	Only the SQLs that meet the criteria
What the month's report says	Sequences live, invites accepted, open rates	Each SQL, with its account, buyer and signal
If the pipeline does not move	The retainer is invoiced anyway	No invoice

One small setup instalment covers infrastructure, data and operator time, and it is capped before launch. Everything after it moves with the result.

07 How we start

From agreed outcome to live outreach in 1-2 weeks.

The first cohort is defined before infrastructure goes live. Nothing is sent until the definition of an SQL is signed off, because that definition is what the invoice is checked against.

Stage 01 · define

Agree the outcomefree · one call

You leave knowing what gets paid for

The segment, the buying roles and the thresholds, written down
The definition of an SQL, and how each one is evidenced
Attribution, the exclusion list and the monthly cap you set
Both figures, out of your own deal economics, in writing
A written read of your market, yours to keep either way

Book the 30-minute call

Stage 02 · build

Build the motionsetup instalment · 1-2 weeks

Accounts, data, signals and messaging

The first cohort sourced and checked against the exclusion list
Domains and mailboxes registered in your name, warmed and paid for by us
Signals wired, and a reason to write per account
Sequences per micro-segment across LinkedIn, email and Telegram
CRM connected, so every record lands on your side from the first day

Stage 03 · run

Deliver and improveper SQL · cap is yours

We work the replies, you take the calls

Replies classified and answered, objections logged
Each SQL delivered with its evidence and counted against the cap
Lists and copy revised weekly against what actually came back
Call review, so the next cohort is chosen from evidence
A weekly working call and a statement at month end

Both figures are set on the first call and fixed in writing. If the market is too thin to carry the model, we say so before anyone signs.

08 Straight answers

The questions everyone asks. Answered without fog.

"You will send us junk just to get paid."

The definition is written before launch and every SQL arrives with its evidence: the account, the role, the signal and the correspondence. You have five business days to reject one by naming the condition it fails, and a rejected SQL is not invoiced. Volume against a cap you set is the opposite of a reason to pad.

"What if our market turns out to be empty?"

We size it before anyone signs, counted against real databases rather than estimated, and we tell you when it is too thin to carry the model. That risk sits on our side: an empty month costs you the setup and nothing more, and costs us the work.

"Why is the setup paid, if you work on results?"

Because it buys real things before a single reply exists: domains and mailboxes warmed for weeks, the contact base, the enrichment and verification, and the segment work. It is capped before launch and it includes the first SQL. Everything after it moves with the result.

"How is this different from a shop selling meetings at $100?"

Compare the definitions, not the numbers. A volume shop counts anyone who agreed to a call; here the company has to sit in the agreed segment, the person has to hold the agreed role, and the conversation has to have happened. A cheap meeting with no definition behind it is an argument in month two.

"Who writes the messages, and on whose infrastructure?"

We do, on our own operators and our own sending estate. Domains and mailboxes are registered in your name, we pay for and run them during the term, and on exit control passes to you. Your main corporate domain is never used for cold outreach.

"Outbound is dead."

Spray-and-pray outbound is dead, and deserved it. Writing only to accounts carrying a dated public signal, with the reason named in the message, is quietly booking the meetings your competitors think are impossible. The volume is a tenth of what a volume shop sends.

"We tried an agency. It flopped."

Probably a volume shop: scraped lists, one template, your domain burned. Different lists, different messaging discipline, and this time a month that produces nothing produces no invoice. If the same thing happens again, it happens at our expense.

"Our niche is too specific."

Specific niches are where signal-led targeting wins hardest: fewer accounts, richer signals, messages that could not have been sent to anyone else. Broad is what breaks outbound, not narrow. What matters is whether the niche holds enough accounts to carry a cap, and that is the first thing we count.

"We are too early. No sales team yet."

If a founder is selling, there is something to write to. Early engagements are smaller: one tight segment, founder-calibrated messaging, a low cap. You attend the calls and we do everything before them, which is the part that does not scale with a founder's evenings.

"What about our data?"

We source contacts ourselves and do not ask for your customer database. You give us the exclusion list, in any form you like including hashes or domains, and we check it before the first approach. Records land in your CRM under your access control, and you can revoke access in one click. One DPA covers the engagement.

Performance-based outbound.