You are an engineer or operator, not a salesperson. You have a product that works and a list of companies that should buy it, and you are spending 15 to 20 hours a week sending messages that mostly land in spam. You have probably tried an agency before and ended up with mis-qualified meetings and nothing you could keep when the contract ended. So when someone pitches a "3-month pilot," the fair question is: what actually gets built, and in what order?
This is the week-by-week version. No vendor benchmark gloss, just the real sequence we run on a fixed-scope 3-month outbound pilot, where the deliverable is an owned system that gets handed to you at day 90. Some weeks are unglamorous. The unglamorous weeks are the ones that decide whether the campaign works.
Weeks 1 to 3: Infrastructure and warmup (the part everyone skips)
Nothing goes out in the first few weeks, and that is the point. Most outbound fails before the first send because it runs on shared infrastructure or a freshly bought domain. We register dedicated sending domains, separate from your primary domain so a deliverability problem never touches your real email, and stand up 52 warmed mailboxes across Google, Microsoft, and Azure. SPF, DKIM, and DMARC get configured per domain, then the mailboxes warm gradually so the receiving servers learn to trust them.
This is the difference between roughly 60% inbox placement on shared infra and the 98.5% average placement we hold across accounts. Warmup cannot be rushed. A mailbox that sends 200 cold emails on day two gets flagged, and you spend the rest of the pilot in the spam folder. If you want to understand why this stage matters, our deliverability breakdown covers the infrastructure in detail, and you can run your draft copy through the free spam words checker while the mailboxes warm.
Weeks 2 to 4: Data, enrichment, and the target list
Running in parallel with warmup: building the list. This is where most bought-data outbound quietly dies. Apollo exports decay fast, and a list with 15% bad emails will wreck your bounce rate and your domain reputation in one send. We build the target list in Clay with waterfall enrichment, which means checking a contact across multiple data providers in sequence until we get a verified result, rather than trusting one stale source.
The output is a list with verified emails that holds bounce rates between 0.15% and 0.9%, not the double-digit bounces that get domains blocklisted. For one client we built a proprietary 66,000-school database that did not exist in any off-the-shelf tool. That is the kind of thing Clay enrichment makes possible: you are not renting a list, you are building an asset. We have run this across more than 1,800 production Clay tables and 950,000+ enriched contacts, so the patterns that cause bounces are known and engineered out before send.
Weeks 4 to 6: Signals, copy, and first sends
A verified list is necessary but not sufficient. The reason to message someone is timing. We layer signal-based triggers onto the list: recent funding, active hiring, tech-stack changes, and job changes among your buyers. A founder who just raised a round or just posted three roles in your category is a different prospect than a name on a static list.
Copy gets built per segment, with AI personalization using Perplexity for research and Claude for drafting, so the opening line references something real rather than a generic merge field. Then the first campaigns go live: cold email plus LinkedIn through HeyReach on real profiles only, never fake accounts that get banned. Early sends are deliberately small so we can read reply data before scaling volume. The goal share we target is 25 to 30% positive replies, and the way you get there is matching signal-based timing to relevant copy, not blasting volume.
Weeks 6 to 12: Scale, automate, and reply handling
Once early campaigns show clean deliverability and real replies, volume scales across the warmed mailboxes. This is also when the system stops eating your time. Self-hosted n8n automation handles the routing: replies get classified, positive ones surface to you, and the rest are handled or sequenced automatically with AI reply handling. Everything syncs to your CRM, whether that is HubSpot, Salesforce, Pipedrive, Attio, or Monday, so a reply becomes a tracked opportunity without manual copy-paste.
The practical result is the founder-time track: your involvement drops from 15 to 20 hours a week to roughly one hour reviewing qualified conversations. The campaign keeps running on triggers, not on you remembering to send. For context on the build, our sales automation overview walks through how the n8n layer fits together with the CRM and the sending stack.
Day 90: What you own when it ends
The pilot ends and the system stays with you. The domains, the 52 mailboxes, the Clay tables, the n8n workflows, the CRM integration, and the campaign logic: all of it is yours. This is the part that separates a pilot from a pay-per-meeting arrangement, where you own nothing and the leads stop the day you stop paying.
This approach is what produced 78,000 emails and $300,000+ CAD in pipeline at a 37% positive reply rate for one retail-tech client, 177 interested wine importers reached on trade-fair timing for another, and 143 interested physicians from a single insurance campaign. You can see the full numbers on the case studies hub. If you are weighing this against hiring an SDR, the in-house comparison lays out the cost and ramp difference. The point of the pilot is not a batch of meetings. It is a working outbound system you control after we leave.
Questions, answered.
Why does the pilot take three weeks before sending any emails?
What do I actually keep when the 3-month pilot ends?
How much of my time does this take once it is running?
Want this built and run for you?
LongRun builds the outbound system, runs it, and hands it over at day 90. Book a strategy call to scope yours.