How to turn a market event into territory-wide outbound

Yesterday, Deel's Mozambique director messaged me about one account in my territory.

By the end of the morning, I'd mapped the accounts in my book with exposure to a $20 billion project, identified the right person inside each one, built segmented outbound sequences, and staged everything in Amplemarket.

Ten minutes of actual work inside Amplemarket. Everything before that was reasoning, done in Claude, connected directly to my Amplemarket account through its MCP integration.

This post is that build, prompt by prompt.

If you sell into any market where local news creates sudden, category-wide demand, which, if you're in B2B sales, is more often than you'd think, this is the exact workflow.

When is a news event actually a sales trigger?

A news event is a sales trigger when it forces a whole category of companies to solve the same problem at the same time. It is not a trigger when it only changes the situation of a single company, however dramatic the headline.

Our director pinged me: one of my accounts is tied to a $20 billion energy project restarting after four years. Worth a look.

Most news like this gets a thumbs-up reaction and dies in a Slack thread. Interesting, not actionable.

I almost did the same thing.

Then I asked a different question: If this one account is exposed, how many others in my territory are sitting on the same news without me knowing it?

That question is the whole method, and it's also why this workflow needed Amplemarket rather than a spreadsheet. I don't just need to know that one account is exposed, I need to search my entire owned territory, across every industry tag, in the time it takes to ask.

Here's the test I use before building anything: Does this event force an entire category of companies to solve the same problem right now, not just one company?

If yes, it's a business case. If no, it's a nice-to-know.

This event passed that test immediately.

The workflow, in five steps:

1. Qualify the event: does it force a whole category of companies to solve the same problem at once?

2. Research the facts in Claude with web search, flagging anything a prospect could correct

3. Turn the research into a segmented sequence structure with dynamic fields

4. Sweep your territory through the Amplemarket MCP, scoring each account's exposure using enrichment data

5. Identify the right persona per segment, then build one sequence per segment in Amplemarket and review before activating

What made this event a business case?

TotalEnergies and the Mozambican government jointly announced the full restart of the Mozambique LNG project on January 29, 2026, at Afungi. That followed the consortium's decision on November 7, 2025, to lift the force majeure it had declared in April 2021 after security incidents in Cabo Delgado forced a full evacuation of the site. More than four years of suspension.

The numbers that made this a business case rather than a headline:

• Over 4,000 workers mobilized at the announcement, more than 3,000 of them Mozambican nationals, rising past 6,000 by the end of April 2026
• The project 42% complete at the end of March 2026, with first LNG targeted for 2029
• Up to 7,000 direct jobs expected during construction
• More than $4 billion in contracts earmarked for local companies
• A revised budget of roughly $20.5 billion

And the detail that ended up carrying the whole first email. On July 22, 2026, MSC launched a dedicated Afungi feeder service from Nacala and Maputo, carrying construction materials, machinery, industrial equipment and spare parts. Six days before I ran this build.

A shipping line doesn't open a dedicated route for a project that's winding down.

None of that tells you who to email. What it tells you is the shape of the problem.

Every EPC (engineering, procurement and construction) contractor, camp and catering operator, logistics provider, and inspection firm touching that project is about to mobilize crews into a market where almost none of them hold a local entity.

That isn't a TotalEnergies problem. It's every Tier 1, 2, and 3 supplier's problem, at the same time, for the next three years.

Not long before this news broke, we'd launched Deel Field Services, built specifically for organizations moving crews into medium-to-high complexity, on-site environments across markets like Mozambique, where owning the local entity rather than routing through a regional broker is the actual differentiator.

I'd had the product. I hadn't had the reason to lead with it yet.

This was the reason.

How do you research the event before building anything?

First move: get the facts straight before building anything on top of them.

I opened a fresh Claude conversation with web search on and asked for research.

Research the TotalEnergies Mozambique LNG project restart. I'm a Mid-Market AE at Deel selling workforce infrastructure into the companies that supply this project. I need commercial ammunition, not a summary.

Cover:
1. Restart timeline and current status: force majeure lifted, restart announced, completion %, first LNG target
2. Workforce numbers: how many mobilised now, projected peak, local vs expat split, local content quotas
3. Contract value flowing to suppliers and which tiers are tendering right now
4. The supply chain map: which categories of company get work on a project like this. EPC, subsea, camp and catering, logistics, drilling services, inspection and certification, equipment rental, security. Name actual companies where they're publicly confirmed.
5. What's happening in parallel in the Rovuma Basin, Eni Coral and ExxonMobil Area 4, that compounds the labour demand
6. Operating conditions in Cabo Delgado that make staffing hard: security, rotational schedules, remote site, entity requirements

For each fact, note whether it's confirmed and by whom. Flag anything a prospect could correct me on. End with the 6 facts most likely to make a supply-chain executive stop and read.

This step matters more than it looks like it should. The email that gets a reply from a Project Director isn't the one with the best pitch. It's the one that proves you actually understand what just happened to their world.

How do you turn the research into a sequence structure?

I pasted the research back into the same conversation and connected Deel's internal knowledge base, so Claude could pull real Deel Field Services positioning rather than generic feature language.

Using the Mozambique research above and the Deel knowledge base, build me the structure for a five-email outbound sequence. This is a business-event sequence. The news is the reason I'm reaching out, and it carries the whole thing.

Product: Deel Field Services. Pull the actual positioning from the knowledge base, plus any customer examples in oil & gas, mining, construction or heavy logistics.

Rules for every email:
- Subject line references the project or the business event, never Deel
- First sentence is about their world, not mine
- 4-5 lines maximum
- Each email introduces one new piece of information the previous one
didn't have
- No feature lists, no "just checking in," no meeting ask before email 3

Arc across the five:
1. The event and why it lands on them specifically
2. The operational problem it creates: crews, rotations, entity,local content
3. How another company solved it, named or anonymised, from the knowledge base
4. The cost of the default path: regional brokers, no spend visibility
5. Direct, short, easy to decline

For each email give me: subject, body, the specific research fact it uses, and which dynamic fields need filling per company. Mark every personalization point as {{field_name}}.

The output was a five-email arc where every email earns the next one. The dynamic fields Claude flagged became the exact variables I'd need Amplemarket to fill per lead: {{company_project_link}}, {{persona_pain}}, {{proof_point}}.

Worth flagging one thing here, because it's the kind of gap that only shows up when you actually check. Deel's case study library is deep on technology and SMB, and thin on heavy industry.

There was no oil and gas customer story to pull. So {{proof_point}} ended up carrying a mechanism statement rather than a named reference, something like": "operators with existing local entities rarely need another one, they need a way to put rotators on compliant contracts without reopening their HR structure."

‍Less powerful than a logo, but still true, which matters more.

How do you find which accounts in your territory are exposed?

You query your book by industry category rather than by company name, filtered to accounts you own, then enrich and score each result on how plausibly the event touches it.

This is where the workflow moves from "one account" to "how many," and it's the step that doesn't work without Amplemarket's MCP integration.

I connected it directly inside my Claude conversation and asked it to query my book the same way I'd query it manually, except across every account I own rather than the handful I happened to remember.

Using the Amplemarket MCP, pull the accounts I own that plausibly have exposure to the TotalEnergies Mozambique LNG project supply chain.

Start by calling get_industries so you're filtering on real values, then list_accounts filtered to me as owner.

Target categories:
- Oil & gas services and equipment
- Engineering, procurement and construction
- Marine, offshore and subsea
- Heavy logistics, freight forwarding, port operations
- Facilities management, catering, camp and accommodation services
- Inspection, certification, testing, HSE services
- Industrial and heavy equipment rental
- Specialist staffing and manpower for industrial sectors

Use domains rather than company names wherever you have them. Name matching is unreliable.

Then for each account returned, enrich it and tell me: does it have existing Africa or Mozambique operations, is it plausibly a Tier 1, 2 or 3 supplier here, and what's the specific reason it would care. Score each 1-3 on likelihood of exposure and sort by score.

Output as a table: account, domain, industry, tier, Africa presence, exposure score, one-line reason. Flag any account where the connection is a stretch. I'd rather cut than send something wrong.

Two Amplemarket capabilities do the real work in this step:

• ‍Company enrichment pulls firmographic depth Claude can reason over instead of guess at
• The account ownership filter means this search only ever touches my book, not the whole database. 

That combination is what turns "I wonder if I have other exposed accounts" into an actual answer in under a minute.

The scoring is also where the exercise earns its keep, because it cuts as much as it qualifies. Two things surfaced that would have wrecked the copy if I hadn't checked. Both of my strongest accounts already held local entities in the project country, one of them two.

So the headline pitch, employ people without setting up a local entity, was wrong for both of them. The angle had to shift to rotators and the compliance layer underneath an entity they already ran.

Who do you actually contact inside an exposed account?

A supply-chain exposure score is useless without a name attached to it. Same conversation, same MCP connection.

For each account scoring 2 or 3 above, use the Amplemarket MCP to identify who I should reach out to. Call get_job_functions first.

Persona logic by segment. Apply the right one per account, don't use one list for everything:

EPC and construction: Project Director, Mobilisation Manager, Head of Site Operations first. HR Director Africa second.

Camps, catering, facilities: HR Director and Finance Director first, high headcount, thin margin, labour spend is the pain. Ops Director second.

Logistics and marine: Country Manager or Regional Director first, then Head of HR EMEA/Africa.

Technical services, inspection, drilling: Head of Mobilisation or Resourcing first, then HSE Director.

Across all of them, add the CFO or Finance Director as a secondary thread on accounts where headcount exposure is large.

2-3 people per account. For each: name, title, seniority, LinkedIn, which segment logic applied, and why them over the alternative. Skip anyone we've already contacted in the last 90 days.

Then group all accounts into segments: EPC/construction, camps & logistics, technical services.

The "skip anyone contacted in the last 90 days" instruction works because Amplemarket already holds that engagement history across every channel in the sequence. Every person returned came back with a recent-contact flag attached, so suppression happened at the point of search rather than as a manual pass afterwards.

The segment logic earned itself immediately. On one technical services account, "Country Manager or Regional Director first" surfaced a country manager physically based in the project country.

Out of everyone the search returned, he was the only person sitting in the country where the event was actually happening. A generic C-suite list would have sent me to Paris.

Choosing the right two or three people per account is its own discipline; the buying committee mapping guide goes deeper on it.

How do you structure sequences when a whole segment is exposed?

This is the step I almost got wrong. My first instinct was one sequence per account. That doesn't scale, and it isn't actually more personalized; it's just more objects to maintain in Amplemarket.

The real personalization lives in per-lead dynamic fields, not in duplicated sequences.

So: one sequence per segment, built and staged directly through Amplemarket's sequence and stage tools, with per-lead fields filling in the specifics.

Build the sequences in Amplemarket, one per segment, using the structure from earlier. Same five-email arc, but the industry framing shifts per segment. An EPC mobilisation lead and a catering CFO don't
have the same problem.

For each sequence:
- Adapt each email to the segment's actual operational pain
- Keep the 4-5 line limit
- Keep the news as the opener
- Add a LinkedIn touch after email 2 and after email 4, referencing the email subject
- Add a call task on the same beats

Per-lead personalization via dynamic fields, not rewritten copy: {{company_project_link}}, their specific connection to the project or the basin. {{persona_pain}}, what this title owns that the restart disrupts. {{proof_point}}, the closest customer example from the knowledge base to their segment.

Then: create_sequence for each segment, append_stage_to_sequence for each email, LinkedIn touch and call, add_leads_to_sequence with the dynamic field values populated per person.

Show me everything in full before creating anything. Once I approve, create as drafts. Do not activate.

On the calls: I put phone_call stages after emails 2 and 4, on the same beats as the social touches, so the call always references a subject line the prospect has already seen. Calls arrive as manual tasks rather than automated sends, and the notes field accepts the same Liquid fields as the email bodies, so {{persona_pain}} populates the script per person rather than just the copy.

One constraint worth planning around. Stages append to the end of a linear draft, and a stage type can't be changed once created. If you want calls interleaved rather than bolted onto the end, decide before you build. I didn't, and interleaving properly meant a rebuild.

If you want the field-formatting detail that stops your paragraph breaks collapsing in the prospect's inbox, the ABM outreach guide covers it, along with the account-level version of this workflow.

Prompts like the ones above are also worth saving; the GTM Skills Library has ready-made versions you can run without rebuilding from scratch.

I reviewed everything in full before anything went live in Amplemarket. Every subject line, every body, every dynamic field value per lead. That review step isn't optional. The entire premise of a business-event sequence is that the facts are right.

Amplemarket makes it fast to build, but the accuracy check still has to be human.

Pressing play

Once the sequences were approved, I opened the Amplemarket interface for the first time in this entire process.

Not to build anything: everything was already there, staged, personalized, ready.
Just to activate.

‍

That's the actual shift. The reasoning, research, structuring, mapping, writing, and personalizing happened entirely inside a Claude conversation connected to Amplemarket through MCP.

Amplemarket became the execution layer: the place holding my account ownership, my engagement history, my contact data, and now live sequences. Not a tool I had to manually operate at every step, but the system of record the whole workflow ran on top of.

What's next

Sequences are live. In September, once the full arc has run across all three segments, I'll share the results here: replies, meetings booked, and which segment responded fastest. My guess is EPC, because mobilization pain is the most immediate and visible right now. I could be wrong, and that post will say so either way.

The bigger takeaway isn't the tools, though the combination is why this took a morning instead of a week. It's the filter. News is only a business case if it creates the same problem for a category of company, at the same time. Most news doesn't clear that bar.

When it does, and when your CRM data lives inside a platform an AI agent can actually query, the build from ping to staged sequence is a morning's work rather than a project.

If you want to see how far this goes, I also wrote up how I built an AI lead sourcing agent with Claude and the Amplemarket MCP.

This workflow uses Amplemarket's MCP integration, available to any Amplemarket customer connecting Claude or another MCP-compatible AI assistant.

How this was researched

Every project figure in this post comes from primary sources: TotalEnergies' restart announcement and Q1 2026 earnings call, and MSC's press release on the Afungi Shuttle, linked from the section where they appear.

The account and contact examples are real results from my territory, anonymized. The exposure table shows actual scores with account names removed.