Monday morning, Luke asked the community a question a lot of you have been circling.

He wanted to build a territory plan in Claude. Not a summary, not a first draft of an email. A real one: accounts surfaced by intent, account research done for him, and a dynamic heat map across a defined patch. He listed his stack so people could answer properly. Common Room, ZoomInfo, Sales Navigator.

What he asked for was a prompt.

What he got back was better, and it is the reason this issue exists.

There is no prompt that saves you

Joe went first, and he refused to hand over a prompt. His point was that the prompt is the last thing you should be worrying about, because the quality of what comes out is decided by three things you sort out beforehand:

• How your back end data is structured for mapping. If your account records do not cleanly resolve to a territory, no instruction fixes that. Claude will make something up that looks plausible.

• What first party data you actually have. Not what you have licences for. What is populated, current, and reachable.

• Which internal and external triggers you use for account tiering. If you have never written down what makes an account a tier one, you are asking a model to invent your strategy and then trusting it.

Get those three wrong and you get a confident, well-formatted, useless output. Which is exactly what most people are getting, and why so many leaders have quietly decided this stuff does not work.

It works. The prep is the job. The prompt is just the interface.

Joe then did something more useful than explaining. He built a version with dummy data, recorded his screen, wrote a getting started guide, and dropped the lot in the thread.

​ApolloNext 2026 is the defining event for go-to-market leaders. One day in San Francisco to put the AI hype aside and see what's actually working: first access to Apollo's biggest product launches, tactical sessions from teams driving real results, and a room full of the operators shaping modern GTM. September 30, SFJAZZ Center.

Get your ticket now using our exclusive SDR Leaders Global promo code HERE!

The second thing that stops people

Joey added the other half of it. He has built an agent for exactly this internally, and his warning was blunt: the model is not the hard part, the connectors are.

If Claude cannot reach your systems, you are exporting CSVs and pasting them into a chat window, then calling it automation. It will work once, for a demo, and never become part of how your team operates.

So get the connectors in place first. Which means talking to your security team early rather than presenting them with something you have already built.

What happened next

Tess asked to be looped in. Drew offered to join. Corbin asked to sit in and listen. Within a day the thread had stopped being a question and turned into a live build session, now set for Tuesday 18 August.

Nobody scheduled that. One person asked a specific question in public, with enough detail for people to answer it properly, and five leaders decided they wanted to be in the room.

Worth remembering next time you are tempted to work something out on your own.

The short version

1. Fix your account tiering logic and your data structure before you open Claude. That work is the deliverable.

2. Sort your connectors, and bring security in at the start.

3. Build with dummy data first, so you can judge the shape of the output before you trust it with real accounts.

4. Ask in public. Luke got a screen recording and a personalised build session because his question was specific enough to answer.

Blake Robinson joins as Head of Research and Intelligence

While we are on the subject of not guessing.

Blake Robinson is joining SDR Leaders Global as our Head of Research and Intelligence. Many of you have already seen his work. He built the framework behind the SDR Capacity Planning Benchmark.

Now he owns the whole thing:

• Creating and curating the data, benchmarks and reports this community has never had access to

• Leading the State of Sales Development, our flagship report (More to come soon!)

• Arming you with evidence rather than opinions when you are defending headcount and budget

Every one of us has had to justify this function to an exec team, a board, or a PE firm mid-diligence. Anecdotes lose those arguments.

Blake has asked for one thing from you, and he means it. He wants to hear your "it would be helpful if I had..." Whatever number you have been guessing at, or arguing about without evidence, tell him. It shapes what gets researched.

Reply to this email or find him in Slack.

Content is going live on the new site

sdrleaders.global is filling up.

We are starting to add content properly now: events, insights and benchmarks, with cross-region access so you can see what EMEA, USA and APAC are each working on rather than just your own patch. Salary and tech stack insights are in there already, and this is where Blake's research will live as it lands.

If you have not claimed your account yet, look for the email from noreply@notify.sdrleaders.global, or go to Account and use the password reset.

Have a look around. Tell us what is missing.