Your pipeline shouldn't live inside client folders: one CRM, stage as the structure, brand as a label

I found the problem while reading a filter, not a report. An automated follow-up engine I run for one of my brands was, on paper, allowed to email a contact that belonged to a different brand of mine. Not because the engine was badly built, but because nothing in the contact record said which brand owned the person. The engine asked "has this contact gone quiet?" and that was the only question it knew how to ask.

Nobody got the wrong email. But the only thing standing between that contact and a follow-up from the wrong sender was the fact that the lists happened not to overlap yet.

Leads were filed by where they came from

Here is how it got that way, and I suspect it is extremely common. Each piece of work I do has a project: a client, a site, a practice. When a lead arrived through one of them, it went into that project. A prospect who found the lead-gen site sat in the lead-gen project. A prospect who came out of my own practice sat next to the delivery work for that practice.

It feels tidy. Everything about a thing lives with that thing. The trouble is that it files business development on the same axis as delivery, and those two are not the same axis at all. Delivery work is organised by who you are doing it for. Pipeline is organised by how close someone is to deciding. Put them in one tree and you cannot answer the only question that matters on a Monday morning: who is waiting on me right now, across everything?

I could tell you how a client project was going. I could not tell you how many live conversations I had. The number existed, but it was spread across five folders and a task list, and it changed shape depending on which folder I opened first.

The simultaneous visibility of these project materials helps us identify patterns and encourages creative synthesis to occur much more readily than when these resources are hidden away in file folders, notebooks, or PowerPoint decks.

That is Tim Brown in Change by Design, writing about design studios rather than sales pipelines, and it is exactly the failure. Nothing was lost. Everything was hidden away in file folders.

Stage is the structure, brand is a label

The fix was one decision, and it took about ten minutes to make and an afternoon to apply. Every lead, prospect, inbound enquiry, partner conversation and outreach reply moves to one place. That place is structured by stage: active, waiting on them, nurture, partners and network, admin. The brand becomes a label on the record, not a folder around it.

Stage-as-structure is not my invention; it is what every serious CRM already assumes. HubSpot's lifecycle stages and Salesforce's separation of leads from accounts and opportunities both exist because a record's position in the buying process is the thing that changes, and the thing you act on. What I had done was invent a second, competing structure on top of it — and the second one won, because it matched the shape of my file system.

Making brand a label rather than a container changes what the automation can see. The follow-up engine now reads the brand field and refuses anything that is not its own. The record-level check replaced a convention I was holding in my head. That is the whole point of writing a fact down in a field: it survives the week I am not paying attention.

Why this matters more with automation than it did without

When a human runs the pipeline, a mis-filed lead is a small annoyance. You notice the mismatch, you fix it in passing, and nobody outside the company ever sees it. When a scheduled routine runs the pipeline, a mis-filed lead is an email from the wrong sender with the wrong signature at the wrong moment — and the recipient is the one who finds out.

Gartner's 2025 buyer survey found that 73% of B2B buyers actively avoid suppliers who send irrelevant outreach. Irrelevant is not only the wrong offer. It is also the right offer from a brand they never spoke to. That is a data-model problem dressed up as a messaging problem, and no amount of better copy fixes it.

This is the part I keep coming back to in my own work. The leverage in AI tooling is not that it writes a better email. It is that it will faithfully execute whatever your records tell it, at a volume you would never reach by hand. Which means the records stop being admin and start being the product. I have written before about qualifying inbound before the sales team acts and about a lead that went to a company struck off the register two years earlier — both are the same lesson from a different angle. The automation is only ever as honest as the field it reads.

What I would check in any business doing this

Open the place your leads live. If the first level of the tree is the name of a product, a site, a brand or a client, business development is filed on the delivery axis and you have the same problem I had. Ask who is waiting on a reply right now, across everything. If answering it means opening more than one place, the answer does not exist yet — you are reconstructing it each time, which is a different and much more expensive thing.

Then check what your automations key on. Anything that sends on your behalf needs a field it can refuse on, not a convention you remember. I run several different ways in, and each one needs to sound like itself when it reaches someone.

The move cost me an afternoon. Eleven tasks renamed, one project created, five sections, four labels, one field added to the contact records, and three scripts taught to read it. What I got back was a single question with a single answer.

The folder structure was never the filing system. It was the thing stopping me from having one.

Sources & further reading

External
HubSpot — Use lifecycle stages
Salesforce Help — Leads
Gartner — Sales Survey Finds 61% of B2B Buyers Prefer a Rep-Free Buying Experience (June 2025)
Tim Brown, Change by Design

Related posts
Qualify inbound before the sales team acts
The lead went to a company struck off two years earlier
Three ways in, none of them a funnel