A HubSpot portal holds 6,559 property definitions across 35 readable CRM object types (27 standard, 8 custom). A sweep of every one of them, run this week, found twenty-six properties whose name or label contains the word "trap": a coincidence of arithmetic with the twenty-six words below, not the same twenty-six. Read all twenty-six and five different names for the same idea emerge, written by people who never knew the others had reached for it.
One property, on Deals, is called Primary Trap Addressed. The identical idea, on Marketing Events, is Trap Focus. On Services, it is correctly called Primary Constraint Addressed: the right name, on the one object where somebody happened to land on it. On Listings, it is Trap Relevance. On Interests, it is Trap Indicated. None of the five defines what a trap actually is, because no canonical definition existed anywhere in the portal for any of them to reach for.
That gap is where Constraint Traps came from. Not a plan. A discovery: three separate people, at three separate moments, each needed a word for what's actually in the buyer's way, and each invented their own. The audit that found this happened mid-build on a HubSpot quote template, under a contest deadline, not in a strategy session, which turns out to be exactly the kind of moment a fragmented vocabulary actually shows up in.
The Trap Was Never the Constraint
VFT already publishes twelve Complexity Traps: the patterns that explain how an organization got complex in the first place, industrial-age thinking made visible. Constraint Traps are not a thirteenth. They live at a completely different moment: not how the org got here, but what a specific person says, in their own words, at the instant they're deciding whether to act. No time. No budget. The data's a mess. We tried this before.
The two are designed to compose, not compete. A Constraint Trap is what someone reports; a Complexity Trap is often what produced it. Two of the twenty-six collide with a Complexity Trap closely enough that mixing them up sends the wrong work entirely, and both are flagged below, at the word each one meets.
Here is the turn that makes twenty-six words a piece of methodology instead of a list a HubSpot admin wrote on a Friday: the trap is not the constraint. The trap is treating the constraint as fixed. Not having time is not the trap. Believing time cannot be reclaimed is.
That distinction decided the shape of the set. An earlier draft tried to reduce the six original dimensions (time, money, capability, energy, confidence, relationships) to an orthogonal minimum, on the theory that fewer words is cleaner. Chris ruled against it:
Forcing someone whose pain is capacity to decompose it into time-plus-capability isn't precision. It's friction, at the exact moment friction costs the most, and two agents asked to do the decomposing will do it two different ways. Twenty-six words, not six, is the design working as intended: recognition, not derivation. The reader's job is to find their own word on the list. Not to build it.
Not every word on this list gets an AI-changes-everything line, and holding that discipline matters more here than anywhere else in the set. A few of these are exactly as fixed as they've always been. Knowing which ones is the actual skill.
What "Not Enough" Actually Measures
Three of the twenty-six are the oldest excuses in the room, and each one used to be a straightforward fact about headcount.
Time. Not enough hours to do it properly. The hours bound when every pass needed a human. Much of the first pass no longer does, so "no time" is increasingly a question of what still has to be yours — and treating the whole of it as yours is the trap.
Money. Budget won't cover what's actually needed. AI didn't add budget; it changed what a budget buys. The trap is scoping at pre-AI labor rates and concluding the outcome is unaffordable.
Capacity. Not enough people or throughput, regardless of budget. Headcount was the throughput ceiling. Staying capacity-bound now usually means the work itself was never shaped for anything but a person to do it.
None of these three is false as a description of the org today. The hours genuinely aren't there. The money genuinely isn't either. What changed is what those facts are allowed to imply about next quarter.
Expertise Doesn't Have to Arrive First Anymore
Four words about what the building itself can do, independent of who's in it.
Capability. The skills or knowledge aren't in the building. Expertise used to have to arrive before the work could start. Now it can arrive with the work — waiting to hire it first is the trap.
Tooling. The systems in place can't do what's being asked. "The system can't do that" was final when changing the system meant a project. Platform limits are more negotiable than they look, and accepting the sentence as the end of the conversation is the trap.
Manual work. Work done by hand that shouldn't be. Manual used to be the safe choice, when automation was brittle and expensive to build. That trade has inverted. Manual is now the path accruing the risk.
AI readiness. Not prepared for agents to work safely on this. Readiness isn't a prerequisite you achieve and then begin — it's produced by doing the work. Waiting to be ready is how organizations stay unready.
Not Knowing Used to Cost a Quarter
Three words about whether anyone can actually see what's true.
Clarity. Nobody is sure what's actually true. Not knowing used to cost a quarter of analysis. Now it costs a question — but only if the truth sits somewhere a machine can reach, and still budgeting the quarter instead of moving the truth somewhere reachable is the trap.
Data quality. The data exists but can't be trusted. Dirty data used to slow humans down, who could usually tell when something looked wrong. It makes AI confidently wrong instead, which is worse: every agent deployed against it inherits the data's honesty, or the lack of it.
Visibility. Can't see what's happening without building a report by hand. Building the report used to be the work. Now asking the question is. The trap is still budgeting for dashboards instead of the data underneath them.
The Trap Under the Trap Is Always Human
Five words, and this is the group where the reframe has to be earned most carefully, because none of these clear up in a single afternoon.
Energy. The team is drained; nothing left for another initiative. Burnout has many causes, and one of them is carrying complexity no human should have to hold — that part is exactly what AI-Human Partnership is for. Leaving that share of it on people while it no longer has to be theirs is the trap.
Confidence. Doubt this will work, or that the team can pull it off. Confidence used to require precedent, so anything new stayed undone until somebody else proved it first. When the cost of trying collapses, an organization can find out instead of deciding in advance.
Trust. Not willing to hand this work, or this data, to an AI or an outside party. Nothing makes AI or an outside party inherently trustworthy, and starting from no is the correct position. What changed is the cost of checking — what a system may touch, what it read, and what it wrote, are observable per action, so trust can be extended in increments against evidence instead of granted up front. Treating a correct first no as a permanent one is the trap.
Change fatigue. Burned by past initiatives that didn't stick. Change used to hurt because it failed slowly and expensively, over quarters. Shorter cycles teach a different lesson about change entirely, but only if the next attempt is genuinely shorter, not the same project wearing an AI label.
Adoption. The team won't actually use what gets built. Adoption failed for years because software made people conform to it. AI can meet people in their own language instead. Shipping another tool that demands conformity and calling it AI is the trap, not the tool itself.
Somebody Is Still Choosing the Wall
Five words about the boundaries between people, and every one of them turns out to be more chosen than it looks.
Relationships. Misalignment between the people who have to agree. AI doesn't align people, but it makes it cheap to put the same truth in front of everyone — which is what most misalignment actually needs. Treating every disagreement as needing another meeting, when it needs a shared view, is the trap.
Decision rights. No clear owner; decisions stall. Decisions used to stall because assembling what they needed took weeks of someone's calendar. When context assembles in minutes, the bottleneck gets revealed for what it actually is: ownership, not information.
Silos. Teams or data separated in ways that block the work. Silos existed because moving context across a boundary cost real human effort, every single time. That cost is near zero now, so a silo that's still standing is a choice somebody is still making.
Process. No agreed way of working to build on. Undocumented process was tolerable when only humans had to improvise around it. An agent can't improvise your intent. Adding AI before deciding how the work should actually go is the trap.
Dependency. Reliant on an outside party to function. Dependency was the price of scarce expertise, and expertise was scarce because it could only arrive inside a person. Capability can now transfer with the work, so continuing to treat your own dependence as a permanent market condition — rather than something a different engagement shape can end — is the trap.
The Debt You Assumed Was Permanent
Six words, and this is the group where the discipline this whole piece is built on has to hold the line, not loosen it.
Integration. The pieces don't connect. Connecting systems used to be a project with a vendor and a line item. Increasingly it's a conversation. Treating every connection as a capital expense is the trap.
Migration. Getting off the current thing is the hard part. Migration used to be expensive enough that organizations stayed on tools they'd outgrown for years, just to avoid it. Moving the data is no longer the hard part. Deciding what deserves to move is.
Customization. Past customization now limits what's possible. Customization became permanent debt because unwinding it meant re-learning a system nobody documented. AI can read the system and tell you what it actually does — treating old decisions as immovable because nobody remembers them is the trap.
Scale. What works now won't hold as the org grows. Scaling used to mean hiring proportionally, so growth and cost moved together automatically. That link is broken. Designing for a headcount curve nobody has to follow anymore is the trap.
Compliance. Regulatory, security, or privacy limits. Compliance is real, and it does not go away. What changed is the cost of proving it. Using compliance as the reason not to build is the trap, not compliance itself. Compliance is the third of the twenty-six, after money and trust, that makes no promise of dissolving anything. That restraint is what keeps the other twenty-three credible.
Vendor lock-in. Trapped with a provider. Lock-in worked because switching used to cost more than staying put. As moving context between systems gets cheaper, that math weakens. Renewing out of inertia and calling it strategy is the trap.
What the List Is Actually For
The properties that started this didn't fragment because anyone was careless. Naming what's actually in someone's way, honestly, in their own words, at the moment they're deciding whether to move, turns out to be hard enough that three separate people solved it three separate times before anyone noticed they'd all solved the same problem.
The list existing at all is most of the fix. A reader finding their own word on it, and then finding the sentence next to it that says why the word has quietly stopped meaning what it used to mean, is the rest.
A constraint is a fact. A trap is a decision dressed as one.
Worth passing on?
