Separate Hard Rules From Interpretation

A request arrives at GTM Lab from a genuinely enormous company. The revenue figure clears every threshold. The matched account is in the target segment. On the numbers this is the best inbound of the month.

The message field reads: asdfgh test test please ignore.

Every quantitative rule says route it to a seller. Every human who looks at it says obviously not. The rules and the judgment disagree, and the system has no way to represent that, so it routes it, and a seller spends twenty minutes finding out that a form got tested.

Two kinds of rule, two kinds of failure

A hard rule is checkable. Anyone applying it to the same record gets the same answer, today and next year. Revenue at or above a threshold. The email domain matches a known vendor list. The record already has an open opportunity. An interpretive rule is a judgment where two competent people could reasonably differ. Does this message read as a genuine enquiry? Is this use case one GTM Lab actually serves?

Keep them in separate sections of the card, for three reasons that are all practical. They fail differently: a hard rule fails by being wrong about the world, which you find out sharply, while an interpretive rule fails by drifting, which you find out slowly. They are owned differently: a threshold is a business decision someone can approve in a meeting, while a judgment standard needs worked examples before anyone can agree to it. And only one of them can be tested by example. You can prove a hard rule with arithmetic. The only way to pin down an interpretive rule is to write down the cases, which is why interpretive rules that live only in someone's head are the most expensive part of any decision system to hand over.

Order matters, and a later pass does not overturn an earlier fail

Rules are evaluated in a stated order, and the order is part of the definition, not an implementation detail. GTM Lab's order runs the two interpretive vetoes first: is this person attempting to buy from GTM Lab at all, and does the submission read as genuine? Both fail to not a prospect and stop evaluation. Nothing below them can rescue the record. Then the hard rules: is there a matched account already prioritized for pursuit, and is the company at or above the scale GTM Lab sells to. Then a final interpretive rule on use-case fit, where fit alone does not qualify, because a good fit below the scale threshold is keep and nurture rather than route to a seller.

A qualitative rule may veto a quantitative pass

This is the answer to the opening. The enormous company with the test message clears the matched-account rule and the scale rule comfortably, and never reaches them, because the genuineness rule already stopped it.

Write the veto down explicitly, because the alternative is that somebody eventually “optimizes” the order for speed by checking the cheap numeric rules first, and the behavior changes without anyone believing they changed a rule.

The same record, two orderings

Run the test-message request through the order above: the genuineness rule stops it, the outcome is not a prospect, and the reason recorded is that the submission does not read as genuine. Now run the same record with the hard rules first, which is the ordering almost every team reaches for because it is cheaper to compute: the matched account satisfies scale, the record is routed, and the interpretive rules are never consulted because the answer was already found.

Same record. Same rules. Two outcomes. The ordering was the decision, and if it is not on the card then nobody decided it.

Carry this into your business

Write down the order your existing rules run in, then ask whether anyone chose it. Ordering is usually inherited from whatever was easiest to build first, and it is doing more to determine your outcomes than the thresholds everyone argues about.