Skip to main content

How MarginArc decides when to show plays

When you finish your pre-scoring questions on a deal, sometimes you see a candidate menu of strategic plays, and sometimes the deal goes straight to a standalone score with no menu shown. The thing that decides this is the eligibility gate.

What the gate does​

The gate is a fast model call that runs after your questions and before the per-candidate scoring. It looks at:

  • The deal context (OEM, customer, list price, quote shape, competitive pressure, deal-reg status)
  • Your free-form context dump
  • The Phase A assessment (the model's first read on the deal)
  • Your customer's history with you
  • The catalog of plays that survived the deterministic hard-rule filter

And it decides: are any of these plays actually worth showing the rep on this deal?

If the answer is yes, you see a menu. The menu's header reads "We see N strategic plays for your deal" and includes the gate's reasoning paragraph: a short explanation of why MarginArc thinks plays are relevant here ("Race-to-bottom on a 12-month subscription with no incumbent gives this an incumbency setup", or similar). The reasoning is collapsed by default and expands when you click "Show more."

If the answer is no, MarginArc auto-scores the deal as standalone (no play attached) and you go straight to the results page. No menu interrupts the flow on deals where it doesn't apply.

Why the gate exists​

The hard-rule filter is fast and deterministic but blunt. A play with a $50K minimum deal size will pass any deal over $50K, even when the deal context makes the play obviously irrelevant. The gate is the layer that catches "deal passes the rules but the play doesn't fit": a Cisco-AP-loss-leader play might pass on a Cisco deal of the right size, but if your context dump says the customer is consolidating to Aruba and we have no AP play here, the gate filters it out.

The gate is also where MarginArc avoids menu fatigue. If we showed a menu on every deal, reps would learn to ignore it. The gate reserves the menu for deals where a strategic frame actually changes the answer.

The escape hatch​

Even when the gate decides no play is relevant, you can still pick one if you disagree. The candidate menu has a small text link at the bottom: "or score this deal without a play →." The gate calls that direction; the link is the override.

When the gate showed you a menu but you don't think any of the candidates fit, click that link. The deal scores standalone with no play attached.

What's in the gate's reasoning​

The reasoning paragraph is intentionally brief: 2 to 4 sentences that point at the specific signals from your deal that mapped to a relevant play. Things like:

  • "12-month SaaS subscription with race-to-bottom pressure looks like an incumbency-setup deal."
  • "This is a follow-on at GSK where you ran a Buy-the-Subscription play 11 months ago. The OEM and timing align."
  • "Active Cisco AP play at this customer expects the DNA renewal in this window. This deal's product mix looks like that payoff."

If the reasoning doesn't match what you know about the deal, that's a signal to pick the standalone link. Don't feel obligated to engage with a menu that's reading the deal wrong.

What happens on the auto-scored path​

When the gate decides no play is relevant, MarginArc skips the menu and runs straight to standalone scoring. From your perspective, the deal flow is identical to scoring without plays at all: you see the scoring spinner, then the results page. There's no notification that "we considered plays and decided none fit." The decision is silent.

When the gate fails​

If the gate model call errors out, MarginArc defaults to showing the menu (any plays the hard-rule filter passed are shown). Better to show too many candidates than to silently skip plays that might have been right. You'll always have the standalone link to fall through.