Lesson 3.4: Primary, Secondary, Valued: Decision Rules
By the end of this lesson, you'll be able to run through the decision tree for any conversion action in your account and assign it the right role, with a clear rationale you can explain to a client or stakeholder.
Here's a practical decision tree for figuring out which role each conversion action should play.
Decision 1: Is spam or junk a large share of your form fills?
If yes, raw submit should not be your long-run primary. Full stop. Move toward qualified, bridge metrics, or stricter firing conditions. The "but we need volume" argument isn't wrong, but volume of junk isn't volume. It's noise at scale.
If no, you're still not done. Ask whether a form submit actually equals a buyer. The answer is almost always no. Move on.
Decision 2: Do you have enough truth events per month per campaign?
Think dozens, not ones.
If yes, the truth event can be primary for Smart Bidding.
If no, keep truth as secondary, use a bridge metric as a temporary primary, or apply values to the softer event so not all submissions look identical to the algorithm.
Decision 3: Is revenue imported with acceptable delay?
If yes, work toward value-based bidding on real dollars.
If no, use qualified plus points, or stage-based values, until imports are stable. "Stable" doesn't mean perfect. It means consistent enough that the signal isn't randomly missing for weeks at a time.
Decision 4: Are stakeholders attached to CPL on raw leads?
Stop debating in their vocabulary. Instead of arguing about why their CPL metric is wrong, add cost per qualified, cost per booked call, or pipeline per $1,000 spent to what they see, alongside legacy CPL, not instead of it. Let the numbers make the argument for you. A footnote that says "legacy CPL optimizes for a different animal than we're hunting" is gentler than a lecture, and it tends to land better.
A note on learning thresholds. Smart Bidding needs enough conversion events in a given window for each campaign or strategy to learn reliably. If "qualified" only generates a handful of events, everything feels like starvation mode. This is not an argument for lying with primaries. It's an argument for bridge metrics, portfolio bidding, or a temporary primary with a written exit plan. Written down. With a date. Not "we'll switch when it feels right."