Primary vs Secondary Conversion Actions: What Each One Actually Changes
The only thing that really changes is whether a conversion feeds bidding. Everything else is reporting. Here is how to set it without starving your algorithm.
The short version
- Primary feeds the "Conversions" column and your bidder; Secondary is report-only. There is no middle setting.
- Promote an action to Primary only if you would pay more for the click that produces it.
- Demoting your only high-volume action can starve Smart Bidding, especially under tROAS or long conversion lags.
- The "30 conversions in 30 days" idea is a loose rule of thumb, not a Google rule; value strategies and long lags need more.
- Audit what each action actually records before changing status; a duplicated or mislabeled tag poisons every downstream bid.
Setting a conversion action to Primary means it counts in the "Conversions" column and feeds your Smart Bidding algorithm. Setting it to Secondary removes it from that column and from bidding entirely: it still tracks, but only in the "All conversions" column for reporting. That single distinction (does this action train the bidder or not) is the whole decision. Everything else people argue about is downstream of it.
The one thing the toggle actually changes
In the old setup every enabled conversion action either counted or did not, and "Include in Conversions" was the switch. Primary and Secondary are the same switch with clearer labels. Primary actions roll up into the "Conversions" column. That column is what tROAS, tCPA, Maximize conversions, and Maximize conversion value all optimize toward. Secondary actions are stripped out of "Conversions" and dropped into "All conversions," where they are visible for analysis but invisible to the auction.
The mechanism matters because Smart Bidding does not read your intentions, it reads the "Conversions" column. When the algorithm sets a bid for a given auction, it is predicting the expected count (or value) of Primary conversions for that click and bidding to hit your target. A Secondary action contributes zero to that prediction. So demoting an action from Primary to Secondary is not a soft "count it less" move. It is a hard "count it as nothing for bidding" move. There is no partial credit.
This is why the common question "will making my newsletter signup secondary reduce how much the bidder chases it" has a blunt answer: it reduces it to zero. The bidder will no longer steer spend toward clicks that tend to produce newsletter signups. If those signups happened to correlate with real revenue, you just severed that signal.
The decision rule: what belongs in each bucket
Put an action in Primary only if you would genuinely pay more for the click that produces it. That is the test. A purchase, a qualified lead form, a booked demo: yes. A PDF download, a 30-second page dwell, a newsletter signup: almost always no, because you do not actually want to bid up clicks just because they tend to download a PDF.
The trap is mixing actions of wildly different value inside Primary. If a 2 dollar lead and a 2,000 dollar closed sale are both Primary and both count as "1 conversion" under a tCPA strategy, the bidder treats them as equal and will happily buy a flood of the cheap thing. Fix this one of two ways: move the low-value action to Secondary, or switch to a value-based strategy and assign real values so the 2,000 dollar action pulls more weight. Do not leave unlike actions competing as equal units.
Use Secondary for everything you want to watch but not optimize for: upper-funnel signals, micro-conversions you use for diagnostics, and lead actions you have not yet validated. Secondary is the honest home for "I want this in my reports but I do not trust it enough to spend against it."
Starvation: when demoting an action backfires
The real failure mode is volume. Smart Bidding needs a steady flow of Primary conversions to learn. When you promote a hard, low-frequency action (closed-won deals that happen a few times a month) to sole Primary and demote the frequent proxy (lead form) to Secondary, you can drop the bidder below the data it needs to set bids well. It stops learning and your performance gets noisier, not cleaner.
People reach for the figure "you need about 30 conversions in the last 30 days." Treat that as a crude rule of thumb, not a Google threshold, and know that it moves. A tROAS strategy optimizing on value needs more data than a simple tCPA, because it is learning a value distribution and not just a yes/no. A long conversion lag (B2B deals that close in 45 days) also demands more trailing volume, because recent clicks have not had time to convert and the bidder is working from a thinner recent sample. Short-lag ecommerce can learn on less. The number is a direction, not a line.
So the backfire is specific: you demote the only high-volume action you had, your remaining Primary action is sparse, and conversion lag is long. Now the bidder is guessing. In that situation keep the proxy action Primary (or use a value-based setup that credits both the proxy and the real sale at different values) until the high-value action alone clears a volume you are comfortable optimizing on.
How to make the switch safely
Changes to Primary and Secondary status apply at the account or conversion-goal level and feed bidding going forward, so treat any change as a learning-period event. The bidder will re-adjust to the new definition of "Conversions," and your CPA or ROAS numbers before and after the change are not apples to apples. Mark the date and compare like-for-like periods, not across the switch.
Before you touch anything, read what each action is actually recording. Go to Goals > Conversions > Summary and check the conversion count and value for each action over the last 30 and 90 days. An action you assumed was high-volume may be firing twice per purchase because of a tag duplication, or a "lead" action may be counting every form view. A conversion tracking audit here pays for itself: if the numbers feeding your bidder are wrong, the Primary vs Secondary decision is built on sand.
When you do promote or demote, change one thing at a time and give it a full conversion window plus the learning period before you judge it. If you flip three actions at once and performance moves, you will not know which change did it.
A quick worked example
Say you run lead gen. You track: form submit (frequent), phone call over 60 seconds (frequent), brochure download (very frequent), and "became a sales qualified lead" imported from your CRM (rare, lagged by a week or more). The instinct is to make only the SQL Primary because it is closest to money. If SQL volume is thin, that instinct starves the bidder.
The safer setup: keep form submit and qualified phone call Primary so the bidder has enough signal to learn, make brochure download Secondary (you watch it, you do not pay up for it), and either import SQL as a Secondary action for now or move to a value-based strategy that assigns a higher value to SQL than to a raw form submit. That gives the bidder volume today and a path to optimize toward real quality as SQL data accumulates.
The point is not a template. It is the sequence: identify which action you would truly pay more for, check whether it fires often enough to steer bidding, and only then decide what goes Primary. Volume constrains ambition.
Questions people also ask
Does a Secondary conversion action affect Smart Bidding at all?
No. Secondary actions are excluded from the "Conversions" column, which is the only input Smart Bidding optimizes toward. They appear in "All conversions" for reporting and analysis, but the bidder does not steer spend toward producing them. If you want an action to influence bids, it must be Primary.
Can I keep multiple Primary conversion actions?
Yes, and often you should to maintain enough volume for the bidder to learn. The risk is that under a count-based strategy like tCPA, each Primary action counts as one conversion regardless of its real worth. If your Primary actions differ a lot in value, either demote the low-value ones or switch to a value-based strategy and assign accurate values.
Will changing an action from Primary to Secondary reset my bid strategy?
It does not wipe your history, but it redefines what "Conversions" means, so the bidder re-adjusts and you should treat it like a learning-period event. Numbers before and after the change are not directly comparable. Change one action at a time and wait a full conversion window plus learning period before judging the result.
How much conversion volume do I need before making a rare action my sole Primary?
There is no official threshold. A common rule of thumb is roughly 30 conversions in the trailing 30 days, but treat it as a direction, not a line: tROAS and long conversion lags need more, short-lag ecommerce can do with less. If your rare action is below the volume you are comfortable optimizing on, keep a frequent proxy action Primary alongside it.
Should micro-conversions like PDF downloads ever be Primary?
Rarely. The test is whether you would genuinely pay more for a click just because it tends to produce that action. For most micro-conversions the answer is no, so they belong in Secondary where you can monitor them without letting them distort bids. The exception is when a micro-conversion is a validated, strong predictor of revenue and you lack enough real conversions to bid on directly.