Incubator campaigns
How BidSmart's incubator, a separate Maximise Clicks campaign, ramps new and thin-data products cheaply, and the criteria by which they graduate to your main tROAS campaign, always via your approval.
New products and thin-data products have no reliable performance history, so putting them straight into a target-ROAS campaign either starves them or lets them drag the average down. The incubator is a separate, low-cost campaign that buys them clicks and data cheaply until they've proven themselves. Once a product clears its graduation criteria, BidSmart proposes moving it to your main campaign, and the move only happens when you approve it.
The thin-data problem the incubator solves
Target ROAS is a prediction engine, and predictions need history. A product with four conversions has no usable ROAS: put it in a tROAS campaign and Google is guessing. Two bad outcomes follow: either the product gets starved (no data to justify a bid) or it draws erratic spend and drags down the campaign average that your proven products depend on. Neither is what you want for a product you're trying to get off the ground. See Target ROAS explained for why thin data breaks the algorithm.
What the incubator is
The incubator is a separate Maximise Clicks campaign, kept apart from your main tROAS campaign, with its own budget. Its job is deliberately different: it isn't chasing a return target, it's buying clicks and conversion data as cheaply as possible so a new or thin-data product can build the history it needs. Because it's ring-fenced, an unproven product can gather data without touching the target average your established products rely on.
New products and products BidSmart flags as too thin to bid on go here: always as a recommendation you approve, never automatically.
Graduation: from incubator to main campaign
Deep dive The graduation criteria and how the move is staged
A product graduates out of the incubator once it has earned a reliable target: not on a timer, but on evidence. BidSmart watches each incubator product against a graduation bar built from:
- Conversion volume. Enough conversions to compute a target that isn't noise: in practice this is the binding constraint. A product with a handful of conversions hasn't proven anything yet, however long it's been running.
- Time in market. Enough elapsed time that the data isn't a single lucky week: this guards against a product that spiked once and won't repeat.
- Consistency. Conversion behaviour steady enough that a product-level target derived from it will hold up when the product moves into a tROAS campaign. See product-level tROAS analysis for how that target is formed.
When a product clears the bar, BidSmart proposes graduating it: moving it from the incubator into the main tROAS campaign, with a suggested starting target derived from the history it built up while incubating. That proposal is staged for your approval: it appears in the approval queue like any other change. The move executes only when you sign it off, records its before/after state (which campaign it left, which it joined, its new target), and can be reversed.
Products that never gather enough data don't silently graduate on thin evidence and don't silently vanish either. They stay flagged so you can decide: keep incubating, pause, or accept that the product just doesn't get enough traffic to advertise on a target. The incubator automates the watching and the proposing; it never automates the decision.
What stays under your control
- Creating the incubator is a proposed setup you approve, not something imposed on your account.
- Moving a product in (a new or thin-data product) is a queued recommendation.
- Graduating a product out is a queued recommendation with a suggested starting target.
- Every move records before/after state and is reversible.
The incubator handles the busywork of spotting which products need data and which have earned promotion. The approval-first rule still holds at every step: nothing enters, moves within, or leaves the incubator without your sign-off. See how BidSmart works for the full model.
Frequently asked questions
Why not just put new products straight into my tROAS campaign?
Because tROAS needs conversion history to bid well, and a brand-new product has none. Google ends up either starving it (no data to justify bids) or spending erratically on it. The incubator sidesteps that by optimising for cheap clicks and data first, not for a target it can't yet hit.
Does BidSmart create and move campaigns on its own?
No. Setting up the incubator, moving a product into it, and graduating a product out of it are all recommendations that wait in your approval queue. The move executes only after you sign it off, and it's reversible.
How long does a product stay in the incubator?
Until it clears its graduation criteria, enough conversions over enough time to support a reliable target. That's typically a few weeks, but it depends on the product's traffic. A product that never gathers data stays flagged for your decision rather than graduating on thin evidence.