DEMAND GEN BIDDING STRATEGIES GOOGLE ADS: WHAT HOLDS UP IN THE LEARNING PHASE

When people ask me about demand gen bidding strategies google ads gives you, they usually want a ranking best to worst. That's the wrong frame, and it's the frame that's cost more than one account real money. The choice between tCPA, tROAS and Target CPC isn't about which is strongest; it's about which one the campaign can actually support given how much data it has. Pick a strategy the campaign can't feed and you don't get a slightly worse result, you get erratic delivery and wasted spend. This is the view from someone who manages live budgets, not a settings tour: the strategy that holds up in week one is rarely the one that holds up in week six, and the most expensive errors I've made here were all about sequencing. If you want the platform background first, the primer on what Demand Gen is sets the stage before this.

It helps to be precise about what these strategies are before arguing about when to use them. tCPA chases a cost per conversion, tROAS chases a value-weighted return on spend, and Target CPC chases a cost per click. Three different goals, three different data appetites. The learning phase the stretch right after launch where Google's system is still modelling your audience and conversions is where those differences become brutal, because a strategy that needs data it doesn't have yet simply stalls. That's the whole story here: matching the strategy to what the campaign can carry, and not jumping ahead.

The three strategies, and what each one actually asks for

The demand gen bidding strategies google ads exposes split cleanly into efficiency strategies and volume strategies, and the distinction matters more than the names suggest. tCPA and tROAS are the "target" strategies: they're built to hit a number you set, a cost per conversion or a return on spend respectively. Maximize Conversions and Maximize Conversion Value are the "max" strategies: they spend the budget to drive volume without holding a target. Target CPC sits apart as a click strategy, optimising toward a cost per click rather than a conversion outcome at all. Each of these makes a different demand on your data, and that demand is the thing to look at first.

tCPA needs a base of conversions to model a cost target reliably, but it's relatively forgiving it only has to predict whether an impression converts, not how much that conversion is worth. tROAS is the hungry one: it needs reliable conversion values, not just counts, and enough of them to model return per user. Target CPC needs the least conversion data because it isn't optimising for conversions; it just needs to land clicks at a price. When I think about which of the google ads demand gen campaign bidding strategies a campaign can support on day one, this hierarchy of appetite is what I'm reading, and it explains almost everything about what happens next.

Which one holds up in the learning phase

Here's the practical answer from my accounts, and it's the one that reframes how I rank the google ads demand gen campaign bidding strategies: a brand-new Demand Gen campaign should not open on tROAS, and usually shouldn't open on tCPA either. It should open on Maximize Conversions. The logic is the data appetite from the last section. A cold campaign has zero conversion history, and a target strategy laid on zero history has nothing to optimise toward, so it throttles delivery while it waits for signal that the throttling itself is preventing. Maximize Conversions breaks that loop: it spends the budget freely, gathers conversions fast, and lets the system build the model the target strategies will later need. I treat it as the learning-phase engine, not as a permanent setting.

The threshold I work to is roughly 50 conversions before I consider switching to a target strategy, and that's not a number I invented it's the volume at which Google's system has a foundational read on what a good impression looks like for that campaign. Below it, a target is guessing. There's a budget dimension here too that catches people out constantly: if the daily budget is too low, the campaign caps out on daily conversions and never accumulates the volume the learning phase needs, which is, in my experience, the single most common reason a new Demand Gen campaign underperforms. I keep the budget at roughly ten times the target CPA so the system has room to find conversions. Starve it and even the right strategy stalls, which is the same dynamic I watch for when I think about overall Demand Gen best practices at launch.

The tROAS mistake that burns budget

This is the part I learned the expensive way. On one ecommerce account the early Demand Gen numbers looked strong ROAS was tracking above target after only a couple of weeks and I made the move I now warn everyone against: I switched to tROAS at around twenty conversions because the data looked good enough. It wasn't. Within days the delivery went erratic: some days the campaign barely spent, other days it overspent and chased low-quality conversions, and across the week the cost per conversion climbed while volume slid. The campaign wasn't malfunctioning. I'd handed a value strategy the job of optimising on conversion-value data that didn't exist yet, and it did exactly what an underfed value strategy does it lurched.

The fix was humbling in its simplicity: I reverted to Maximize Conversions, let the campaign rebuild past the 50-conversion mark, and only then layered tROAS back on with a deliberately relaxed target. Google's own guidance is explicit that you should accumulate conversions with value first and shift to tCPA or tROAS only after the campaign has received at least 50 conversions, and that switching too early is precisely what causes the volatility I saw the detail is laid out in the official Google Demand Gen bidding guidance. The early-tROAS switch is, hands down, the fastest way I've found to set money on fire in this campaign type.

The sequence I run now

So the workflow is a sequence, not a single pick. I launch on Maximize Conversions and leave it alone through the learning phase, resisting the urge to optimise on a handful of early conversions. Once the campaign clears roughly 50 conversions and the cost per conversion has stopped swinging wildly day to day volume alone isn't the trigger, stability is I step up to tCPA, because it's the forgiving target strategy that tolerates the data I now have. Only when tCPA is steady and I've got a genuine base of conversions with value do I consider tROAS, and even then I set the target deliberately loose to give the system room rather than clamping it to my historical best.

When I do change a target, I move it gently on the order of 10 percent at a time and then I leave the campaign through a fresh settling period instead of nudging it daily. Big jumps re-trigger learning and throw away the stability I waited for. This patience is the single biggest difference between the campaigns of mine that hold up and the ones that thrashed, and it's the same discipline I apply when reading whether a campaign is genuinely fatiguing or just mid-adjustment, the kind of signal I track alongside creative fatigue in Google Ads. The strategy that holds up isn't a brand of magic; it's the one I didn't rush.

Where Target CPC actually fits

Target CPC is the most misused of the demand gen bidding strategies google ads offers, because it sounds like a safe, controllable middle option, and it isn't a conversion strategy at all. It optimises for clicks at a price you set, full stop. That makes it genuinely useful for a specific job: upper-funnel consideration, traffic generation, and seeding audiences I'll later remarket to in a separate conversion-focused campaign. When the objective is "get relevant people onto the landing page at a predictable cost", Target CPC does that cleanly and needs almost no conversion data to function.

What it must not do is carry a campaign you're judging on cost per conversion or return. Pointing Target CPC at a CPA goal is asking a click strategy to deliver a conversion outcome it has no mechanism to optimise toward, and you'll get clicks that don't convert at a cost that looks fine until you check the downstream numbers. I keep it firmly in its lane click-led objectives and audience building and I never let it stand in for tCPA. Slotting each strategy to its actual job is, in the end, the same mindset I bring to picking among broader Demand Gen campaign setups: the tool has to match the goal, not the other way round.

The practical takeaway

If you remember one thing: with the demand gen bidding strategies google ads provides, the question is never which is best, it's which one the campaign can currently support. Open on Maximize Conversions, let the learning phase gather real volume, step up to tCPA once you've got a stable base around 50 conversions, and only reach for tROAS once you have genuine conversion value to feed it. The error that burns budget is switching to tROAS too early, and I've paid that tuition so you don't have to.

So I no longer treat the bid strategy as a setting I lock in at launch. I treat it as a sequence I walk the campaign through, sized to the data at each step, with Target CPC reserved for the click-led jobs it's actually built for. Do that, and the bidding question stops being a gamble and becomes a fairly predictable progression.