Step 1: Non-Shopping Campaigns: Start with manual CPC with eCPC enabled (because eCPC is the option that uses a bit of Google’s automation, but is mostly manual). This is nice because you can control the bids, but it’s not very scalable because it’s high maintenance, and because a machine will do it better than a human ever could. After you have around 100 conversions with high frequency (like a sale every other day, for example), at this point we’ve given the algorithm enough data for it to pick up where we left off and make it better. Now you can switch from eCPC to a smart bidding strategy like Target CPC. It leverages the data that you’ve already provided.
Shopping Campaignss: Start with a regular Shopping campaign before switching to a Smart Shopping campaign, because regular Shopping campaigns are inbound (meaning people are searching for the product when the ad appears) whereas Smart Shopping is outbound (meaning Google will try to show it to the type of people who might be interested), so the Smart algorithm can’t do outbound well, and therefore will never prove ROAS, without plenty of data. So, start with Shopping campaign using Manual CPC (usually w/enhanced enabled unless you have super low profit margin and you're trying to prove concept)
After 35 - 45 days of gathering consistent conversion data (aka converting at a normal cadence with a reliable pattern), then pause all shopping campaigns & replace it with a new Smart Shopping campaign using Maximize Conversions (which is the only bidding option). After 35 - 45 more days if the campaign isn’t hitting an acceptable ROAS, you can turn on the optional ROAS goal within the Maximize Conversions bidding strategy.
You don’t want to start off with a ROAS goal because it won’t get any conversions because it doesn’t have enough outbound data yet. Once you've given it that data, turning on the ROAS goal will restrict it to only spend when it thinks it can hit that goal, so spend starts to decrease, the campaign gets less activity, you get less clicks per day, but you should get a higher quality click and conversions that are making your ROAS.
One additional strategy is to spend more $/budget to force Smart Shopping campaigns to go find new markets by removing the Target ROAS goal.
Step 2: Now you can begin to step the ROAS goal higher. For example, if the campaign was at 300% when you first turned on Target ROAS, then try for 325%. If the campaign hits that or surpasses it, then try for 350%. At some point, you will hit a ceiling. Once you know how high the automation can go, get it back to the top and keep it there steady.
Step 3: Now that your ROAS is high, you can begin to give it more daily ad spend to see if you can increase the amount of conversions. So now, instead of spending $100/day and getting $300, you can try spending $500/day and getting $1500. Just make sure the ROAS doesn’t dip.
Display Pay Per Conversion - This is a new type of campaign/bid strategy in which you only pay for a conversion, and it’s only when the conversion doesn’t go above what you said you would pay. Begin with a super high CPA and then decrease it slowly. The idea is to get the CPA as low as possible (at which point you'll just turn off any regular Proactive Display campaign and just use this one). These types of campaigns can get lots of fake conversions generated by bots. Captcha prevents the bots from submitting forms and wasting ad spend.