Smart Shopping, Google’s completely automated, all-in-one delivery system for product listings on search, has been out for just about a year. What have we learned about Smart Shopping campaigns in that time? There’s good, there’s bad, and there’s just plain ugly.

The Good

When Smart Shopping works – and it doesn’t always – it can get great results. I’ve used it successfully to salvage underperforming product category campaigns for larger accounts with multiple shopping campaigns. I’ve created a Smart Shopping campaign for a category with a 200% target return on ad spend (ROAS) goal and let the learning period do its thing for a minimum of 15 days. Once it was out of learning and the goal was met, I was able to raise the ROAS goal and sustain the volume.

I’ve seen excellent results with this strategy because you aren’t relying on Smart Shopping to control all the shopping results – your efforts are diversified. If Smart Shopping doesn’t work for a specific campaign, you haven’t brought your shopping results to a screeching halt. Starting small and targeted will allow you to experiment with Smart Shopping to see if it works for your goals.

The Bad

When Smart Shopping doesn’t work, there’s not much you can do to improve it. Google usually gives one of four reasons why it’s not working:

  • You didn’t give it good enough data – the feed.
  • You didn’t give it enough money – the budget.
  • Your target ROAS wasn’t set low enough – the only bidding option you have.
  • You didn’t give it enough time – 15 days plus the average lag from click to conversion.

But what if you gave it good data, a generous budget, a nice low target ROAS to start with, and plenty of time … and it still doesn’t work? Your only option is to try lowering the target ROAS even more.

In other instances, you can get the campaign to run, but the volume is considerably less than your standard shopping campaign. Again, your only option is to lower your target ROAS, then slowly bring it back up. Sometimes that works, but sometimes it doesn’t.

Once you’ve pulled all the levers you have – data, budget, time, target ROAS – there’s really nothing else to be done except return to standard shopping campaigns.

Since this is new and fully reliant on automation, common issues can completely mess up the campaign. For example, if conversion tracking breaks, or there is a mobile site issue, or you need to pause the campaign for any reason, Smart Shopping performance suffers drastically.

Google did try to anticipate this with Seasonality Adjustments, which allow you to give the system more information about conversion rates. Seasonality adjustments let you tell Google’s algorithm if you have a situation, like a sale, that may cause the conversion rate to jump so that it knows this is a temporary blip. You can also use seasonality adjustments to blackout dates on which you’ve had issues by showing a conversion rate drop. However, we’ve found that while performance improves somewhat, it most likely won’t recover quickly or completely.

Another issue with Smart Shopping is that we can’t exclude mobile, which has been an issue more times than you’d think. We’ve had clients that have issues with their mobile sites, but since we can’t turn mobile off, we have to turn off the entire Smart Shopping campaign. We’ve also had clients that want to test turning off mobile targeting, and their only option is to turn off the Smart Shopping campaign and go back to standard shopping campaigns.

The Ugly

Here’s the truth about why I don’t adopt Smart Shopping more widely: These campaigns lack transparency.

Perhaps it’s because they are new, or maybe because Google is still learning, but Smart Shopping is like saying, “Here Google, do whatever you want with the money I give you. Try to meet my goal. I trust you completely.”

And that’s the issue. If a Smart Shopping campaign fails, we can’t analyze out why.

Google doesn’t provide any data to help us analyze the results. We can’t see where the ads were shown. We can’t see the keywords that people searched. We can’t see the percentages between the spend and revenue from search or what’s being shown on display, between remarketing or prospecting.

The lack of transparency is a legitimate issue. We use search query data to help expand search keywords, but we can’t see the data here. We look at breakdowns between remarketing and prospecting on display, but that data is also not available to mine.

The Google teams that we’ve talked to say the engineers won’t budge on the issue – why show something that we can’t manipulate – but this response just makes us wonder what Google is hiding. Google also says that automation gives us more time for strategy, but when they make data unavailable it’s harder to develop the strategy.

I’ve been managing paid search accounts for more than 17 years. Google’s stance on giving us less data with Smart Shopping reminds me of when Google advertising was new. They didn’t share search queries then, and they were a lot less transparent than they are now.

If Google wants greater Smart Shopping adoption, particularly among experienced PPC practitioners, they need to work with us. We’ve got our clients’ best interests at heart, but moves like this – as well as the recent blurring of match types and turning on conservative automation for remarketing campaigns –make it feel like Google doesn’t.