February 13, 2014
Google Shopping is the next generation of PLAs. Offering a more retail centric advertising model, advertisers can easily segment their product feeds by categories and labels. This is all well and good but what do you do when performance is lacking? Do these new campaign types require new tactics or can you treat them like your other AdWords campaigns?
The account in questions is a medium sized e-commerce account. Performance is reasonable in search and this company has a large inventory as well as a few standout products that deliver the welcome pair of high sales and great returns.
So what was the issue? Google Shopping was lagging in performance. It was not an issue of the ads not showing. Impression share was high and conversions were decent. All metrics were steady and appeared healthy except for a stagnant ROAS.
On the search side, this account is very competitive and has great returns. Assuming we had the same products, shouldn’t these shopping campaigns be doing much better? Since these were newer campaigns that had not been running long prior to on boarding, perhaps it was just an issue of basic optimizations.
Negative keywords helped push numbers in the right direction. There were some easy wins and by eliminating overly broad keywords we were able to save a bit on wasted spend Since performance was not overtly terrible and these campaigns drove a large number of conversions there was the hope that gradual tweaks week over week would push the campaigns in the right direction.
This worked reasonably well but consistency was still an issue. Bids were now down to the benchmark levels and we were closing in on goals. What would happen if we ignored this benchmark though? Would undercutting this number actually lead to a decrease in conversions and loss of impression share?
Attacking the CPC’s
The graphs start at the point where the bids had been lowered to the benchmark CPC levels provided by Google. This number supposedly gives you and idea of how much your competitors are bidding on products, giving you some insight on the competition’s bid aggression.
If bids dropped even further could we still maintain adequate volume without succumbing to more aggressive competitors? We implemented steady roll out of decreasing bids. This kept us in control and protected us from doing anything to drastic. Adjusting for a higher ROAS lowered the bids in some campaigns significantly from the benchmark and the last thing I wanted to do was over correct and lose too much.
Surprisingly average CPC was not strongly correlated with decreased conversion volume over this period. Bids were halved on average, leading to a drop in impressions and a decrease in clicks but end of the line sales held relatively steady
More exciting was the fact that total cost dropped dramatically creating much higher ROAS ratios and much lower CPAs than the original goals we had aimed for. Through this we were able to both cut spend, prove the worth of continuing these campaigns and gained a greater initiative in expanding past the few thousand products we are currently listing.
Of course the last week was great in terms for performance and won’t hold forever. As anyone in paid search knows, there is more work to be done and more corrections to be made. While returns were fantastic the last few weeks, long term it by be better to increase bids again to recoup some incremental conversion volume while keeping a nice padding from the lower spend.
Nonetheless, I would not be surprised to see others benefit from this same tactic. Oddly enough staying competitive here actually meant bidding down by a large margin. Although it is tempting to keep pushing when you have the budget, some areas in paid search are more forgiving than others for potential overspend. Google Shopping in this case is not one of those, taking any money you put in, even for very marginal gains.
This should serve as a general reminder that nothing is sacred in paid search. By pushing bids lower and lower even past the supposed benchmarks, the gains kept piling up. It can be a trite expression but sometimes it really comes down to “maybe I should test that”.
Have you seen similar results, or even seen the opposite in your campaigns? Feel free to leave a comment. I’ll of course keep working on this, hopefully distilling some long term insights regarding Google Shopping performance.