NEWSFLASH: Google turns off Google Product Search in One Box….

Starting yesterday, June 27, our data indicated that Google started to send the bulk of the “one box” traffic for commercial searches to the new ‘step 1’ Google Shopping experience.  In calls with retailers we are getting a ton of questions about the change(s) going on at Google, so wanted to continue sharing common questions and our answers along with news updates as they happen.

What is One box traffic?

Internally at Google called the integration of product search and main search is called Universal Search, and externally One Box – this is where historically the product results from Google Product Search (GPS) were integrated right in with the organic and paid search results.


For pretty much every commercial/product search you will now see this new treatment.  Here are some examples.

Here’s the search for ‘electrical generator’:











Here’s the result for ‘mascara’:


Here’s the result for ‘samsung phone’:


Why does this matter?

The original timeline that Google published for the beginning of some testing of traffic was June 1 with July 1 as the ‘change over begins in earnest day’ and Oct 15 as the move is complete day.

This abrupt change in traffic on June 27, indicates that Google has moved that July 1 date up.  In fact, many retailers were thinking there would be a slow change from July 1-> Oct 15, but this one box move signals that the bulk of the traffic will now flow through the new system.

Is GPS totally turned off?

There are still a couple of paths that consumers can take to get to the old GPS experience and thus generate free traffic to retailers.  The first path is through the new One Box Google Shopping Experience.  In the PLA below, if the user clicks any of the product images, or retailers, they are taken to the retailer an the PLA CPC is incurred.  However, if they click on the green ‘’ link they still go to the old GPS.

The two other paths consumers can take to the old GPS:

  • Direct URL navigation to
  • Clicking on the shopping tab from Google SERPS or home page

How much traffic is now diverted?

By our measure, the one-box delivered 80% of the traffic to Google Product Search.  While you could argue that there is still a path there from the onebox, we have seen traffic drop about 80%, so we believe that not many consumers are actually clicking on the green shopping link.  This makes sense because the other options are much more visual and appealing.

What should retailers be doing?

We continue to have a lot of  retailers calling us with a lot of questions on the changes.

Here are our recommendations:

  1. Education – First it helps to fully understand the changes and what they mean for your business.  We have a lot of content here on the CSE Strategies blog and ChannelAdvisor also had a Google co-hosted webinar that was an excellent source of information.
  2. Become a Google Shopping and PLA expert – The PLA system has flown under the radar since 2009 and we’re finding a lot of retailers have not understood the program very well nor have they optimised it.  More information on common misconceptions is below.
  3. Start optimising Google Shopping and PLAs – While we haven’t seen the complete secondary and tertiary experience roll out yet (except for tents),  even with this big first step change in one box, you can start optimising your PLAS.

Google Shopping + PLAs are not search marketing

One common mis-perception we are finding is that retailers assume that because a) this is Google and b) it’s CPC and c) you manage part of it in the adwords user interface, this is another program just like Adwords.  That couldn’t be further from the truth.  Here are some of the reasons it is NOT like adwords and therefore should be managed differently:

  1. The User Experience is a CSE – See our walk through here for details.
  2. Products not keywords – Instead of thinking at a keyword+ad levely, you have to optimise at a target+product level.  Google’s Adwords system has no concept of products.  You  control the product-level information via the datafeed which you have to produce.
  3. Targets?!– The atomic level at which you bid in PLAs is the target.  You group products into targets and then assign a bid to that set of products.  This creates some important and strategic questions:
    • Should you group products by brand?
    • Should you group by category?
    • Should you group by ASP?
    • Should you group by promotional activity?
    • Should you group by margin?
    • Should you group by ROAS?
    • Should you group by Conversion rate?
    • What do you do if a product goes out of stock – keep it in the target?
    • Maybe you should group by ‘price competitiveness’?

Here’s the trick – the answer to the above questions is sometimes yes, sometimes no and sometimes it changes on a daily basis.  Most frequently you want to matrix and optimise against 2-3 factors.

  1. Datafeed – The way you create and optimise targets is via datafeeds.  You’re going to want to create as tight a loop between your optimization strategies and your datafeed.  If you have to go to IT every time you want to optimise your datafeed for PLA, you will get eaten alive by competitors that can out cycle you.
  2. Target transparency – you will want to be able to ‘look into’ the targets to see the metrics on a per-product level.  Google does not allow you to do this, SEMs don’t understand products and will not be able to help you here.
  3. Search bid optimization strategies won’t work.  In the keyword search world, some advanced strategies have been developed (that are generally questionable for retailers to begin with, but we’ll save that debate for a later day).  These approaches do not make sense with Google Shopping and PLA:
    • Efficient Frontier For CPA-based business models, where you are focused on acquiring a customer or an event (e.g. a mortgage) it makes sense to overpay for some leads and underpay for others and optimise your budget.  This makes no sense in PLA – what if you overpay for a target that is lower margin products and underpay for a target that has higher margin costs?
    • Portfolio bidding – Actually this is very applicable and is essentially what google has implemented.  Targets are your different portfolios and you treat them differently with different bids.
    • Predictive bidding – These approaches allow you to ‘simulate into the future’ and answer questions such as: If I bid $X, how much more traffic will I get and consequently revenue?  This approach doesn’t work, because unlike a keyword/ad combo in Adwords where you know if I bid on the keyword “lcd monitor” I will show up, in PLA, you don’t know what target will show up for which search and you also don’t know which product within the target did/would/could show up.  In other words, by its very nature, PLA is unpredictive.  You would do much better looking at past performance than trying to predict the unpredictable future performance.

(Note: Typepad won’t let me fix the numbering above.)

Ten questions to ask your Google Shopping/PLA solution provider

We’ve suggested that retailers use questions like the following to help understand if your current PLA or proposed provider is going to be able to really help you with this program:

  1. How should I set up my Targets?  What is your initial recommendation?
  2. Will you need us to make frequent datafeed changes?  How much time will you allow/need?
  3. Can you provide sku/product level details on PLA/GS performance?  (sales/sku, cost/sku, CR/sku, margin/sku, etc.)
  4. How do you plan to optimise – against spend, revenue, margin, ?
  5. Do you have business rules that automatically change datafeeds?
  6. What is the ROAS your e-commerce clients are seeing with GS/PLA?
  7. Do PLA’s convert worse, or better than adwords?
  8. If I have a fixed budget, what % should be adwords and what % PLA?
  9. If a product I don’t want is showing up for a GS result, can you turn that off or optimise the product that does show up?
  10. What are some common optional attributes I need to make sure my feed has to improve my GS/PLA performance?
  11. What’s an example where you would bid-down a Target?
  12. How does predictive modeling work with GS/PLAs?


Examples of unoptimised retailers that are burning money on Google Shopping/PLA

One step you can take immediately is start doing searches for your top selling products.  You will start to see some very interesting behaviours which really points out the importance of tightly managing this program.  Here are a couple of examples we have found that are interesting.  When talking about the ads, we use a numbering system from left to right like this:


Example 1:  front loading washer lg:

In this example, we are looking for a front-loading LG washer.  You can imagine for a $1000+ appliance, you would be willing to pay a pretty big CPC, so to compete for this type of product, you have to be bidding north of $5/click most likely (assuming the market is rational).  If you are paying that much, you want to make sure you are showing a high-end product:


Here advertiser 5 would not want to pay a high CPC to have their $160 product.  When you see these large swings of ASP in one set of results, it usually points to a program that is running on autopilot and the retailer is probably losing not only a lot of $ on a cost/product basis, but most importantly the opportunity cost is very high.  What if they had the WM2250 at a competitive price?


Example 2: Lawn sweepers anyone?

One of the top selling outdoor items is a lawn sweeper (who knew?!). Here’s a search result:


One of these things is not like the other….Yes, you guessed it, Amazon (well Google) has served up a floor sweeper and not a lawn sweeper.  Again, you would think the economics would take are of this.


Example 3: 42 inch TVs

Here, we are starting to look for a 42 inch TV and are looking at panasonics:


Spot number 2 looks like a great price, but the image is a little weird.  When you click through, this is some kind of an internal part for a panasonic 42″ TV.  What you’ll find is that due to their complex and unorganised catalogueueue, eBay has some of the wackiest things they are advertising in Google Shopping.  It will be interesting to watch and see if they keep this level of spend up or not.  Also, Google has to realise it is a bad user experience at some point and do something about it.


Example 4: Mustang bumpers

Here we have been in a bit of a fender bender and are looking for a new bumper for our classic Mustang.  Boy there sure are some great deals!  BTW, this one wins the ‘biggest swing in ASP’ that I have found – a whopping $1,665!



Based on what we are seeing, here are some closing thoughts:

  • Google seems to be moving faster here than anyone initially thought
  • A lot of retailers are really scrambling
  • Unfortunately many are mistaking the new Google Shopping CSE as another search offering
  • They will be at a competitive disadvantage.
  • Heading into Q4, this creates a big opportunity for cutting-edge retailers to get ahead of the competition and get aggressive with this program (and smart) and take some serious share.

Are you seeing your GPS traffic drop off a cliff? Have you seen any wacky PLA examples? Sound off in comments.


SeekingAlpha Disclosure – I am long Google and Amazon. eBay is an investor in ChannelAdvisor where I am CEO.