Google Shopping goes paid: Product Listing Ads for retailers

Google Product Search has been a free comparison engine since it launched as Froogle. On May 31, Google announced it will become Google Shopping, a commercial program where every listing is a Product Listing Ad paid for through AdWords. Google's stated reason is data quality: merchants who pay will keep feeds accurate. The practical effect is that one of the largest free sources of ecommerce traffic is becoming a paid channel.

Timeline for US merchants as it stands:

• Now through the summer: Product Search results begin to be replaced by PLAs. Both appear during the transition.
• August 15: deadline to have PLAs running to qualify for Google's early adopter credit. Merchants who create PLAs by then receive a 10% monthly credit on PLA spend through the end of the year, and existing Product Search merchants also get a $100 AdWords credit.
• Fall 2012: transition complete. No free listings remain.

Amazon has already pulled its products from Google Shopping rather than pay. Expect less competition from the largest marketplace in some categories and more from mid-sized retailers who are moving budget here.

A PLA shows a product image, title, price and your store name, either in a block on the main search results page or in the Shopping tab. Unlike text ads there are no keywords. Google matches the search to your product data, and you control what shows through two things:

1. The Google Merchant Center feed. A spreadsheet or XML file listing every product with required attributes: id, title, description, link, image link, price, condition, availability, brand, and for most categories a GTIN (UPC in the US) or MPN. Google product category and your own product type are also expected.
2. The AdWords campaign. A campaign using the Product Listing Ads extension, with ad groups containing product targets. A product target is a filter on your feed, such as brand = X or a custom label like adwords_labels = clearance, with a bid attached.

You bid either cost per click or, if you qualify, cost per acquisition, where Google takes a percentage of the revenue from the sale instead of charging per click. CPA bidding needs conversion tracking and a clean checkout, and Google has been selective about who gets it.

Most retailers who have had products in Product Search for years have a feed that was set up once and forgotten. Under the paid program, every feed error is now a wasted click or a disapproved product. Go through this checklist before touching the campaign.

• Titles that describe the product. Brand, product name, key attribute such as size or color, in the first 70 characters. "Blue Widget" loses to "Acme Blue Widget 12-inch Stainless Steel" every time.
• Unique identifiers on every item. UPC, brand and MPN. Google requires at least two of the three for most products. Missing identifiers mean disapproval or poor matching.
• Accurate price and availability. The ad shows your feed price; the landing page must match. Price mismatches are the most common reason for suspension. Submit the feed daily, or more often if prices move.
• Images that pass review. White or neutral background, no watermarks, no promotional text, at least 250 by 250 pixels. Images with text overlays are being rejected.
• Google product category. Assign the most specific category from Google's taxonomy. It affects matching more than any other field.
• Custom labels for bidding. Add adwords_labels and adwords_grouping columns so you can bid differently on margin tiers, seasonal items or bestsellers.

Feed management tools such as GoDataFeed, SingleFeed and ChannelAdvisor charge $50 to several hundred dollars a month and are worth it above a few thousand SKUs. Below that, a well-built spreadsheet and a scheduled upload usually suffice.

Start simple and segment later. A reasonable first structure for a mid-sized US retailer:

Rules that keep this from going wrong:

• Keep the "All products" group as a catch-all, bid low, so nothing is left out.
• Use promotional text, the 45-character line under the ad, for free shipping or a sale, not generic slogans.
• Add negative keywords. PLAs have no keywords, but negatives still apply and they are the only way to stop your products showing for irrelevant searches.
• Set a separate daily budget from your text campaigns so one cannot starve the other.
• Turn on conversion tracking in AdWords and link Google Analytics before spending a dollar.

Early reporting from retailers running both suggests PLAs deliver click-through rates two to three times text ads for the same products and cost per click 20% to 40% lower. That advantage will shrink as more merchants bid, which is why starting before the fall matters.

Look at your Google Analytics referral traffic from google.com/products and shopping.google.com over the last 12 months. That is the free traffic you are about to lose. Multiply those visits by $0.40 to get a rough annual cost to replace them with PLAs at current prices. For most small retailers this is a few hundred to a few thousand dollars a month, which usually compares well against the revenue those visits produced.

Then decide how to fund it. Common moves this summer:

• Shift budget from underperforming comparison shopping engines. Nextag, PriceGrabber and Shopzilla typically charge $0.30 to $1.00 per click with lower conversion rates than Google.
• Trim text ad spend on product-name keywords where the PLA now takes the click.
• Pause the daily deal experiments that have not paid back.

Watch margin, not just revenue. The CPA bidding option is attractive for low-margin categories precisely because you cap the cost as a percentage of the sale.

Waiting until October. The August 15 credit deadline is the real one. After that you pay full price and compete against everyone who moved early.

Bidding on everything equally. A single product target with one bid treats a $12 accessory and a $900 appliance the same.

Ignoring disapprovals. Merchant Center emails disapproval notices. Most go to an inbox nobody reads. Assign an owner.

Sending clicks to a category page. The feed link must go to the exact product page. Category links are a policy violation and convert poorly anyway.

Forgetting Bing. Bing Shopping is still free for now. Keep feeding it while that lasts.

Our method is discover, plan, build, improve. Discovery means auditing the Merchant Center account, the feed and the Analytics history to see what Product Search has been delivering and where the feed is weak. Planning sets the campaign structure, the margin-based bidding tiers, the budget derived from the traffic you are replacing and the targets we will manage against. Build covers the feed fixes, usually in the store platform itself rather than in a spreadsheet, the AdWords campaign, conversion tracking and the daily upload schedule. Improve is the weekly cycle of bid changes, negative keywords and feed refinements once data arrives.

Because our ecommerce, performance and web teams work together, feed problems that originate in the product catalog get fixed at the source: better titles, complete UPCs, clean images and product pages that load fast on the phone a shopper is holding in a store aisle.

A representative engagement is on our ecommerce work page. An online retailer with several thousand SKUs had relied on free Product Search traffic and had a feed full of missing identifiers and generic titles. We rebuilt the feed from the catalog, added margin labels, launched a tiered PLA campaign alongside the existing text ads and set up tracking so the two could be compared. Paid Shopping traffic replaced the lost free visits within weeks and did so at a cost per order below the retailer's text ads. The pattern holds for most catalogs: the feed does the heavy lifting.