Pre-Launch Testing of Catalogs on the Web

You can pre-test items for an upcoming catalog, without having to print the pages or stock the merchandise, by asking customers to view images from the catalog on a website.

Though simple in concept, catalog pre-tests need attention to subtleties in both design and interpretation to get reliable results.  Research Boston's process automates most of those details for you.  For far less than the cost of a preview mailing, you can have an advance sounding of the market for your catalog's items -- gaining a month or more of extra time to adjust stocks.  

Example webtest pages
(Move your mouse around the page to see the item descriptions.)

Example report of webtest results

Benefits

• Improve fulfillment.  Provide better customer service.
• Capture sales currently lost to stockouts.  Some clients have estimated that 10% or more of the demand they've already created often evaporates because of stockouts.
• Cut backorder costs.
• Reduce firefighting, friction with suppliers, and disappointed customers when an item "runs away".
• Get an early jump on canceling orders on "dogs", and dropping them from later books.

Will webtests fit your situation?  Consider these factors...

• If at all possible, test the page layouts and images that customers will actually see in the mailed catalog.  Tests based on the "thumbnail" pictures typical of many websites are less accurate, for predicting a mailed catalog, than tests of the actual pages.
  
• The best combination of response rate, time schedule, and predictive power usually comes from "mock" tests.  That is, customers know they aren't actually buying anything.  Instead,  they are compensated for their efforts by cash, discount coupons, or similar incentives.
  
• "Live" tests, where customers actually do buy the items for future delivery, will often work but their design and interpretation tend to be more complex.
  
• Raw customer responses must be adjusted for out-of-season effects, price sensitivity, competing items, and similar factors.  Plan on at least one "calibration run" for mock webtests, and 2 or 3 for live tests, to tune the adjustment factors.
  
• Once the system is setup and calibrated, a webtest takes about 2 weeks from the time you provide images until you get final forecasts.  (Raw tallies are reported daily.)  Surprise items, "runaways" and "dogs", usually become apparent in a test's first few days.
  
• Accuracy of the webtests varies with many factors, but as a rule of thumb, forecast errors average about half those of buyers' judgmental ("manual") forecasts.