The second day of #SESCHI is in full swing, and the Panda vs. Human: Advanced eCommerce SEO & UX, featuring speakers Greg Nudelman, CTO of Design Caffeine and author of Designing Search (@DesignCaffeine) Inc and Jaimie Sirovich, CTO of SEO Egghead Inc (@SEOEgghead) tackled some seriously deep tactics for pleasing the Panda without compromising user experience.
Greg and Jaimie tag-teamed the session discussing how what is good for Panda can be good for people… but not always. Panda is your visitor, too, and these guys consider that furry black and white critter an important demographic. Read on for the tasty takeaways.
Sliders: the Usability Challenge
The User Side
You can pop them on the page & they look good… or do they? Often, a drill-down or box-select work better for user experience and the Panda. Sliders fail when trying to find things around the range.
Sliders create ineffective system states. For most people, 1-2 stars are pretty darn useless (eg: Hotel reviews) most users don’t even want to go there.
Where there is a large range of price with consumers looking at a tight budget, the user experience is poor, as the range buttons are right on top of each other and users cannot easily identify how many option they have within that range.
Prisjakt, a Swedish eCommerce site, is implementing something new: a histogram + slider! Leave it to the Sweeds to come up with something really cool.
The really big elephant in the room: accessibility.
Colorblindness is something marketers must keep in mind. Some fancy interactivity may have issues.
The Panda Side:
Do nothing, see nothing. Sometimes this is best because some data is not useful (price). Instead create a hierarchy of values based on string prefixes, numerical ranges, etc.
- Does not for work for “true” Aristotelian trees
- These are often tested at Wal-Mart (and they go between switching them on & off quickly)
- Jaimie has implemented a tweaked version
When it comes to making a decision: K.I.S.S. …!
Incorporating all the solutions is like shooting yourself in the foot. When users can see the inventory within the ranges it’s quite powerful.
Facet-Enabled-Landing Pages: Where’s the Beef?
It can be difficult when users cannot quickly find products on landing pages which creates a poor user experience.
Take for example a “Harry Potter” search on Amazon. Amazon has all products, then there’s a tile of lists, which users really like.
Considering the Panda:
Pandas like to see real content (a LOT). Solution: do both, products and content.
Create facet-enabled (FE) landing page:
- Add content at the top for Pandas and humans
- Not just categories, for categories + facet-filters
- Eg: “material: leather” + sofas
A few eCommerce platforms (adeptCommerce, Endeca etc) support FE landing pages out of the box. These are really important for PPC and SEO.
Tangent: Don’t make me think DRILL.
- Modern FE design leads to shallower less-detailed/deep category tree.
Often, this forces users to make ambiguous and or frustrating subcategory decisions. Humans love facet categories, but they’re very bad for robots and Panda.
Now, many subcategories are turned into facets and facets used for details in any application order and combo are a huge spider trap and bad for Panda.
Superfluous Content (Not really duplicate content, but almost…)
Useless content consists of unique questions that no one cares about (yes, Virginia), there ARE stupid questions. Eg: search: “cameras & red or blue” vs. “cameras in red” … or “cameras in blue”
Why is this a problem? Humans have intent: “I want a blue shirt,” and many URL parameters are useless to pandas, just as some are also useless to humans.
Google’s New Parameter Tool
New addition to Web Master Tools:
- Indicate to WM which URL parameters filter content
- Still does not indicate which are more useful
- What about facets that used to be subcategories
- Does not work if you’re an avid re-writer
Same Content, Multiple Access Points
- Duplicate controls
- No way to be sure you’ve found everything
- Confusing and unsatisfying if you are trying to repeat the same search task.
The URLs are also different, though the product set is the same, which makes Panda sad, and makes for a poor user experience when trying to recreate a search.
Panda and humans agree: Too many ways to get to the same thing is confusing for people & pandas. It truly is duplicate content.
Rel=Canonical is Not Your Solution
Panda does not appreciate thin repetitive content, and though some may think so, Rel=Canonical is not your SEO messiah.
It’s probably not a good idea to leave it up to Google to figure out these different URL strings.
Good Solutions: Present states for selections in a consistent order within a pages URL, this eliminates really ugly spider traps.
- Eliminate the number of selections
- Exclude multiple-disjunctive-select URLS
- Whitelist exceptions (especially landing pages with added content)
But, really, is this anything new? Nope.
This is the same SEO your mom did. SEOs have done this since hotbot & altavista etc.
[editors note: not really my mom]
- Hide everything ( you’ll miss out on long-tail traffic)
- Keep spiders out
- Just use AJAX
- AJAX does not create index-able pages
- Same problem
- #! might work/help, but will not save you
Variation on Same Idea:
- Selective robots.tx
- Selectively include faceted pages
Why Not Use Keyword Search Pages?
- Search doesn’t allow for easy exploration, because there are no sibling values
- Noise: polysemy is a problem
- Google (officially) frowns upon this
- Less clear when adding content/value. (But why chance it?)
Facets: Single or Multi-Select
We used to unconditionally recommend single-select/drill down, now, multi-select pretty much rulez. Caveat: you really must know what you’re doing.
Decide if you’re going to stick with drill down or parallel selection. Provide an obvious and consistent way to undo selections. Lastly, always make ALL filters easily accessible.
What about Panda?
Multi-select & SEO is even worse than single-select “standard” facet. Remove and get rid of it. On very large sites with multi-select the silly combo-space is even larger. Robots are fast, but cannot index terabytes of data.
Watch Your Logs! (Not just analytics)
- Don’t be an armchair SEO (theorizing all day)
- This is not in your analytics
- It’s very easy to miss spider traps
- You must look at your logs
- You’ll find stuff you missed (trust me)
But… Search & Navigation is SO COMPLICATED! Why not use SaaS search?
Not so fast!
SaaS isn’t so SaaSy sometimes:
- Only works off search box
- If users lands on a category page, no advanced navigation
- They must restart from search box
- Content on subdomain
Certain vendors have solutions:
- adeptCommerce Hosted Search platform (in Beta) has solved this
- Category pages with grafted in facet filters
- Content should not live on a subdomain (you really don’t want this)
- Searchspring has a solution for this as well
NEW: Facet-Enabled Breadcrumbs
Browse & search work in tandem on the best sites. Integrated facet-breadcrumbs (IFB):
- Integrates facets into breadcrumb
- Maintains hierarchical order
- Labels breadcrumb aspects (this is essential if users are coming from search space eg: Google)
- Direct keyword manipulation
- Makes pivoting easy
MAYA: most advanced yet acceptable
Faceted-breadcrumbs hold the promise to become the key component in the next-gen of intuitive finding interfaces that fully integrate the best of faceted search and browser capabilities.
If you want to implement IFB or similar, you must think about Panda:
- Unless care is taken this will now create a new spider trap on all pages
- Once again, rel=canonical will not work
- Jr.com implements session based
- Pandas and humans do agree on a few things
- When they disagree, pay attention!
- Don’t forget user experience for humans AND pandas
- Panda does require some consideration as a “demographic”
Big thanks to the speakers for an amazing session. Stay tuned to @merrymorud @ericasendros and @meganlichty for more live coverage from #SESCHI, and keep a finger on the pulse of aimClear blog for more posts from the event!