Joe Lamantia.com

A recent ques­tion on the mail­ing list for the Tax­on­omy Com­mu­nity of Prac­tice asked about search ven­dors whose prod­ucts han­dle faceted nav­i­ga­tion, and men­tioned Endeca. Because ven­dor mar­ket­ing dis­torts the mean­ing of accepted terms too often, it’s worth point­ing out that Endeca’s tools dif­fer from faceted nav­i­ga­tion and orga­ni­za­tion sys­tems in a num­ber of key ways. These dif­fer­ences should affect strat­egy and pur­chase deci­sions on the best approach to pro­vid­ing high qual­ity search expe­ri­ences for users.
The Endeca model is based on Guided Nav­i­ga­tion, a prod­uct con­cept that blends ele­ments of user expe­ri­ence, admin­is­tra­tion, func­tion­al­ity, and pos­si­ble infor­ma­tion struc­tures. In prac­tice, guided nav­i­ga­tion feels sim­i­lar to facets, in that sets of results are nar­rowed or fil­tered by suc­ces­sive choices from avail­able attrib­utes (Endeca calls them dimen­sions).
But at heart, Endeca’s approach is dif­fer­ent in key ways.

• Facets are orthog­o­nal, whereas Endeca’s dimen­sions can overlap.
• Facets are ubiq­ui­tous, so always apply, whereas Endeca’s dimen­sions can be con­di­tional, some­times apply­ing and some­times not.
• Facets reflect a fun­da­men­tal char­ac­ter­is­tic or aspect of the pool of items. Endeca’s Dimen­sions may reflect some aspect of the pool of items (pri­mary prop­er­ties), they may be inferred (sec­ondary prop­er­ties), they may be out­side cri­te­ria, etc.
• The val­ues pos­si­ble for a indi­vid­ual facet are flat and equiv­a­lent. Endeca’s dimen­sions can con­tain var­i­ous kinds of struc­tures (unless I’m mis­taken), and may not be equivalent.

In terms of appli­ca­tion to var­i­ous kinds of busi­ness needs and user expe­ri­ences, facets can offer great power and util­ity for quickly iden­ti­fy­ing and manip­u­lat­ing large num­bers of sim­i­lar or sym­met­ri­cal items, typ­i­cally in nar­rower domains. Endeca’s guided nav­i­ga­tion is well suited to broader domains (though there is still a sin­gle root at the base of the tree), with fuzzier struc­tures than facets.
Oper­a­tively, facets often don’t serve well as a uni­fy­ing solu­tion to the need for pro­vid­ing struc­ture and access to het­ero­ge­neous col­lec­tions, and can encounter scal­ing dif­fi­cul­ties when used for homoge­nous col­lec­tions. Faceted expe­ri­ences can offer gen­uine bidi­rec­tional nav­i­ga­tion for users, mean­ing they work equally well for nav­i­ga­tion paths that expand item sets from a sin­gle item to larger col­lec­tions of sim­i­lar items, because of the sym­me­try built in to faceted sys­tems.
Guided nav­i­ga­tion is bet­ter able to han­dle het­ero­ge­neous col­lec­tions, but is not as pre­cise for iden­ti­fi­ca­tion, does not reflect struc­ture, and requires atten­tion to cor­rectly define (in ways not con­fus­ing / con­flict­ing) and man­age over time. Endeca’s dimen­sions do not offer bidi­rec­tional nav­i­ga­tion by default (because of their struc­tural dif­fer­ences — it is pos­si­ble to cre­ate user expe­ri­ences that sup­port bidi­rec­tional nav­i­ga­tion using Endeca).
In sum, these dif­fer­ences should help explain the pop­u­lar­ity of Endeca in ecom­merce con­texts, where every archi­tec­tural incen­tive (even those that may not align with user goals) to increas­ing the total value of cus­tomer pur­chases is sig­nif­i­cant, and the rel­e­vance of facets to search­ing and infor­ma­tion retrieval expe­ri­ences that sup­port a broader set of user goals within nar­rower infor­ma­tion domains.