Search Recommendation Engine

 

   Overview : At certain instances in user journey it becomes vital to intuitively ‘Suggest’ the users about possible steps they could take to experience the product in more suitable way. In context of housing.com it was observed that many users faced two kind of issues, the broad search and the narrow search. Under this project solutions were conceptualised for ‘suggestions’  which should be made to the users to persuade them to convert their narrow or broad search to normal search.   

Purpose : For solutions team,  Housing.com
 

Duration : Oct'15 - Nov'15 

Credits : Sayanee Halder for helping in final visuals

BROAD SEARCH
1

Use CaseUser searches in parent locality ( big localities which contain multiple sub localities) and see overwhelming no. of listings 

Solution : Shift to sub-locality within a parent locality. Here a new concept of 'spotlight'  (purple pins on the map) is introduced. By hovering over spot light pin on the map user can see the comparative price index with respect to parent locality and the no. of listings in sub-locality. Quick access to key information helps user to make an informed choice 

2

Use Case - Budget Filter is not applied by the user which is very imporatnt to show adequate no. of results.

Solution : On the basis of user past search analysis (i.e. user profile) and polygon profile (i.e. locality profile) it is possible to guess a budget range user might choose and therefore could be the by default budget range suggestion for the users. It could reduce their effort of selecting a range. Secondly in the proposed solution when a user adjusts a  budget range she is aware of the no. of listings she could expect.

Apply this budget to see

listings in Powai

NARROW SEARCH
3

Use Case - User searches in a sub locality or parent locality which has few or no listings

Solution : One possible suggestion for the user could be to search in nearby localities. It is important for the users to quickly access 'some key information' about the locality helping them in deciding the next locality she could pick. This piece of information is price point comparison and no. of listings. Here there are two possible scenarios :-
1. BHK type is selected - then the price is shown as a definite range. For ex. Rs.10K  to Rs. 15K for 1 BHK
2. BHK type is not selected -In this case, the price is shown as how expensive/cheap a locality is as compare to other locality. 

Problem 

Solution

5

Use Case - User has applied multiple soft filters (filters other then BHK type, budget and furnishing type ). This reduce the no. of listings matching the criterion. It was found through user research that generally these filters are not deal breakers but the 'good to have' amenities in the house. users are willing to see listings not strictly matching the criterion.

Problem

Issue : For user it is difficult to understand  to whether the filters applied by her are removed or user has to remove them. Secondly it is not important to show all the soft filters.

Solution 1 : With the help of existing data a  'preference order' of most important soft filters for general users was created. As per this order a user is most likely to remove microwave as a filter and least likely to remove refrigerator or washingmachine. User is persuaded to remove one or the more soft filters

Applicable when : User has applied more then two soft filters. Instead of showing all the filters user would be only show maximum of three filters low in preference order.

Solution 2 : Instead of a user is asked to remove a filter to see more results, flats which doesn't match the criterion defined by the filters are also shown along with the listings strictly matching the criterion. As per the user study for ex. if a user has applied microwave as a filters he/she also wants to see listings without microwave as it's not a major deciding facor.

Applicable when : User has applied only one soft filter and there are significant no. of listings available if user  removes the applied filter.

4

Use Case - User searches in a small sub locality which has few or no listings

Solution :If the nearby localities also have very few listings then a clear suggestion is made to search in parent locality.