Input for Kiva staff on the "Popularity algorithm"

[Guest]

.

[alan]

Actually, it shouldn't be "New and Popular" but "New or Popular" because "new" trumps "popular". If there is anything new, it gets showcased. Otherwise, popular begins to show up.

[YowieFreak]

Can we find out from them why they chose to continue to include "newness" in their calculation of "popularity"?  Hopefully they will have a valid reason.  (Other than "if we took it out, Matt would be unhappy because loans wouldn't expire".)

[AccountAbility]

And just when I was beginning to think that Kiva really wanted to listen-- as the title of this thread implied.   

Ample examples were given on this forum demonstrating that their own formula actually causes some loans to fund and others to expire.  They still want "new" to muck up the sort. 

Dan

[Jeremy]

It seems to me like the summary of this thread is that Kiva has two conflicting goals:

1. To encourage visitors to the Kiva site to become lenders.
2. To fund as many existing loans as possible.

Both of these goals are part of Kiva's (and our) overreaching goal of making more money available to third-world borrowers. As I see it, Kiva believe that #1 is a bigger problem in reaching our overarching goal, while the Kiva Friends generally see #2 as a bigger problem. While I tend to agree with Kiva that #1 is the bigger problem (as there are many people who visit kiva.org without lending, but we have yet to see a loan expire), I think that goals 1 and 2 aren't mutually exclusive.

What if Kiva changed the default sort so that users who weren't logged in would see the Popular sort, and users who were logged in saw the Ending soon sort? If Kiva is correct that the popular sort increases new lender conversion (perhaps an A/B test could be used to test this assertion), then they could keep the Popular sort, but also encourage users with accounts to fund the loans that are in more risk of expiring.

[alan]

It seems to me like the summary of this thread is that Kiva has two conflicting goals:

1. To encourage visitors to the Kiva site to become lenders.
2. To fund as many existing loans as possible.

Both of these goals are part of Kiva's (and our) overreaching goal of making more money available to third-world borrowers. As I see it, Kiva believe that #1 is a bigger problem in reaching our overarching goal, while the Kiva Friends generally see #2 as a bigger problem. While I tend to agree with Kiva that #1 is the bigger problem (as there are many people who visit kiva.org without lending, but we have yet to see a loan expire), I think that goals 1 and 2 aren't mutually exclusive.

We have seen 14 loans expire: 2 in the first 4 years of Kiva's existence, and 12 this month.

See here for a list.

[alan]

Can we find out from them why they chose to continue to include "newness" in their calculation of "popularity"?  Hopefully they will have a valid reason.  (Other than "if we took it out, Matt would be unhappy because loans wouldn't expire".)

I imagine they've decided to tinker a bit in advance of making any wholesale changes based no this thread. It's sensible, actually, simply to eliminate 1 of many factors to see what the effect is. Minimal, in this case, I think.

To be fair, they haven't said they aren't making further changes, just that this is the one they're making now.

[alan]

When 300 loans are added, none are popular, all are new. What is the sort order then ?

You can always stick in a secondary sort of old-to-new, so if two loans have the same popularity ranking the older one shows first. In my proposal, new loans have demonstrated no popularity, so they go to the end of the list.

[JoanW]

It seems to me like the summary of this thread is that Kiva has two conflicting goals:

1. To encourage visitors to the Kiva site to become lenders.
2. To fund as many existing loans as possible.

Both of these goals are part of Kiva's (and our) overreaching goal of making more money available to third-world borrowers. As I see it, Kiva believe that #1 is a bigger problem in reaching our overarching goal, while the Kiva Friends generally see #2 as a bigger problem. While I tend to agree with Kiva that #1 is the bigger problem (as there are many people who visit kiva.org without lending, but we have yet to see a loan expire), I think that goals 1 and 2 aren't mutually exclusive.

I agree with you on this (except that there have been loans go unfunded) - although I think if it was not for the devoted efforts of a group who work really hard, we would see a lot more loans actually go unfunded.

What if Kiva changed the default sort so that users who weren't logged in would see the Popular sort, and users who were logged in saw the Ending soon sort? If Kiva is correct that the popular sort increases new lender conversion (perhaps an A/B test could be used to test this assertion), then they could keep the Popular sort, but also encourage users with accounts to fund the loans that are in more risk of expiring.

I'm reading this...and just finished reading the other thread about converting web hits...Apparently the current method is NOT converting lenders.

Remember the Magic Formula For Success?
1) What do I want?
2) What am I doing to get it?
3) Is it working? - if "No", then you must change  either #1 or #2

So that bring us to -
1) I (Kiva) want to convert browsers into Lenders
2) I (Kiva) am showing them a "Popular" assortment of loans
3) It's not working as well as  I (Kiva) expected it to -   I (Kiva) need to change either #1 or #2

It really is *that* simple.

[Jeremy]

First, I apologize for not realizing that we had some unfunded loans - don't know how I missed that.

Second, I totally agree that the assertion that the popular sort is effective in conversions may not be true. I don't know what sort of testing Kiva has done, but it certainly makes sense to start doing some A/B testing with different sorts as the default, or even with different pages that introduce Kiva to new lenders.

[David2051]

It seems to me like the summary of this thread is that Kiva has two conflicting goals:

1. To encourage visitors to the Kiva site to become lenders.
2. To fund as many existing loans as possible.

I don't see the conflict here.  Encouraging site visitors to become lenders, and hopefully core lenders at that, helps to fund as many loans as possible.  I don't see anything about funding as many loans as possible that would discourage visitors from being lenders.  It seems to me that these two goals reinforce each other.

So why is the conversion rate low?  Maybe any combination of the following...

• visitors don't understand how it works
• the explanations provided are not clear and concise
• the economy is bad and they don't have the money
• they don't understand or trust PayPal
• they don't trust Kiva with their money
• maybe they understand perfectly but Kiva risk, MFI risk, borrower risk, currency risk, etc. seems too much to tolerate
• any number of other things???

I've been on some retail sites where if you try to leave the page a window pops up that makes some last ditch offer.  Maybe a little survey should be offered for people to indicate why they aren't lending?  (groan...)