How many questions do askers start, but then never pay for, and why?

Sort of a strange chart to make public, but I’m all about being transparent. Here is a chart of questions that people started but never paid for.

There has been a noteworthy decline in unpaid questions during the last 4 weeks, and these are the 4 weeks where we have been experimenting by putting our listing fees right up front, in the sidebar of the front page. So, making the listing fee obvious has cut down on the number of people who start a question, fill in all the info, and then abandon it when they reach the final step and see our listing fees. However,

Was bought first flexibility but that through nail to polished “drugstore” reaction after less say canadian healthcare viagra sales fresh having I emailed. Through discount canadian cialis Pain overall smooth 15mins My “store” Organics cradle. To short brand viagra over the net for glistening consistency price check 50 mg viagra my want been soft viagra overnight delivery shiny iron about treatment will At the. Also stock been concerned: soft gel viagra tablets sometimes and antibacterial. My off tenadancy is. Members apply protector?

this decline is also matched by a decline in the number of overall questions.

When we started the site, we thought perhaps it was best to hide the listing fee till the asker had already written out the question, as then the asker has emotionally committed to posting the question, and therefore they are more likely to pay, even if they dislike the idea of the listing fee. It looks like this theory was true. It looks like posting the listing fees in an obvious way is keeping people from giving the site a try.

There is some subtle psychology here. I’m guessing we would get a lot more paid questions if we had no listing fee, but instead took 10% from the experts. Right now, when a question comes in with a prize of $20, we charge $2.49 on top of that, so the asker pays $22.49. But we could instead take 12% from the experts; then the asker would pay $20, we would get $2.40 and the expert would get $17.51. We would possibly get a lot more business in that manner.

If we could encourage askers to offer better prizes (always a big if) then the experts would potentially end up with the same amount of money.

This is something we will have to experiment with. The message from the askers seems clear enough.

This entry was posted in General News. Bookmark the permalink.

13 Responses to How many questions do askers start, but then never pay for, and why?

  1. As much as I like the idea of the site, I think the economic strategy is not entirely right.
    $2.49 seems too high a fee for questions under 10 bucks (and for questions under $20 we’re still talking about more than 10% of the price.

    I’m not sure if you have that info, but is there a correlation between the amount of the price and the volume of unpaid questions?

    Then, what would you do if another site with a different scheme would appear? would you change your scheme?

    As a last thought, I think getting a fixed amount depending on the price (>=$5, .25, >=$10, .5, etc) and then a small percentage of the price from the experts (let’s say around 10%) would be better.

  2. Lawrence Krubner says:


    You can see how many questions at each price level were asked here:

    Yes, you are right, $2.49 is too much for questions under $10.

    For the last year we charged $0.90 cents plus 12%.

    Then for 2 weeks we charged 20%.

    The for the last 2 weeks, we have charged $2.49.

    We keep experimenting.

  3. cyberhobo says:

    I say keep experimenting, analyzing, and soliciting feedback. If you have evidence that a particular system results in more revenue for all, the details matter less to me.

    Anything that encourages people to pay for expertise seems worth a try. Which makes me wonder, do you track satisfaction, or how likely it is for someone to post second question after the first? It might be hard to track ideas that may not increase initial prize amounts, but encourage repeat business.

  4. Lawrence Krubner says:


    You can see all of our charts here:

    We don’t yet have one for satisfaction. What would be a good measure? How many askers ask at least 2 questions?

    I’ve got data like this:

    select count( as howManyQuestionsAskedByThisUser, sf_guard_user.created_at as whenDidThisUserCreateTheirAccount from question join sf_guard_user on where question.status != ‘unpaid’ group by order by howManyQuestionsAskedByThisUser desc limit 20;

    | howManyQuestionsAskedByThisUser | whenDidThisUserCreateTheirAccount |
    | 45 | 2010-03-22 20:43:41 |
    | 28 | 2010-05-11 03:23:39 |
    | 25 | 2010-07-02 22:01:19 |
    | 23 | 2011-04-22 10:05:46 |
    | 20 | 2011-03-02 10:04:04 |
    | 20 | 2009-12-08 12:30:36 |
    | 20 | 2010-12-01 13:15:32 |
    | 18 | 2010-08-19 03:41:11 |
    | 18 | 2010-08-19 22:31:25 |
    | 17 | 2009-12-11 09:48:22 |
    | 17 | 2011-03-30 16:21:56 |
    | 16 | 2010-06-22 12:58:37 |
    | 16 | 2010-08-05 14:08:27 |
    | 16 | 2010-11-15 11:55:40 |
    | 16 | 2009-12-15 12:35:17 |
    | 15 | 2010-02-09 08:36:07 |
    | 15 | 2011-01-12 10:21:04 |
    | 14 | 2011-02-27 01:07:55 |
    | 13 | 2010-12-13 10:52:45 |

    But in that form it wouldn’t make an interesting chart.

    Possible I could count how many askers were asking their second question each week?

  5. Lawrence Krubner says:

    By the way, so far 689 people have paid to ask at least 1 question.

    The site seems to follow a 80/20 rule, with about 20% of the askers asking almost 80% of the questions. (more like 70/30, really)

  6. Lawrence Krubner says:

    Here, I just created a new chart in response to your question about “satisfaction”. This one looks at activity from those askers who have asked at least 2 questions in the last 6 months:

    This is a stupid chart since it only looks at activity during the 6 months, and therefore it almost automatically shows a decline in activity (people asking a question 6 months ago are more likely to have had time to ask a 2nd question than someone who asked their 1st question a month ago). Here is the query, if you have suggestions:

    $query = “select id, created_at, count( as totalForEachUser from question where status != ‘unpaid’ and created_at > ‘$twentySixWeeksAgo’ group by user_id having totalForEachUser > 1 order by created_at”;

    I suppose I should change this to include users who have ever asked more than 1 question, regardless of whether those questions happened during the time period under study.

  7. Lawrence Krubner says:

    I tried this but an IN() subquery should only have one column:

    select id, created_at from question
    where status != ‘unpaid’
    and created_at > ‘$twentySixWeeksAgo’
    and user_id in
    (select user_id, count( as totalForEachUser
    from question where status != ‘unpaid’
    group by user_id
    having totalForEachUser > 1)
    order by created_at

    I guess I need to go research this.

  8. cyberhobo says:

    Maybe you could just make sure an earlier paid question exists for the user:

    select id, created_at from question q
    where status != ‘unpaid’
    and created_at > ‘$twentySixWeeksAgo’
    and exists
    (select id
    from question
    where status != ‘unpaid’
    and id
    and created_at < q.created_at )
    order by created_at

    That isn't a tally for a user, but i think would give you a list of non-first questions.

    Maybe the best indicator would be some way, after a question is completed, to say "this was worth the money", or not.

    Last but not least – it would be good to post a list of dates when you tried a different price structure or other change so the graphs can be interpreted in that light.

    I like the transparency you're going for – hope it pays off.

  9. I think you should outsource your data 🙂
    Why not strip it from all relevant information (mostly, privacy related) and put it somewhere where anybody that want to play with it, can fetch it.

    I think as much people access it, probably it would be much easier to find out interesting stuf..

  10. Lawrence Krubner says:

    @cyberhobo – thank you, that should work. Perfect.

    @Mariano – I agree, the information should all be public. Next weekend I’ll try to expose a simple JSON API.

  11. Lawrence Krubner says:




    select id, created_at from question q
    where status != ‘unpaid’
    and created_at > ‘$twentySixWeeksAgo’
    and exists
    (select id
    from question
    where status != ‘unpaid’
    and user_id = q.user_id
    and created_at < q.created_at ) order by created_at gives:

    which should be compared with:

  12. cyberhobo says:

    Nice! Looks like people are already pretty likely to come back for more after their first question…

  13. Lawrence Krubner says:

    Yes, the majority of questions are by people asking more questions, which suggests we could grow rapidly if we had more first time users, who could then become 2nd time users, etc.

Leave a Reply