Hacker News new | comments | ask | show | jobs | submit login
Ask HN: Please review my site - PickFu.com (pickfu.com)
104 points by justinchen on June 3, 2009 | hide | past | web | favorite | 68 comments

Looks great. I think the price is right. The website is very clear. Maybe the examples should be more business oriented. You should also include the time frame to get the 50 answers: 1 hour, 1 day, 1 week?

Can we choose, or at least influence the demography (only women, only people below 35, etc.)?

Thanks. Good point about more business-oriented examples. The time frame varies since it depends on the people answering the question, but it's usually within a couple hours - good idea to note that on the site.

Unfortunately, can't choose demography but we try to break out the responses in a bunch of different ways so you can visually filter which demographic you're interested in.

Justin, I've been mucking around for awhile in MTurk trying to use Qualifications to create an survey that would target a specific demographic. Have you cracked this nut? Is this something you're considering adding to extend the product? Or is it simply not within the scope of MTurk's capabilities?

We looked at using Qualifications a bit, but quickly decided that it wasn't worth our time to deal with them, at least for PickFu in its current state.

I quite like that it's ultra-simple right now. 1 price, few options. Not something you need to think about too much. Just do it.

I'm not saying that choosing your own demographics, number of participants, more complex surveys might not be useful sometimes. But I love that I can get my head around this in 10 seconds.

another one of the creators here, thanks for the warm comments.

a follow-up question - how much more compelling would this be if you could choose/influence the demographics and # of responses (potentially at a higher cost)?

I would personally find it more compelling, but I'd suggest that it shouldn't be a premium feature.

When it comes to consume research, demographics are everything. Statistical significance of sub-samples can plummet quickly.

If I'm marketing a product specifically to, say, 30-44yo men, it would be valuable to restrict the sample exclusively to those men and I think i'd have a better chance of seeing statistical meaning in the sub-samples I'm looking for, eg, income, marital status, etc.

It seems to me now that I'd have to post it "at large" and the 30-44yo demo I'm looking for would itself be a sub-sample.

I am not a marketing professional, but as a small business owner its one of my many hats. I have used both Maritz and, more recently, Pinecone, and I'm probably wrong but I don't remember even being presented with an option to open the surveys to everybody.

And what I'm saying is.. I wouldn't even consider using your product without that ability. You have a fantastic 1.0, though, so I'm not knocking the product, just offering my $0.02.

And some demographics of your own: We've got a dozen of us on the payroll with 2009 sales looking to be a bit over $2MM.

And no joke, if I were you, I'd eat your dogfood and put up a survey to determine the importance of this feature!


I do want to add that I get that your site is supposed to be intro-level market research. But if you're going to limit to 50 replies, demographic targeting would be very important to me, for anything other than a "should we get thai or pizza for lunch" questions.

If you do let me collect 500-1000 replies, I'd probably have more statistically meaningful sub-samples, so as long as you give me a nice drill-down interface I'd find it useful, even if i couldn't target my question.

To take the other side...there is currently a beautiful simplicity about this idea and execution.

I pretty much just fall asleep when I start trying to deal with A/B testing, market research, ad effectiveness, etc. This is so simple that all I have to ask myself is, "Do I have a question I'd like to ask a bunch of people?" and "Do I have five dollars?" That's a really good place to be, because everybody in business has questions, and everybody has five dollars.

I will absolutely use this site. I do question where the people are coming from, which I guess is a demographic question...but I think any time you ask of the user to choose it should be later in the sales process. For statistics wonks, mention it in the FAQ, or something. But I wouldn't clutter up the initial offer with too much stuff like that. I wish I had a product that was this simple of a sell.

And, of course, the more involved things get, the more expensive it will have to be. I really like five dollar market research. I might not be so enthusiastic about $25 market research.

Fair enough... but if this is just a completely random, not-even-remotely-scientific man on the street approach, then what problem does this solve that a poll i can add to my website/blog/favorite forum via Fantastico in 180 seconds doesn't?

It puts it in front of people that don't already know about your product.

I can ask usability questions of my users, but their existing knowledge will effect what is "easy" and "intuitive" (intuitive generally means, "what I'm used to"). I can ask marketing questions of them, but I've already pulled them into my site, so my current message will be the one that gets reinforced. I can ask "compare these two products" questions, but I already know a preponderance of people on my site prefer my product to our competitors (one of which is Fantastico, by the way). I can ask any number of things, and we do frequently ask our customers opinions on things, but by virtue of them being on our site the results are known to be skewed in exactly the way we don't want them to be for a large number of questions.

Also, I don't want to clutter up my site with constant polls. When someone is on our site, we have very clear goals we want to achieve: Teach them about the benefits, help them choose the right products, show them how to use it, answer their questions. Asking them random questions isn't my idea of good design, when it comes to those goals.

Example: I am designing a logo for a market research site. I put up the pickfu logo against mine (or just use two of mine), & ask users which company does market research.

Examples 2-100: A poll on my site or blog would work fine but... I don't want people on my site to know what I'm polling. I don't have a site. I don't know how to do polls on my site. My IT department says it will take 8 weeks. My web design guys say they don't have the module for this particular CMS - they can build one from scratch for $2k. I told the web design guys to go to hell & found an embeddable polling service, it'll take 4 weeks & they won't be responsible if it doesn't work. I worked it out with the web design guys but my boss says we need to use the companies standard colour scheme - the widget won't change its colour.

I really like the site, and it is well executed. I think I'll certainly use it at some point. Good price point.

I completely agree that there needs to be some way to restrict who answers questions.

Another example would be if someone was making a Twitter application, they might want to restrict people who answer the question to only those that use twitter.

Or if you were doing something aimed at webmasters, you might want to restrict the responses to people employed as such.

As also mentioned in other comments, use your poll system to do your own poll on how desirable this could be.

I would be a little wary about reading into comments about this. Immediate thoughts brought on by a quick browse are often different from while-using wishlist items. There are also a wide range of possible uses for this. I am guessing mine is different from other people's.

I thought of using this as part of a sales process for consulting services. That would put you in some sort of competition (or symbiosis) with compete.com. For that, immediately obvious conclusions (however simple), low price & easy to use (time is an issue here) are all essential. precision, advanced options & such, not so much.

I'd pay twice as much to be able to target a specific demographic. Nice job John.

Hmm, would you pay more than twice as much to target a demographic segment that was less that 50% of the population?

Why does it matter how big the segment is?

It would require more effort to get 50 responses from a tiny demographic segment than a large one, so I was wondering if having the price inversely correlated to the segment size was reasonable to an end user.

I just used this site, and learned that 18-34-year-old women find a picture of me without a baby more attractive, while the 50+ demographic definitely prefers the picture of me holding my niece. Biologial clocks at work? Anyway, worth $5! ;) Well, sort of... I wish there was a way to view the votes by (gender,age-group) pair, but most of the respondents were female, so I think I'm interpreting the results correctly.

Also, I just emailed a friend about this and shared the PickFu URL for the results of my photo survey on facebook, which according to Derek Sivers is a great sign (http://sivers.org/purplecow).

Finally, note that you can see all the PickFu questions as they are distributed to Mechanical Turk (even the "private" ones, obviously) at


I hope your site can scale. I predict growth.

Actually, there is a way you can view the votes by (gender,age-group,etc.) - click on the "Option A vs. Option B" tab right under "Who responded to your question?" Is that what you were looking for? If so, it's our fault for not making it more visible.

Thanks for the heads-up about the mturk job search - we're making tweaks so that those previews show a sample question from now on. Helps keeps the responses more random too.

Thanks for sharing it!

No, "Option A vs Option B" isn't quite what I was looking for. Let us stipulate that, once I see the results, I decide that I don't care what guys think of my photo. I just want to compare the difference in response between 18-34-year-old women, 35-49-year-old-women, and 50+-year-old-women. I don't see a way to get that information from your current interface.

Ah yeah, I get what you're talking about. It's one of the features I've been wanting to add for a while too - filtering the responses by any demographic combination. Not pri 1, maybe in the next version. =)

I would be happy just to be able to access the raw responses in a .csv file or something, and do the filtering myself.

I may not be typical, but of all the "review my site" postings, both good and bad, this is the first one I actually plan to use. Pickfu.com is now a Favorite.

1 vote with $ = many votes without $

I think simplicity is the key here. It is going to be useful to masses and not constrained to just big companies. I personally used it as soon as I saw it and will be using it again and again.

My first thought: Why can't I answer the questions?

I know you're going to say something along the lines of "we want to ensure that users are unique, and demographically diverse, blah blah blah."

Yeah, I know, but still. When I see "Mac or PC", I instinctively click on the "Mac" option, and when nothing happens, I think "Well, THAT's clearly broken!"

I know you probably don't want to create just another hot-or-not site. But, since you have public commenting that's separate from the "official" responses, why not also have public voting that's segregated? Maybe if I give you my demographic info, you could let me answer questions.

The bonus is that you don't have to pay me ;) So, perhaps you could offer a few options: $5 for "valid" demographically-distributed statistically-significant mechanical turk answers, or you can get all the answers you want for free from the PickFu.com community (still organized by demographics.)

Good idea. We've heard similar feedback before but haven't had a chance to get around to it. We'll revisit it.

Reminds me of a design client of mine... http://www.feedbackarmy.com/

Similar concept but a bit more information from the Turks at a slightly higher cost.

A user friendly MTurk front end. Not a bad idea. I wish you the best of luck.

Micro-surveys. Cheap Price, Fast response. I could actually see using this in the future.

I misread it the first time and wondered what you had against Pi.

Great idea guys! If you have to time to come up to SF you should definitely drop by our Crowdsoucing Labor/Mechanical Turk meetup on June 10th. You can RSVP here: http://www.meetup.com/Amazon-Mechanical-Turk-Crowdsourcing-W...

The Amazon VP of Mechanical Turk is going to come by to answer questions, and I'm sure she would love to talk to you all.

Thanks Lukas, I plan on being there. Looking forward to meeting up - you guys are doing some awesome things with MTurk.

Rockin' idea. I was just looking for something like this to do some quick-and-dirty market research for my startup.

I don't have much experience with mturk, but my one concern would be that respondents may simply click randomly rather than honestly answering the question; do you have any data on how honest and accurate your responses are?

With any survey you always have the possibility of disingenuous answers, but we require them to put a comment/rationale for their answers, which is where the most interesting data lies. Here are some examples: http://pickfu.com/IOHSK8 http://pickfu.com/JWQFLA

Our respondents don't get paid unless they actually explain their answers - which both reduces false responses and provides unique value and insight to the service.

For example, we used this feature to dogfood our own FAQ page: http://pickfu.com/4UQM7U

I still think doesn't help. I liked the idea and did the turk for one of your questions (this was my first time doing the turk, btw). I was asked to visit a myspace webpage, listen entirely to 2 songs and say which one I like better and why. All this for 3 cents? I did it, cause I am a good fellow but I have hard time thinking of people who are willing to spend 10 minutes of their time for 3 cents.

the FAQ example is a bad example: I can always answer "yes" and then supply as reason "every questions I had was answered". No need to read the FAQ page, of course.

I think if you want to insert any kind of validation system you would have to count the time it took for the turk to answer your question and make sure we are in that range for every hit.

As noted before, it's tough to know that 100% of the responses are truthful on any survey. With PickFu, you can read the explanations to get a feel as to which respondents put some thought in.

I'm sorry you had a rough experience with that specific music-listening job - we've never had a more time-consuming question submitted. Most are in the order of "which site do you prefer?", like this current one: http://pickfu.com/AGSIBT. These get answered far faster by the MTurk workers.

You're basically going to get the same response quality as Amazon Turk. We use them a lot, but the quality is pretty average so there's still some amount of manual work or quality checking needed.

What would be truly amazing if this service could work with non-mechanical turkers...

"You can spend hours at a coffee shop polling 50 random people or bug everyone you know, but you'll be getting a skewed population"

Not to rain on your impressive parade, but I'd consider the type of person that wants or needs to use MTurk to be a pretty skewed population too, just in a different way.

Not saying it's not valuable, it may well be, and certainly more honest than asking friends - but I would expect a fairly low-income or bored-teen demographic.

Actually, an interesting meta-survey might be to survey the MTurk population itself, see what they think, their incomes, political beliefs etc; might be an interesting use of this very service.

But enough nit-picking - congratulations on a great idea, swiftly executed.

Another similar site:


Here is a controversial thought: Your price is too low. You tell me $5 for 50 comments, and I think, these are low quality comments.

You're underselling yourself IMHO. Position your service as a more premium offering.

I'd do $50 for 50 comments. At least.

I disagree. I am almost guaranteed to use this in the near future because it's so ridiculously cheap. At $50, I'd think about it a lot more and probably attempt to do it myself. Are there more people like me, or more people who wouldn't use it because it's too cheap to trust?

I also disagree for 2 reasons.

- At $5, you could try it out & develop trust as appropriate. This isn't market research with a consultant who can guarantee anything. This is DIY & you decide for yourself what it's worth. There's room for both kinds.

- Neither 5 or $50 is going to provide high quality trustworthy anything. That's not the point of this. The point is something interesting & potentially useful at a price that means you don't need to think about it too much.

Nice site. Clear goals. One suggestion: Why limit at 50 responses? What if I want more than 50 people to participate in the survey/market research? Can I setup something? It wasn't obvious from the web site.

It's limited mainly for simplicity. Since it's intended to be a cheap and quick answer, we figured $5 and 50 questions seemed reasonable for most. Perhaps we can keep the standard, advertised solution at 50 for $5 but when you buy offer the option to upgrade to more questions.

I personally think $5 is a great price point, but would make 3 tiers of plans.

1) Free, limited number of responses (maybe 5/month?), ad supported.

2) $5, Unlimited number of responses, no ads.

3) Skinable version for companies, allows them to manage questions/responses asked by employees.

If you go that route, make sure to make it more discoverable. If people write off your product because they need more responses, they're never going to figure out that they could get more.

Currently no, you can't set up a higher # of responses.

We found that 50 responses offers a good set of opinions without taking too long to finish. What are some other options you'd like to see?

Great app. A few thoughts.

- The demographic segmentation just doesn't work with 50 answers. Maybe package it (say $5, $20, $50) & either only offer segmentation to the higher ones or recommend against reading too much in to the lower ones.

- Host the images & otherwise make the image part easy. I'm guessing this is an upcoming feature. Since you clearly set out to make market research cheap & easy, this is part of that. You'd be surprised how many people don't know an image has a URL. Also make it easy to get online images from websites without knowing images have URLs.

Really nice overall, but I have one UI issue:

When I go to an example page, it says something like, "Survey says: B! (18 of 33 answers)"

And then I never remember which was A and which was B, so I have to look.

Why not "Survey says: Gmail!" or "Survey says: Obama" instead? Especially for the text ones. For the image ones it kind of makes sense to do A/B since they don't necessarily ever tell you a name for each image (though you could change that), but for the ones that are just text, it would be way more user-friendly to put the name in.

Impressive! Keep up the good work.

Just wondering, how has RPX ( https://rpxnow.com/ ) Login outsourcing/integration worked for you guys? Would you recommend it highly?

Edit: Btw, the Yahoo OpenId login prompts: "Warning: pickfu.rpxnow.com has not forwarded their Privacy Policy."

For a site as simple as this one it's worked out pretty well. Easy to drop in and 1 less thing to think about it.

Thanks for the privacy warning - we'll look into it.

I know domain names are hard BUT


your domain looks like :: PickFuck :: to a casual glance.


Similar to how the FCUK brand (intentionally) looks like FUCK.


I think your site is good, although it seems the demographics are exclusively American. Hopefully you could add some multinational demographic later.

Good luck and well done!

This is an excellent idea. Flying way, way under traditional solutions.

Who is your target demographic (whom do you expect to purchase these surveys)? Small companies?

Anyone on a budget or looking for a quick, cheap answer. Small businesses and entrepreneurs like ourselves were the initial target.

What is the geographic distribution of your responders?

It's almost entirely US-based.

I'd make that clear if you haven't already :)

Err hang on. Are you using Mechanical Turks? That can't possibly be a true representative of the online population, much as some superficial demographics might match.

Wow, this is pretty upsetting because I have been working on the exact same idea as you. How long has it been up, what kind of volume have you been doing, and how long did it take you to make it?

How about change "upsetting" to "motivating" and you're good.

The world is a big place, and competitors mean the market exists.

Competitors doesn't mean a significant market exists, it means multiple people have come up with the idea.

According to the developer's blog, the site went live in December '08 and took about two weeks to build.


Yeah, we initially built it a few months ago to solve a need we had internally. Didn't see any easy way to do simple A/B testing so we decided to build it since we figured it might be useful to other people too. Haven't done much marketing for it yet, but glad to hear there's interest for it. Some awesome feedback here too.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact