We have to block several users per day for creating multiple accounts. I can tell you no amount of wording on the website will work, people have multiple email addresses and many use burner/disposable/temporary emails for signup. As first line of defense add a captcha, then rate-limiting by IP address. Some download website have a counter "You can continue after waiting 10 minutes ... or upgrade now", that might work here.

Hmm, some interesting suggestions there. Thanks!

Captcha is a good one and easy to implement for now. Will do that.

I may be misunderstanding, but if "keep uploading the same document" is an issue, retain hashes of recently-uploaded documents and refuse to handle anything seen too recently too many times, unless being paid.

Sorry, to be more precise, "same document" could mean the same content but in different formats. OR just slightly tweaked to result in a different hash each time. Harder to abuse for sure, but still has holes.

Sure.. but at some point you've got a diminished return on efforts to stop users. 99% of abusers won't work around the limits.

use a cheap algo to detect % similarity?

Any sufficiently-determined (and smart enough) user will find a way around any such check; this is just meant to make it a bit harder to do the wrong thing so that some people will stop trying...

BTW, the web app is at https://clarifypdf.com, if you need to see what it is first.

Beautiful typography, I must say! Is it built on React or NodeJS?

Thank you. That's just an off-the-shelf template from https://tailwindui.com!

$1.99? At that point don't even provide a trial.

This is what I'm debating too. Thanks.

You might want to just put a signup on it, charge monthly per X number of PDF's/questions (whatever combo nets you cash flow after your service charges, etc.), have different plans, take their CC info and charge them at the end of a 30 day free trial.

Your price point of $2 is going to bring abusers of the service and if you have something promising your best bet is to go after smaller document based professionals (i.e. 1-4 person law firms) that want to speed up their processing/knowledge of documents but also don't know anything about ChatGPT, etc. They would likely give it a try and not think twice about a signup + credit card info if they think it will help them reduce costs.

Thanks for that.

What about it being $2 is going to bring abusers?

Ask for a credit card(not debit cards because you can get unlimited virtual debit cards with a lot of banks)

No Web site is getting valid card numbers from me until after I have decided that I trust it. And how are you propsing debit/credit status be checked anyhow?

there are APIs to check that

OK, I suppose that is plausible, maybe by BIN, if someone has put together a database!

Tie to a phone number, is basically it. Even that's not full proof with virtual number apps akimbo

Like others have mentioned, there’s many ways:

- Hash each pdf, check if it’s been run for free before

- Hash IP addresses, check before allowing free trial (can lead to many false positives though)

- Use local storage or cookies to detect if they’ve done the trial or not (easily circumvented, but not by general users)

Probably a combination of a few of these would give what you want

I like the cookie idea. Thanks.

Random advice from the internet:

+ Don't offer a free tier because people who won't spend $1.99 to try something out are not going to spend $1.99 regularly.

+ Raise prices significantly. Nothing filters high demand low revenue users more effectively...and that's what a free tier attracts.

+ Don't waste time trying to make crappy user experiences. Just focus on making good experiences and price your products so that you can provide high quality.

+ When you have more than enough money, then you can give it away to deserving causes. That's more likely to be a food bank than software engineers. But maybe not.

Good luck.

You've put in words what's been in my gut. Thank you.

Thanks everyone for your input. More than anything, it's been really great at making look at the bigger pricing picture.

My conclusion:

I'm not going to put in a free tier, because as @brudgers mentioned below, if they can't spend $1.99 to try it, then they're not my target market.

So, in words I read somewhere else before, instead of downgrading my product, I should upgrade my customers.

I will instead spend time creating a high quality product.

And slowly raise prices as I improve the product.

Thanks again, everyone.

Track by IP and check a hash or text of the pdf to verify each user/IP can only upload a pdf once.

I think $2 is too low, I'd be more likely to sign up with you for $20/mo for up to 10 pdfs. (add larger plans too)

Aren't you being hit with about $0.30 per charge in fees per trans + open ai api costs at the $2 price point?

If they login can you let them see their history for each pdf and go back and ask new questions later?

Building the login and auth isn't so bad if you use something like Laravel or Rails.

If you don't have money to burn i would strongly advise against the no sign-up approach. Providing these services is pretty expensive and the fight against bots/abusers is also very time consuming, so at least use a phone-number check if you want to provide a free tier.