How should I launch an API demo as an MVP?
I have an API that I would like to launch as an MVP, to gauge market interest.
It extracts keywords from text.
Goals:
- Acquire users on my mailing list, so I can market other products and services from my consultancy.
- Gauge potential interest in charging for the API on a per-subscription basis.
- Gauge potential interest in people having the service customized for their particular data. (The API is for general web text, but people might want it for high-value technical domains, on which the demo will not work well.)
Here is a competing API.
They allow you to test the technology from the web (not from an API). If you register, there is a free plan with a low usage limit, and paid plans for more usage.
What is the correct way to launch my demo? Each launch strategy has potential benefits and disadvantages for each goal:
- Just put the API up, explain that it's a temporary demo and ask
people on the honor system not to make more than 1K requests per
day. Benefits: Low barrier for testing. Disadvantages: Doesn't
acquire users for my mailing list, might get abused. - As before, but
also ask that users sign up for my mailing list. - As before, but
don't allow users to try the API until they sign up for my mailing
list. - Force users to create an account on my system. Advantages:
Since users are signed up, it is easier to get them to start paying.
Can directly prevent abuse. Looks more professional. Disadvantages:
The highest barrier to trying my API means fewer people test it. (I
can encourage them by posting sample output though.) Extra coding
time. Less forgiveness from users for kinks I still need to work
out (e.g. isn't scalable to multiple concurrent requests). - Copy the competitor's approach, i.e. demo through a web interface and force users to sign up for a free or paid plan for API access. Benefits: Easy to demo. More clear to the user that abuse is not allowed. Disadvantages: I have to code up something that allows me to sign up users, should they want to sign up. Alternately, I could collect their email addresses.
What are your thoughts on the difference approaches?
-
Joseph, you've asked 23 questions, and have an accept rate of 5%. You don't have to accept an answer to every question you ask, but I encourage you to start accepting more answers. Surely we've done a better job here than just 5%. Accepting an answer helps keep the community organized, it increases your rep, and it encourages people to answer your questions. – Zuly Gonzalez 13 years ago
-
@ZulyGonzalez: How does accepting answers make the site more organized? Please explain. The reason I don't "Accept" answers is because many of the questions are subjective, and often there is no conclusive answer. I don't want to privilege one response over another one that is also very good. Business advice is different than StackOverflow, when it is very clear that an answer is correct. But I am open to changing my approach if you explain more. – Joseph Turian 13 years ago
-
Sorry, I never saw your comment back to me. You don't have to accept an answer if you don't feel like your question was answered. Please don't accept an answer just for the sake of it. My point, which I admit I didn't communicate very well, was that accepting a good answer that addressed your question helps everyone. It's a nice way to say thank you for taking the time to help me to the person that did a good job answering your question. It helps future visitors because it lets them know that was the most helpful answer to your question. If they have the same question they know where to start. – Zuly Gonzalez 12 years ago
-
And it helps you because some people may be less likely to take the time to answer your questions if they feel like you are not giving back to the community. Again, don't feel pressured into accepting an answer that's not helpful, but if you do receive an exceptional answer, please consider accepting it. If you can't decide between multiple good answers, maybe consider leaving comments explaining what was helpful and what was not. That way future visitors can judge whether they want to take a particular approach or not. – Zuly Gonzalez 12 years ago
1 Answer
Here are my three suggestions:
1) No matter what else you do, have some way to collect email addresses. A page like "If you would like to be part of our beta test please enter your email address..."
2) At this point a web based test interface that anyone can try without signing up or providing an email address is a good idea. One doesn't need to make an API call to determine the quality of your keyword extraction, a web interface like Alchemy will work just fine.
3) To prevent abuse, no matter if a HTML or API based test interface, simply throttle requests. You can either limit the number of requests from a particular TCP/IP address (4 per min perhaps) or make each request take a minimum of 15 seconds to respond. If you do have throttling then say so and tell people that they can contact you to arrange for a tryal that is not throttled.
