Get GPT-3 generated AI-Suggestion for the Title and Copy of your landing! 🤖
Hey Guys!
We were invited to the private beta of GPT-3 and we already implement it in our product. It's ready to use on https://abtesting.ai!
You have an example of Stripe on the cover of this post. Those suggestions were made by GPT-3.
Or feel free to leave your landing in the comments and I will share a screenshot with you!
Trending on Indie Hackers
I’d love to see what you get for https://simplebackups.io :-)
I’m jealous... how did you manage to get access !!
Hey! We requested and it took a little time but we just got the invitation :)
Probably having a product called abtesting.AI helped haha :)
It's great I see some big flaws in it.
So converting landing pages are not always about copy.
Sometimes it's Copy + Graphics which are meant to compliment each other so they both can work together to bring final message and this tool just check copy.
Then I noticed that the tool is suggesting the same formula I was using, My formula was Save {n} on {desired output}, the tool suggested Save {$} on {$}. I guess mine makes most sense?
Also the tool suggested different colors for my CTA but those weren't correct too, why would I use a color which neither fits my site color theory nor does it compliments it. (I understand CTA needs to pop out but those colours were pathetic).
Also I belive the suggestions it was making weren't related to the actual content of the webpage.
Please don't take this feedback in a wrong way, consider it a constructive feedback.
P.s. I'm copywriter(not a pro though) and the site I tested(owned by me) was based on copywriting and copywriting formulas.
That's great feedback!
Our tool also get's image for you to test, of course. You will be able to select all the elements yourself, now you just have to ask support. We will also add a way to request more variants, remember we implemented GPT-3 yesterday, and this is our first feedback.
We will keep improving the tool!
Regards,
Gonzalo
WOOOW, terryfying in a way.
It is indeed
A compelling addition to an already interesting product! Well done guys. Keep rocking it. 👍
You are one of a kind man!
Love it :)
Thanks for the comment :)!
Wow thats impressive!
GPT-3 is incredible. Here you have a couple of ideas for Astropay!
If a human made changes rather than 'AI', would the results be the same or different?
Good luck, Gonzalo!
Thanks, man!
Gonza! Escribime. Te empece a seguir en Twitter recien. Sos un capo!
Te escribí por twitter!
Grande! Sábelo! Te escribo hoy! Un abrazo!
It sounds interesting! My landing page: https://undersend.com
I liked the copies more than the titles!
The site is really nicely done, great job!
I'm trying the demo on one of my Shopify product pages (which is essentially my 'landing page' from FB ads for this store) and it wasn't able to identify most of the elements on the page (it got the main CTA and the hero image, but everything else it picked the wrong elements on the page).
Nice that you guys have a Shopify integration though, I've been wanting something like this. :)
Send me an email to [email protected] with the link and I will get the elements right for you :)
Also, how are you tracking "unique visitors" on the site? What indicators do you use?
Really interesting product, looking forward to seeing how you guys proceed.
I would be curious to see suggestions: https://festapi.io/
Your free tier has unlimited visitors while the premium plans are capped? I'd say if you can afford it make it all unlimited or cap the free.
Great feedback if you felt that way.
As it says on the pricing page, the free trial lets you run 1 experiment, and you don't have limits in that first experiment. Then, if you like the experience and the results, you have to upgrade to keep testing.
Being a data scientist I love those approaches applied in so Many different field 😍 congratulations for this product!
I leave here the landing page of IsRicherThanYou.com
Well done, this is very cool.
Do you guys have any way to do cross-domain conversion tracking? Unfotunately my signup page is hosted on a different domain than my marketing site.
If you can get our JavaScript snippet to both domains, you won't have any problem.
Interesting! Would love to get a screenshot for Formito.com
I will get back to you soon, because your title is a little bit trickery.
Awesome! Love to see the NLP application!
Can you try my new landing page? https://inksprout.co
Hey!
I accidentally took the screenshot with the preview on one of the titles. Feel free to go to abtesting.ai if you want more suggestions!
Also, I really thank you if you like this post
Cheers!!
I checked the site. It's really nice! Worth checking out!
Oh nice! It really feels the power being done on your own website! Will check out the site for digging around more! Thanks!
Hey! Thank you!! We extract the data from your site, but the suggestions are by GPT-3. Glad you liked it!!
Feel free to set up a free experiment if you want and give me feedback or ask questions if needed!
Will do! My engineer brain caters my curiosity to how GPT3 is functioning ha. But as far as the overall product, it's very polished with visual images integrated with js snippets for testing (which I imagine is fully automated too)!
This seems rather nice! Here's my newest landing for idea validation: https://www.hudle.live
Thanks for sharing btw!
I really liked the third title suggestion and the second copy suggestion!
A Like in the post would show it to more people and is really appriciated!!
Good luck with your project!
That's quite impressive for sure. Voted and thanks again! Good luck to you too!
Here is my concern. Reproducibility in science is so important. If an experiment is not reproducible, it can't be called science. Would you be able guarantee that your experiments are reproducible? If the experiment is repeated in another month will it give the same result. Are you willing to share the data of your experiments? If not sharing data then it is going to be an unknown black box.
"We run our experiments in batches so you can reap the benefits of A/B testing earlier and without needing too much traffic on your site". Not really understanding this. If a site doesn't get traffice like Quora or Google, are they going to get enough sample size? How are you calculating sample size?
Just wanted to make the audience aware.
Hey!
Those are some great concerns.
We run tests with a multi-armed bandit approach. If you are not familiar with it, there is an awesome article about that here: https://towardsdatascience.com/beyond-a-b-testing-multi-armed-bandit-experiments-1493f709f804
Second, of course, no, I will not share the data of our client's experiments, but the examples in the article are pretty clear, and we run tests like those. About your concerns about clients without so much traffic, I will highlight the conclusions of the article above about when is better to use multi-armed bandit:
When the cost of sending users to a losing arm is high. In this example, matching users with a bad advertisement simply results in less revenue. The loss is affordable. In other cases, such as when testing two different account recovery methods, a failure in each arm means permanent loss of a user, and multi-armed bandit experiment is clearly a better option.
For early startups with insufficient user traffic, multi-armed bandit experiment works better because it requires a smaller sample size, terminates earlier, and is more agile than A/B testing.
When there are more than two variants to test, bandit experiment can help us find the winner quickly.
On the other hand, of course, we are a young product with a lot to improve. We are still arguing what is the best approach to give information and decision power to users because a lot of them don't know much about the math running behind.
If you want to keep discussing, feel free to shout out an email to [email protected] or to [email protected], he is the Software Engineer responsible for the algorithm.
I need to read the link you sent. Lot of things unclear because they are hidden behind the jargon of multi-arm bandit. Did you vote the comment down so that it appears at the bottom and only the positive comments are on the top? That's not nice right?
I want to offer some healthy critique on this a/b testing field. Conversion can be because of multiple factors. The landing page which is converting more could be because of
More here https://www.pranaysite.com/ab-testing/
That’s a really well written article! It has a lot of interesting points. But it kinda seems you just wanted to drop the link for traffic.
First, we don’t do testing with multiple landing pages, only one. Maybe you didn’t try our tool or approach?
You also you could be getting more leads because of improvements. All variants are getting the same opportunities. Our tool even do some re-test of different combinations and variants. We use a multi-armed bandit approach.
Last but not leads, you didn’t give alternatives or suggestions, just a generic copy/paste from your article?
I would love if you test our tool and get back to us to give us feedback!