Organic acquisition on GPTs is much lower than Plugins. What should we do?

Hey there!

As many of us noticed, it seems the organic acquisition of new users is much harder on GPTs than it was on plugins.

To give an idea, when we launched our plugins like TotalQuery or Michelangelo, we had a few thousand new registered users the first week.

On GPTs, after 3 weeks, we registered under 50 users, and I’m not even sure it was 100% organic.

That leads me to the following questions:

  • How do you manage your acquisition?
  • Do you have any acquisition strategies that worked for you?
  • Anything you wanna say about it?

Have a great one

1 Like

Hi @kevinpiac !

Great to see you back in the forum and nice topic!

I just want to say that with plug-ins the competition was completely different. Back then there were two(?) waves of developers allowed to develop plug-ins. Today everybody can publish a custom GPT. Instead of one-hundred offerings that are easily screened by the users, today one needs to have an idea what to look for.

It will be interesting to read about how creators cater to their audiences.

3 Likes

Thanks for the warm welcome!

Yes, I agree with this deduction.

Curious to know as well how people manage their acquisition.

2 Likes

Is there anyone from OpenAI listening?

I think we’re left with having to do the doing marketing for our own creations… which just brings more people to sign up for ChatGPT Plus.

  • I’d like to have a widget so I can embed my GPT on my own website.
  • I’d like to be able to offer a free trial.
  • It would be nice to get the revenue-share sorted out, since we will bring new Plus subscribers to OpenAI

I also think a subscription similar to Netflix would make sense.
you pay and then you have access to all paid GPTs. Kobble.io could build this!

I know I am not answering the original question… I think there is no easy answer.
Every builder needs to reach their own target audience, just like with anything else.

1 Like

Enabling GPT developers to display ads or form a partnership with Microsoft Bing could be mutually beneficial.
By allowing traffic from the internet to access our GPT service for a brief period, such as 3 or 4 interactions, we can initially engage users without requiring immediate signup. This approach encourages users to sign up after they’ve had a chance to see the value of our service. Similar to how Google’s Bard have operated in the past.
By not mandating immediate account creation, either free or paid, we remove barriers that otherwise might redirect traffic to competitors. Paying for traffic that doesn’t convert due to these barriers is ineffective.
Suggestions include not enforcing sign-up immediately but allowing a few uses of the tool before prompting for account creation.
Traffic is money, and by not fully leveraging it due to the signup barrier, both we, as GPT builders, and OpenAI, are losing out.

I am still curious the market for GPTs will evolve in the longer term.

On DevDay when Sam announced it on stage, I thought this is a brilliant idea. However, I see them mostly as a “change management tool” to make building with generative AIs more accessible to individuals without coding skills and thus contribute to and lower the barrier for the overall adoption of generative AI on a day-to-day basis

Looking at what’s available in the store, I have so far failed to see the value-add of GPTs. I’ve rarely used one and I would most definitely not pay extra for any of them at the moment (beyond the USD 20 for the subscription). I think there would be a strong benefit in increasing the threshold for which GPTs should be available in the store and certain other limitations.

Sorry - I know this doesn’t address the acquisition question but I do think there are quite a few structural things to hash out before it’s even possible to successfully execute acquisition strategies.

Thank you all for your input :slight_smile: that’s definitely interesting

There is a considerable effort to build a “proper” GPT. 90% of the work goes into writing code that will be deployed as web services which then become GPT “actions”.
Furthermore, there are hosting costs associated with it. If it really picks up, there can be 100K calls to those web services per day - or more.
Developers need to see that it’s worth doing this… and today they can’t see it clearly.
That’s why most of the GPTs you find in the GPT Store are just built from prompt plus some uploaded documents. The development effort is much lower and there is no operating cost.

The five star rating and reviews will filter out the crap but still do not solve the problem of getting discovered. Builders still need to do that the old fashioned way.

I’m not saying your arguments are invalid.

But I also think you gotta give OpenAI a bit of time to observe the evolution of custom GPTs. I’m sure there have been plenty of learnings since November that have impacted the direction in which they are looking to take this and that has led to some delays in strategic decisions around revenue sharing model etc.

Finding the right governance and oversight model for custom GPTs isn’t a straightforward task. That’s proving difficult already now while OpenAI maintains a close tab on custom GPTs but will get exponentially more difficult (I might be exaggerating) the more freedom you provide builders to deploy their GPTs. Unlike API-built stuff, for custom GPTs OpenAI retains a certain level of accountability.