As time goes by I become more confident that I made the right choice to build my business platform with bubble.
One area that concerns me however, is the page loading speed and fact that most of the content is behind a JS layer, in respect of search engine visibility, as things move rapidly towards AI searches.
From the research that I found, Google is able to penetrate the JS layer, which is good news. However, it seems that none of the other major AI platforms do, not seem to have plans to. This could be problematic for bubble and put bubble sites at a disadvantage compared to competitors.
I realise we can still do quite a lot of SEO optimisation with bubble and for now it can hold its own - for traditional search engines, but my concern here is more about how bubbles sites will fare as things rapidly move towards AI like chatGPT etc
Would love to hear some community thoughts on this, and even better if anyone from the bubble team can let us know if any of this is being taking into consideration or worked on. Obviously some kind of static rendered page ability would be good.
Side note: I looked into the tool called prerender.io but they told me it doesn’t work with bubble apps stored on the main cluster.
I’ve tested extensively and the results have actually been quite illuminating - highlighting just how naughty & sneaky these AI’s can be. I tested:
chatGPT4/5
Grok 3/4
Claude 4
Gemini
Copilot
The short summary is that ONLY Gemini could read the content of our test pages correctly.
The others did a convincing job at pretending they could read it - but when pressed, admitted they could not. They ‘guessed’ at the content and often quoted from similar/rival websites instead!
This is really actually quite a big deal because for example, it gave wrong answers to FAQ questions, including important legal aspects - completely undermining our company and painting wrong impression.
I think we must not brush this under the carpet - it’s a massive area that bubble needs to address because Ai searches will soon completely replace traditional ones.
I’m not sure this is 100% Bubble related. If one of the AI was capable, for me, the fact Gemini could read it, seems to indicate, it is not Bubble, it is the other AI.
Based on the two forum threads on this topic, I asked ChatGPT further questions around it, and that summary above indicates, fault may be 50/50 and that bubble may be able to do some things to help out with the issue, but I think it is also related to the AI systems themselves too.
Bubble pages rely heavily on client-side JavaScript. Some browsers and crawlers will render this in full and give Bubble time to load, others will not and will just load the content that is returned.
That’s why different crawlers will have different visibility.
Yes it’s quite obvious that the issue is that the majority of AI crawlers are not penetrating the JS layer and while that is not bubbles “fault" there is no indication that any of the Ai services are bothered about doing so or have any plans to. That’s why the onus is on bubble to take care of this. The fact that an absolutely basic bubble text page cannot properly be read is an issue. Some kind of native ‘prerender’ option is required.
I’ll check pagerabbit out. Has anyone tried it? Does it get around the cluster-issue that prerender.io suffers from?… finger crossed. Will check it out.
