And also, the scale also benefits from our exploration into the new businesses. That’s your first question. What’s the marketing and other services are? So that’s the initiatives we explored in the past three years. And now it seems that we still recorded high growth, proven our capabilities to enter into these areas. And then the second is about the efficiency. You can see a sequential improvement of our ROI. The ROI means how we measure our marketing and the user retention capabilities using revenue divided by our cost of marketing and acquisition. So we are seeing the continuous improvement of efficiency in this regard. And thirdly is the cost optimization. We continued, during the past quarters, of course, including — that’s including some cost-cutting initiatives and also — it’s also an efficiency gain in terms of we deployed certain AI technologies into our — to improve our cost structure.
I think that’s so far the initiatives and the facts we have — we’ve done so far to improve our — to bring us to the approaching breakeven status. But looking to the future, I think your second part of your question is about how we can break even by the end of this year. I think nothing can be guaranteed, particularly in terms of the uncertainties and volatilities of the market environment. So if we can continue to grow the scale, improve the efficiency and saving the cost, definitely, we will — we can be breakeven in the future, but the volatilities and uncertainties may lead to something we cannot control and not — we cannot expect for now. So yes, of course, our goal is to build a healthy business to make profit, but the visibility of obtaining breakeven is — I think it’s not very near term.
I hope that answers your question?
David Ye: I have a few words to add, this is important. So we were approaching breakeven in Q2. However, I mean our business is heavily dependent on the macro economy of China, especially the health status of Chinese financial markets. We have seen some recent data last week and those data, in terms of the real estate market, the SME and the consumer loan I mean and also amount and also as the deposits. We definitely have seen a big decline of those numbers. So in the nutshell, Chinese consumers, they are paying or even paying off their debt, businesses are slowing down their borrowing and also their risk — credit risk or the ability to pay for consumers are actually — their ability are declining. So as an open platform, we heavily rely on our financial partners, banks, loan bank finance company, credit issuer, data capability to manage credit risk and manage growth and serve their customers.
So we are not giving earnings outlook for Q3 or Q4. It’s just hard. It’s tough. It’s not being in the playbook of [indiscernible] data to do the estimation or the estimation. So that’s just my take, my personal take of the sector and economy. However, we as independent of open platform, right? We have seen in the efficiency gain in the last quarter. We have light asset platform. We have been improving over quarter-over-quarter, and we are confident we, the management team and every one at Jianpu, we are able to execute. We are going to outperform our peers and we are going to do better quarter-over-quarter. Thank you.
Unidentified Participant: Got it. Thank you, management.
Operator: Thank you. [Operator Instructions] And the next question comes from Carl Yang with Junggai (ph) Securities.
Unidentified Participant: Hello, management. Can you hear me?
David Ye: Yes.
Unidentified Participant: Okay. Thank you for giving me the opportunity. I would like to know how has generated AI helped your business? And what will be your plan of AI developments in 2023? Thank you.
David Ye: Oscar, please go ahead.
Oscar Chen: Yeah. Okay. Yes, thank you for the questions. Yes, I think, firstly, for the AI, I think it’s still in the early stage to — I think, almost in early stages to everyone. So for now, what we have done is we created an internal one-stop model that’s aggregating various AI tools, including large language models and other AI technologies for our internal use. So through that we already shared some in our — in our prepared scripts that we see significant efficiency gain in terms of using AI tools in our daily work, particularly the R&D, the enhanced efficiency of the — of our engineer to write code and also in our customer services so on and so forth. So that’s the internal part. And also, we have capturing the wave of AI, we will build — we have AI Hackathon event in the second quarter.
The purpose of this event is to encourage and to find some bottom-up ideas, initiatives that could be further developed or commercialized. In that event, we do find some interesting ideas, but it’s still in idea or demo stage. But we see some potential there and we look forward, we can help our team to further build and enhance these initiatives and hope that can be commercialized in the future. And also one thing we want to share is, you may also heard from the expert and media that if to deploy the AI, particularly the large language model into certain scenario or user case, the sector expertise or the domain knowledge will play a more important role in that regards. We also believe in that theory given what we have accumulated in the financial services industry, including the data, the user, the user behavior, all these could be the pretreating materials to feed into the large model and may turn out into something interesting.
