“Inside sales teams struggling because nobody picks up the phone should use this. We are in an era when people will not answer a toll-free 0120 number, of course, or even an 03 Tokyo number — and you can still book meetings without making calls. After adoption, monthly SQLs went from about 20 to 41–48. And 70 to 80% of the chat-sourced leads turn into meetings; they are all high-quality leads.”
— Shintaro Kuniyoshi, Inside Sales, G-gen Co., Ltd.
G-gen Co., Ltd. is a Google Cloud partner that supports Google Cloud and Google Workspace end to end, from adoption and migration through operations to in-house enablement. It has both a Google Cloud sales team and a Google Workspace sales team, and its leads come mainly from website inquiries, event exhibitions, and referrals from Google. It has worked alongside companies considering the cloud with highly specialized consulting and implementation.
To general business audiences, however, Google Cloud looks like a service for engineers. Many customers are at the "what is Google Cloud?" stage, and how to shorten the lead time to a meeting had become a major theme for growing the business.
Before adoption, the inside sales flow went like this. An inquiry comes in from the website → check the content → search for the company in Salesforce and check past inquiry history and whether it is an existing customer → visit the company's website to understand its size and potential → then phone them and ask what needs they have around Google Cloud or Google Workspace.
That alone took 30 to 60 minutes per inquiry. On top of that, calls often did not connect right away, and time was lost before reaching anyone. Even once connected, a person had to spend effort judging whether "this goes to a meeting with a rep" or "this turned out to be an odd inquiry after all." From there, the team scheduled the meeting while checking availability for each product team in advance (the Google Cloud sales team or the Google Workspace sales team), sent a Meet URL, and requested an assignment in the internal Slack. Only then did a sales rep pick it up.
Some people don't pick up the phone right away, so we were losing quite a bit of time there. And when we finally connected, would it go to a meeting with a rep, or would it end with "that turned out to be an odd inquiry"? We were spending a lot of effort on that judgment.
The other problem was understanding needs ahead of time. Some people write rich detail in the inquiry form; others write only "We want to bring in Google Workspace, what do you think?" and their potential cannot be read. Was the evaluation really moving forward, or was interest low? At the inquiry stage, there was no way to sort them.
The challenges were these two: understanding needs, and scheduling meetings.
Until then, the lead sources were website inquiries, events and Google referrals. To grow the business, the team wanted to add one more channel of leads it generated itself, and carry them through to SQL. That wish coincided with the proposal from Meeton ai. The aim was to find, early, the people standing at the entrance to an evaluation — those who know Google Cloud or Gemini to some degree and "want to adopt it but do not know how" — and shorten the lead time to a meeting.
Not "I took a call, so I said yes." We wanted to find customers who already know about these services and want to adopt them but don't know where to start. If an AI chatbot nurtures them, we get high-quality leads. That was the idea.
G-gen had previously adopted Salesforce Einstein as well. It wanted to win meetings efficiently, and it had plenty of leads and data. But while there was some effect on optimizing emails, whether that led to a first meeting was another matter, and it fell short of the results expected. Including event exhibitions, the team had been operating startup-style on "just do a lot of everything," but was left with the sense that the pipeline was short on both quantity and quality.
We did try Einstein, but it didn't really work for us. There was some effect on optimizing emails, but whether that leads to a first meeting is another story.
Tools from Japan and abroad were on the table for comparison. At the time, though, the overseas products gave a strong impression of "pushing sales hard at the website visitors who lean in," which was not the shape G-gen was after. What G-gen wanted was a style in which the visitor has some conversation in chat and the rep then goes into the meeting having read that history. That is why Meeton ai, whose AI capabilities looked rich, stayed on the shortlist.
Rather than leaning in and pushing sales hard, we wanted people to have some conversation in chat first, and then go into the meeting having read that history. In terms of product differences, Meeton ai was the one whose AI capabilities looked the richest.
G-gen has its own engineers, so building in-house was considered at one point. It still chose Meeton ai because it was overwhelmingly faster, and because people who can design a UI are hard to find internally. Since this was an experimental initiative, being able to try it quickly was a better fit than spending money to build it out.
We did consider doing it ourselves. But in terms of speed, this was overwhelmingly faster. And we didn't really have anyone internally who could go as far as UI design.
Several factors came together in the decision to adopt Meeton ai.
The UI is simple and easy to use. That is what it comes down to. And they apply our requests quickly. That speed built trust, and we were saying internally that they really are engaging with us properly. The reporting is solid too; how many we won each month is laid out clearly, so it is easy to publicize internally.
Because we can see the exchange in chat, we understand what kind of customer it is. "This person doesn't have potential." "People who are harsh with the AI aren't really promising." "This one is an existing customer, so we need to follow up." We can judge that from the conversation. That ease of use was the deciding factor.
G-gen set a POC period and drew an investment line: "if we can't book 10 meetings a month, it will be tough." The figure was worked back from the value of a deal; below 10, it would not pay off. The internal target was 15, with an average of 10 as the minimum. In the end the line was cleared from the first month and met on average as well, so the team decided to continue.
During the POC period, I thought that if we couldn't get 10, it would be a bit tough. But we cleared it from the first month, so we felt this could work.
Before adoption, SQLs from website inquiries alone were 18, hovering around an average of about 20. After Meeton ai was introduced, the numbers started to come in: in October the total of website inquiries plus Meeton ai-sourced leads was 41, and in November it reached 48. The average passed 30, and an additional 10 or more now accumulate steadily.
Before adoption, with inquiries as the only inflow, SQLs were 18, around 20 on average. After we brought it in, October was 41 and November was 48. So the average has moved to around 30. It really is an additional 10.
It was not only the volume that grew. Leads via Meeton ai convert to meetings at a high rate; roughly 70–80% go on to an actual meeting. There were almost no low-interest "hollow SQLs"; these were high-quality leads that, as a rule, reliably turned into meetings.
There aren't really hollow SQLs; basically almost all of them turn into conversations. My understanding is that through DynaMeet, more than 70 to 80% lead to meetings.
Because needs can now be understood in advance, the quality of first meetings changed significantly. Among customers who come via Meeton ai, there is no one who does not know Google Cloud or Google Workspace to begin with. The "what is this?" stage is unnecessary, and the conversation can start from "how do we put it to use?" As a result, conversations move forward easily, and both the length of the first meeting and the lead time have shortened.
Customers who don't know what we do, or who don't know Google Cloud or Workspace — through DynaMeet there are none of those at all. We can start from the premise that they know, and go straight to how they will put it to use. That makes the conversation move forward very easily.
There was a change from the marketing perspective as well. Customer interests that used to be put into words only once they made an inquiry can now be seen at the stage before that. From the chat history, the team can tell what information people are looking for.
We've come to understand quite well what customers are thinking. Until now they wouldn't put it into words unless they made an inquiry, but now we can see the stage before that. That part is good.
With less work needed to acquire a lead, inside sales can put the freed-up time into other tasks. It is paying off in both time efficiency and cost efficiency.
Some Google Workspace deals have limited impact, but on the Google Cloud side the team was able to bring in customers with large ARR — companies of a size that already use another cloud and are considering migrating or switching — and in some cases that led to very large numbers as SQLs. As described below, the chat was also visited by a customer with MRR on the order of 100 million yen.
The biggest change adoption brought was in the team's mindset itself. In Yonekawa's words, it was proof that "we can do this without aggressively chasing deals." Kuniyoshi contrasts it with how hard it is to nurture a house list into meetings.
Nurturing a house list into meetings is, frankly, really hard and a hassle. But when you approach potential customers at the stage when they are researching, it fairly quickly becomes a meeting, or they are willing to listen. When you make contact at a stage where the psychological hurdle is low, people tend to think "this is someone on my side" and lean on you naturally. So it is much easier to do.
In today's customer journey, the time customers spend researching on their own is overwhelmingly longer. Being able to get ahead and approach them at that "researching stage" was a strength that traditional, phone-led inside sales did not have.
Internally, there were skeptical voices at first. "How many people will actually type into a chat in the first place?" But in operation, far more people wrote in than imagined. Rather than suddenly getting a sales call or email, they want to research on their own first and then move forward — that customer mindset was clearly there.
Rather than suddenly getting a call or an email from sales, there's quite a strong wish to look into things yourself beforehand and then move forward. I felt that as a gap from what we expected.
What worked in persuading people internally was showing the actual numbers, and the advantage of "being able to approach prospects 24 hours a day, 365 days a year." The team won the company over with the story that it can approach customers at the stage when they are researching on their own.
What worked well in operation was that the AI's effort is visible through notifications. A long exchange with a given customer means high interest, and that gradient can be seen in real time. Seeing it, Kuniyoshi himself immediately switched into the chat, stepped in, and turned it into a booked meeting. There were about 10 such interventions in total. All went relatively smoothly, and among them was a customer with MRR on the order of 100 million yen.
I can tell from the notifications that the AI is working hard. I can see the gradient — "it's been going back and forth with this person for quite a while" — so it is easy to step in. I switch into the chat right away and take over. That happened about 10 times. It basically goes smoothly.
It was not all good news. There were real difficulties during the ramp-up period — but all of them were within acceptable bounds, and operations are now stable.
For the first one to three months, training the chat model took time. There was a period when I thought, "How long are we going to keep doing this?" But after that it started running stably. It stopped needing effort after about three months.
"We'd like a feature that reports each month's activity in natural language and advises on room for improvement." That request has already been released as Meeton ai's "Marketing Hub." It analyzes the website for a selected period, and the AI presents improvement advice. Conversations are automatically classified by topic, making visible which keywords come up most and on which keywords leads are being missed.
From a marketing perspective, there was also a wish to control finer details in-house. For example, if the team could arrange even a single line of the wording shown before the chat launches, it could try many things with A/B tests. Getting to a state where the team runs its own PDCA cycle — here too, what it can do is steadily expanding.
Even just that one line of wording before the chat launches. If we can arrange that kind of thing ourselves, we can try a lot. It would be good to be able to run the PDCA cycle ourselves.
Some requests came out of real operation: a feature that automatically shuts out "reverse sales" — vendors pitching through the chat — from specific domains or companies. There were a few cases in the past, but none since a sentence was added inside the chat. Going a step further, the team would like mechanisms that do not launch the chat at all for visits from specific domains, and that block complaints and persistent contacts.
It might be good to have a feature where the chat doesn't launch if the visit comes from a specific domain. A mechanism for handling complaints, or shutting out persistent customers, would be fine to have too.
Finally, to inside sales teams struggling because nobody picks up the phone, G-gen's message was simple.
You don't have to make phone calls. Go start a revolution in inside sales.
Inside sales teams struggling because nobody picks up the phone should use this. These days people don't answer a call from a 0120 number, and they don't even answer an 03 number. Some companies still call landlines as outreach, but I want to tell them: you can book meetings without doing that.
Capturing a highly specialized offering like Google Cloud and Google Workspace at the very moment a buyer steps into an evaluation, and turning it into a high-quality meeting. For G-gen, Meeton ai is becoming the foundation of a "new kind of inside sales" that does not depend on the phone.
In a 30-minute demo, you can concretely confirm how it works for your company.
