‘Redefining SaaS CX with Agent-led Onboarding’, by Marieliesse Gouilliard
A common assumption in SaaS is that if the product is good enough, users will find their way to value on their own. In reality, most never do; across the industry, activation rates hover around a third, meaning most signups never experience what the product was actually built to deliver.
In this insightful piece, Women in CX Inner Circle member and Co-founder & CEO at Autoplay, Marieliesse Gouilliard, explores how agentic AI could reshape onboarding for SaaS teams and why product usage may be the clearest signal of customer experience we have.
Drawing on research from Gainsight, Mixpanel and Growth Unhinged, Marieliesse unpacks why product adoption isn't just a side effect of a happy customer but the leading indicator of one, why the market's usual onboarding fixes (tours, chatbots, documentation) keep falling short, and how agents can finally deliver the kind of real-time, personalised guidance that was previously impossible at scale.
This article is a timely reminder that fixing onboarding was never about adding another tool to the stack. It's about giving users the right guidance at the right moment, freeing customer success teams to focus on the conversations that actually protect and grow revenue.
How Onboarding With Agents Could Redefine CX For SaaS
New user onboarding is your platform's one shot at a good first impression — and right now, SaaS has an activation crisis. Across the companies analysed in Userpilot's 2024 benchmark report, the average B2B activation rate sits at just 37.5%, with the median barely above a third.
Put plainly: fewer than four in ten signups ever experience the value their product was built to deliver. Lenny Rachitsky's independent research lands at 36% — so this isn't one vendor's skewed sample; it's the industry norm.
It's getting worse, not better. Attention spans are shrinking, AI is making it trivially easy to spin up new products, and users have more tools competing for the same five minutes of curiosity than ever before. The bar to 'check out a new product' has never been lower — which means the bar to stick with one has never been higher.
“Attention spans are shrinking, AI is making it trivially easy to spin up new products, and users have more tools competing for the same five minutes of curiosity than ever before. The bar to 'check out a new product' has never been lower — which means the bar to stick with one has never been higher.”
Product Adoption Equals Your SaaS CX
This isn't just a feeling. It's been measured. And the finding is specific: product adoption isn't a side effect of a happy customer. It's the leading indicator of one.
Gainsight ran a random forest model against 60+ attributes across their customer base to predict renewal. Two usage signals came out on top: breadth of feature usage and daily active user percentage. Both beat softer signals like NPS, time-to-launch, and event attendance by a wide margin. There was one surprise, too: it wasn't a high number of support tickets that flagged risk — it was too few. Silence meant disengagement, not satisfaction.
In other words, 'I don't know how to use this software' isn't just a frustrated comment. It's the clearest sign a customer won't renew. We've lost customers post-sale over features the platform already had. They simply never found them.
This pattern isn't unique to Gainsight. Mixpanel's benchmarking research found that products with clearly defined activation events see up to three times higher trial-to-paid conversion than those only tracking surface-level engagement. Kyle Poyar puts it bluntly in Growth Unhinged: treat usage as a signal of buying intent, and discovery, activation, and conversion become one chain, not three separate steps.
Put these together, and the conclusion is simple: product usage doesn't just correlate with SaaS CX — it's close to a one-to-one proxy for it.
The adoption curve isn't a side detail of SaaS CX; it is SaaS CX.
What The Market Has Tried, And Why It's Failed
The market has tried almost everything to fix this: videos, documentation, pop-up product tours, dynamic demos, 'ask me anything' agentic interfaces, AI chatbots parked at the bottom of the screen.
None of it has moved the needle, and not for lack of data. Tools like Amplitude and PostHog can tell us exactly where users drop off, down to the click. We have all the data and insights we could want — we just haven't fixed the problem through action, because none of these solutions addresses the real job to be done.
They fail for the following reasons:
Users don't know what they don't know. They can't ask about a feature they've never seen or don't know how it works. Onboarding that depends on the user asking the right question becomes a catch-22.
Users don't want to be told what they already know. A tour walking someone through a feature they've already mastered reads as condescending, and it's often the moment they tune out every future prompt.
Users get stuck and don't seek help. Users expect value with no friction. If they don't get value out of your product, they won't go out of their way to seek help — they'll go check elsewhere.
Users want help when they need it. All of the above solutions sit outside the product experience. None of them helps the user in real time, while they're actually inside the product.
If we could wave a magic wand, with unlimited time and headcount, what would ideal onboarding look like? You'd get on a video call with every new signup, watch them explore, and the moment they got stuck, step in: 'Click here, try this.' That's the gold standard, but it's completely undeliverable at scale.
“Users don't want to be told what they already know. A tour walking someone through a feature they've already mastered reads as condescending, and it's often the moment they tune out every future prompt.”
How Agents Offer Us An Opportunity To Finally Fix This
The magic wand solution we just described — someone watching, noticing the moment of friction, and stepping in — is exactly what an agent can carry out at scale. Unlike a human, it isn't limited by headcount: it can watch every user simultaneously, identify the moment each one gets stuck, and proactively step in.
Agentic AI turns the onboarding experience into a proactive, personalised conversation — surfacing the right action exactly when it's needed, without burying the user under everything they could possibly do. It's the difference between a manual dumped on someone all at once and a guide who tells you the one next thing to try, then checks how you do before suggesting the next best action.
What makes this possible today is context, delivered in real time to the agent, so it can reason about when to step in and what to say — per user, based on what that person is doing right now.
Every user gets a genuinely different onboarding experience, rather than the same tour played on a loop.
The good news is that no new software, and no rip-and-replace, is needed to make this a reality. Teams already capture all the input data on user behaviour through their existing product analytics tools. All you need to do is connect an extra MCP to your existing support agent.
What Proactive And Personalised Agentic Onboarding Unlocks For CX Teams
A huge amount of customer success time goes into answering the same low-level usage questions. Repetitive 'how-to' questions, feature explanations, and onboarding guidance are estimated to make up 60 to 70% of support ticket volume — time not spent moving the business's KPIs.
If customer success teams can delegate the actual teaching of how to use the product to an agent, it frees up their conversations and time with the customer to focus on higher-ROI work: understanding a customer's plans, what else they're evaluating, and how to maximise touchpoints across the account. That's the conversation that actually protects and grows revenue.
This would make room for the customer success role to become more commercial — less about explaining the product, more about driving value and revenue.
“If customer success teams can delegate the actual teaching of how to use the product to an agent, it frees up their conversations and time with the customer to focus on higher-ROI work.”
Keep the Conversation Going
Marieliesse's article is a timely reminder that fixing onboarding was never about adding another tool to the stack; it's about giving users the right guidance at the right moment, without relying on headcount that would never scale.
When agents take on the repetitive work of teaching customers how to use a product, CX and customer success teams can focus on what protects revenue: understanding a customer's plans and maximising the value of every touchpoint.
These are the conversations we're passionate about inside the WiCX Inner Circle.
Through expert-led webinars, member discussions, live podcasts, and collaborative learning, we bring together women who are exploring how AI, technology, and design are reshaping the future of customer experience.
If you'd like to continue the conversation around agentic AI, onboarding, and the future of SaaS CX, you can join the WiCX Inner Circle waitlist today.