The New GTM Playbook: How to Build an AI-Native Go-to-Market
Go-to-market strategies have to be rethought in the age of AI. But how? Here's a playbook that works.
There is a lot of noise around AI, but there are a lot fewer practical use cases on how AI is changing how we market and sell our products.
In this post, I’m going to dive into how HubSpot is reshaping our GTM with AI.
The purpose of this post is to give you the areas we’re seeing success in and some key learnings.
Buyers are changing:
- Authority has gone human. B2B looks more like B2C; buyers want to hear from individuals; not corporate blogs.
- Distribution is moving. AI answers are much better than blue links. Clicks disappear.
- Attribution disappears. All the influence is happening off-site; it’s nearly impossible to track.
3 GTM Trends That Matter Because of AI
1) Influence > Clicks
Clicks are disappearing. A growing share of searches end without a website visit.
“6 out of 10 search results in a non-click. All marketing is getting harder to track.”
AI mode is going to accelerate this. I’ve seen reports that suggest it could be upwards of 8 out of 10 searches resulting in zero clicks.
Tracking gets worse; influence still works.
Influence happens away from your website:
Share of Voice in AI Engines (are you named in answers?)
YouTube views (evergreen discovery + long-form trust)
Podcast listens (intimate, high-intent attention)
Newsletter opens (direct distribution you own)
AI accelerates this trend. It creates more ‘AI slop’. People’s feeds are already drowning in it.
People will default to humans more. They’ll want personality, stories.
Content that moves a market will carry a face.
We were early on this trend. At HubSpot we bet early on human-led media (e.g., The Hustle acquisition). That mix now drives more demand than branded content. Huge Shift.
2) AEO > SEO
AI engines are becoming the front door. Traditional SEO is still important, but AEO (AI Engine Optimisation) will become a key pillar of your AI-GTM.
LLMs will surpass search traffic.
By 2028, LLMs will capture 75% of search traffic share, while traditional organic search drops to 25%. - SEMRush
What’s changing is:
- Traffic mix shifts. AI answers steal discovery from blue links. Clicks continue to drop.
- New distribution scoreboard looks more like ‘influence’. Share of Voice in LLMs is a key metric.
- Entity > keyword. Models care about who/what you are and whether you’re mentioned + referenced across the web.
Visibility in AI answers will generate a lot of brand demand, but it will be hard to see. You’ll focus a lot on your Share of Voice as an indicator of visibility.
At HubSpot, we focused on AEO early:
- We’re the most visible brand (by SoV) in our category
- Demand from AI Engines is up 1850% (referral traffic)
- Citations up 433%
AEO still presents a huge first-mover advantage. We’re very early.
3) Value > Volume
In the AI era, you’ll get fewer visits to your assets. You need to extract more value per visit. More customers per touch.
Our data shows that traffic from AI engines converts at 3x the rate of normal organic traffic.
That’s great, but it’s because work happens elsewhere. The site is for decisions, not discovery.
You need to extract more from less.
In that world, a metric to obsess over is:
MRR per Human Engagement: Is the value of every sales touch trending up over time?
As AI scales through sales, the human touch should be worth more. Higher intent. Better context. Fewer, more valuable interactions.
The value formula is: Value = Context + Sales Agents + AI-Enabled Sellers.
i. Context Layer
There was a great study from MIT that showed 95% AI projects are failing for enterprise companies, despite $30-40 billion in enterprise spending.
One of the core reasons users abandon AI internally is that it requires manual context input every time they use it.
The big reason for the difference between the 95% and the 5% is your context layer.
A great context layer transforms raw signals (internal & external) + (structured & unstructured) - into AI insights.
The user doesn’t start from scratch every time; the context layer gets smarter with usage / making it easy to reuse that context across agents and GTM employees.
That’s how you separate yourself from the 95%.
ii. Sales Agents mean you can close more, faster
We’ve deployed three core AI Sales Agents:
- Website Agent → converts inbound traffic.
- Prospecting Agent → books meetings for sales
- Discovery Agent → assembles full context pre-call.
Agents are more successful if they share the same context layer.
a. Website Agent (Salesbot - converts like a human)
We originally started Salesbot to deflect low-quality chats on our website so humans could drive more revenue.
It worked. Salesbot now handles up to 80% of website chat and has increased our MRR per human engagement by 2.3×.
Then we trained it to sell: qualify, guide next steps, book meetings.
Today, Salesbot manages a large % of English chats and books meetings within ~10% of human conversion.
Some learnings:
- It’s not simply answering a question, it’s deciphering intent to buy
- We trained Salesbot on a HubSpot sales framework, which drastically improved its ability to qualify demand and look for intent signals
- Every chat feeds the context layer: Its intent, its step accuracy.
b. AI-SDR (replaces your website?)
I’m curious about the website’s future. Today, we spend a lot of time on brand and product positioning, and on making sure the website has everything you need to research our product and services.
But AI assistants are a better way to get that information. I don’t see a future where you’ll rely on a vendor’s website to answer your questions.
Your website should morph into a selling experience, and that means your chatbot becomes a true multimodal agent.
We’ve tested multimodal AI-SDRs. Engagement is high (~8 minutes).
Post-interaction conversion can be higher vs. standard flows.
The real aha moment comes when the agent can seamlessly switch into providing a demo of your product. Here’s an example from 1mind:
c. Prospecting Agent
From speaking to a lot of companies using AI for prospecting, I’d say +80% or more of them are unhappy with the results.
I’d split prospecting into two core buckets:
i. Email Only: This is where you can do the majority of your prospecting (inbound/outbound) via email only. We’ve found this is true when prospecting to smaller companies.
We’ve seen a +30% increase in meetings booked from AI prospecting across email.
ii. Mult-touch sequences: This is where you integrate AI into how a sales team does prospecting across sequences, eg, email + call + social outreach.
We’ve so far seen a +10% increase in outcomes for teams using the AI functionality we’ve built.
Some learnings:
- Rep Tone > Marketing Tone: Early drafts sounded like marketing; co-creating tone with reps drove adoption and trust.
- Prompts as Blocks: Breaking prompt chains into modular blocks (intent → tone → message → evaluation) reduced hallucination and made messaging consistent.
- Context & Data Matter Most: Performance rose when the AI pulled live CRM + call data — prompting skill mattered less than data quality.
- Treat Copy Like a Product: Weekly reviews and A/B testing of prompts improved replies > 30 % week-over-week.
- Adoption Is a Skill: Training managers to coach AI usage was as critical as prompt tuning.
d. Discovery Agent
The last agent I’ll cover is very experimental.
Coming back to the metric I shared above - MRR per human engagement.
What if your first conversation after booking an email with a rep was an AI assistant who could answer all your questions, gather context, and pass it on to the sales rep to ensure their call was more valuable?
It should improve both deal velocity and closed-won rates.
This is an experiment we have running now, and it is another place I feel AI adds a lot of value.
AI-enabled sellers will need an entire post to itself. It’s hard!
We’ve never been running so many experiments and pilots. Every part of the GTM journey is being rethought with AI.
An incredible time for those who are curious and enjoy building new playbooks vs. implementing the old.
Until Next Time,
Happy AI’fying
Kieran