Sixty percent of Google searches now end without a click on any website (HubSpot, “HubSpot Introduces the Loop” press release). The B2B buyer has their answer before reaching you, or they never come. Meanwhile, they move between TikTok, Reddit, podcasts and five to eight channels on average depending on the organization. The linear marketing funnel, the one that assumes a visitor who arrives, fills in a form and moves forward step by step, was designed for a world that no longer exists.
HubSpot responds to this observation with Loop Marketing, a four-stage framework (Express, Tailor, Amplify, Evolve), and with Breeze Assistant, the platform's conversational AI. Since the announcement, confusion has set in in many of our conversations with marketing teams: the Loop is treated as one more feature, something you would switch on in the portal settings. That is not what it is. The Loop is a mental model, a way of reading your own marketing. Breeze Assistant is a tool. Confusing the two means adopting a vocabulary without changing anything in your practice.
This article sorts things out: what the Loop actually covers, what is repackaging and what is a real shift, what adopting it changes in practice in how a team works, and where Breeze Assistant fits in.
The classic inbound playbook rests on a simple assumption: the prospect searches, finds content, fills in a form, enters a nurturing pipeline. Each step is measured, optimized, repeated. This model produced excellent results for fifteen years.
It also assumes that search happens on an engine that returns links. This is no longer systematically true. A growing share of buying questions are answered directly in a conversational AI, without going through a third-party site. The buyer who asks ChatGPT their question receives a summary, not a list of results to explore.
The second break concerns channels. 52% of brands now manage five to eight marketing channels at the same time, and 17% manage more than eight (HubSpot, State of Marketing). And 71% of marketers say they struggle to follow how their buyers move from one platform to another (same source). This is not a tracking tool problem. It is a mental model problem: we keep drawing a one-way funnel to describe behavior that constantly loops between search, comparison, AI and coming back.
HubSpot's State of Marketing 2026 adds a figure that shows the scale of the moment: 61% of marketers believe they are living through the biggest disruption in their profession in twenty years (HubSpot, State of Marketing 2026). This is the context in which the Loop arrives. Not as a product. As an attempt to give back a way of thinking to teams whose previous model no longer describes reality.
Let us start by clearing up the confusion. There is no “Loop Marketing” button in HubSpot. Express, Tailor, Amplify and Evolve are not modules you activate: they are the four stages of a line of reasoning. Expressing a brand position and a campaign angle (Express). Adjusting the message to real segments rather than theoretical personas (Tailor). Distributing where buyers actually are, answer engines included (Amplify). Feeding what you learn back into the next stage (Evolve). Then starting again, better informed.
HubSpot's CMO, Kipp Bodnar, sums up the intention: “The Loop meets customers everywhere they are, uses AI to personalize each message at scale, and turns every interaction into a learning opportunity” (HubSpot press release).
Let us be direct on a point the launch messaging does not address: taken one by one, these four stages are nothing new. Positioning, segmenting, distributing, optimizing: this describes any properly executed marketing over the last ten years. The Loop is largely a repackaging of the historical inbound methodology (Attract, Engage, Delight), reworded for the era of answer engines. It needs to be said, because selling it as a revolution would insult the intelligence of teams that already practice all of this.
Two shifts are nevertheless real, and they deserve attention.
The first: the Loop acknowledges invisible touchpoints. A recommendation in a ChatGPT answer, a mention in a Reddit thread, a podcast episode listened to in the car: none of these contacts will ever appear cleanly in an attribution report. The classic funnel acted as if these contacts did not exist, since it could not measure them. The Loop integrates them as first-rank realities, to be worked on (this is the whole point of AEO) even when they cannot be traced individually.
The second: looping becomes the working rhythm itself, not an optional best practice. In the funnel, learning comes in the end-of-campaign review, when everything has already been decided. In the loop, it is supposed to come continuously and modify the campaign while it runs. This is a difference in tempo, not in vocabulary. And this is precisely where AI changes things: looping quickly was theoretically possible before, but the cost of execution (producing variants, analyzing signals, adjusting) made it unrealistic for most teams.
If the Loop were only a vocabulary, it would not deserve an article. If you take it seriously as a framework for analysis, it requires three changes in how a marketing team actually works. These three changes show genuine adoption, and their absence signals mere window dressing.
A campaign designed as a funnel is a project: a start, a launch, an end, a review. A campaign designed as a loop is a system that runs: hypotheses set at the start, regular checkpoints, adjustments along the way, and a body of learning that feeds the next one.
In practice, this shows in the calendar. A team that has really adopted the loop has short, frequent campaign reviews (weekly or twice a month) with a decision-making mandate: cut a channel, reallocate a budget, test an angle. A team that has adopted the vocabulary has kept its quarterly review and renamed it Evolve. The difference comes down to one question: when did you last modify a running campaign, based on a signal, before its end date?
This change of rhythm has a real organizational cost that HubSpot's messaging downplays: looping requires decision-making availability. A loop that runs with nobody to arbitrate the signals produces reporting, not learning. This recurring arbitration role, more than production, is what structures our growth agency support: reviews are for deciding, not for noting.
In the loop, each interaction is supposed to feed the next. This sentence, appealing in a keynote, has a down-to-earth condition: interactions must be captured, linked to the right contact, and usable. A CRM database full of duplicates and empty fields does not loop anything. It accumulates.
This is where the concept meets the platform, and where our field experience joins the vendor's message one step ahead: before running a loop, you need data to loop on. We have explained why CRM data enrichment must come before any advanced automation; everything said there applies here even more strongly, since the loop amplifies what it is given. Wrong segments iterated quickly remain wrong segments. They just become wrong faster.
The third change concerns roles, and it is the most uncomfortable. In a loop equipped with AI, production (content variants, segment adaptations, performance analyses) is largely delegated. What remains human is what has always been the core of the profession and had been buried under execution: positioning, angle, arbitration.
The State of Marketing 2026 puts it well: the gap is no longer about who uses AI, but about the quality of its use. 80% of marketers already use it to create content (HubSpot, State of Marketing 2026). When everyone has the same production tool, the differentiator moves up a level: it is the brand's point of view, the ability to say something a competitor equipped with the same assistant will not say. A team that delegates the Express stage to AI gets a brief that is technically complete and strategically interchangeable. The loop will run. It will run empty.
An honest framework includes its blind spots. The Loop has one, and it is structural: it normalizes invisible touchpoints without offering a way to measure them. This is consistent (you cannot track a spoken recommendation from ChatGPT), but the consequence has to be drawn rather than avoided: in loop marketing, fine-grained channel attribution partly becomes an accounting fiction.
Our position on this is firm: chasing perfect attribution in 2026 is a lost battle, and teams that devote their energy to it take that energy away from what matters. Better to shift measurement up a level: less “which channel generated this click”, more “is the pipeline attributable to marketing growing, and which cohorts convert better”. The loop is managed by trend and controlled experimentation, not by last click. For many marketing leaders raised on channel-by-channel reporting, this is a loss to accept. It is also a liberation.
Then there is the tool. Breeze Assistant is the conversational assistant built into HubSpot, available to sales, marketing and service teams, which works directly on CRM data without exporting to a third-party tool. Its relationship to the Loop is exactly that of a tool to a method: it lowers the cost of execution; it does not provide the content.
For a marketing team, three uses concentrate most of the value.
Generating and adapting content. Email drafts, article outlines, campaign briefs, respecting the portal's brand settings. The output can be published after review, not as is: it is a quick draft, not a final copy.
Guidance within the Loop framework. The assistant draws on HubSpot Academy content, website analytics and the Loop framework to help define an ICP (Ideal Customer Profile) and structure a brief according to the four stages. Useful for asking the right questions in the right order. Not enough to answer them for you, for the reason seen above: interchangeable strategic answers are typical of a tool everyone has.
Querying CRM data. Asking a question about a campaign's performance or a segment's behavior directly in HubSpot, and getting a sourced answer without exporting to a spreadsheet. This is the use that most directly serves the loop: it shortens the time between signal and decision, in other words the cycle time of the loop itself.
Two limits shape what Breeze Assistant can actually produce. First, the quality of the answer depends directly on the quality of the portal's data and permissions: a poorly governed database produces wrong answers stated with the same confidence as a right one. Second, the assistant is anchored in CRM data and HubSpot resources: it will not fetch information from an unconnected external source.
The most frequent mistake we observe is using Breeze Assistant in the hope that it will structure the strategy after the fact. The reverse order works better.
This sequence is not original in itself. Its merit is that it does not confuse adopting a vocabulary (“we do Loop Marketing”) with adopting a practice. The first takes one meeting. The second takes two quarters. For the whole platform part (Breeze configuration, portal governance, reporting rhythm), this path is the focus of the support provided by our HubSpot agency.