The B2B Marketing Table Has Changed

A while back, I went to visit an old friend who has spent more than a decade in B2B marketing. Before we'd even sat down, he opened with a sigh: "The more I spend, the more anxious I get."

Nervous about what, I asked?

He said: three things.

First, website traffic is still there, but inbound inquiries keep shrinking. Second, his budget went up this year, yet the moment his boss asked, "Which deals, exactly, did all this money bring in?" — he went quiet. His exact words: "I know these programs are doing something. I just can't prove it in a report." Third — the strangest one. A client they just closed had, all told, ten-plus different people reading their content and receiving their emails, and to this day his sales team can't explain how that deal actually came together inside the account.

When he finished, I told him: your craft hasn't slipped. The table has changed.

It isn't one card played badly — several cards got swapped out, all at once.

1. How Buyers Find You Has Changed

Let's start with the biggest one.

How did B2B buyers used to find vendors? Search engines. Whoever packed in the most keywords, stacked up the most backlinks, and loaded pages the fastest ranked first. Which is why, for the past decade, the core craft of the B2B marketer was SEO.

And now? Here are some numbers: 79% of B2B buyers no longer start their research on a search results page — they start with AI tools, ChatGPT, Perplexity, and Google's AI summaries among them. AI summaries now appear in 13% of search results. And 57% of searches end with zero clicks.

What does zero-click mean? It means the AI answered the buyer's question right there on the results page. That buyer never entered your website.

This means the role of search has changed. The old search engine was a directory — it lined up web pages in order for you. AI search is a synthesis engine: it reads dozens of sources on its own, distills, summarizes, recombines, and hands you one "definitive answer."

So what you're fighting for is no longer "ranking first" — it's "being cited." When the AI finishes answering, are you in the footnotes?

Then how do you get the AI to cite you? A few things: structure your content — FAQs, comparison tables, checklists — so machines can parse it easily; go deep — one 3,000-word piece that covers a subject thoroughly beats ten shallow 300-word posts; be transparent about sources — link every number back to a primary source; and there's one more, called "entity prominence" — state your brand name clearly and repeatedly throughout, so the AI registers you as a definite entity on this topic, not a vague "we" or "the platform."

But there's a trap here: machines are the ones crawling your content, but humans are the ones signing the deals. If you write it like Wikipedia — every paragraph an entry, no opinions, no stories — the AI might be willing to cite you, but people will read it and feel nothing.

Smart teams build two-layer content: one layer for machines, structured and citable; one layer for humans — case studies, judgment calls, and plain human language.

2. The Person to Persuade Went from One to a Committee

The second change is hidden inside the customer's own company.

B2B used to work like this: win over one or two key people and the deal was basically yours. Now? The buying decision has moved from a few key individuals to 5 to 16 people signing off together — the users, the technical owners, the money people, the final signature — each with their own set of considerations. And 74% of buying committees argue among themselves.

Many teams react: easy — just personalize. Send the CMO brand content, send the CFO a cost calculator, send the CTO a security whitepaper. How thoughtful.

Gartner did the research, and the results sting: precision-personalized content aimed at individuals actually damages consensus 59% of the time.

Why? Think it through. The CMO's materials say this thing lowers acquisition costs; the CFO's say it's a new expense; the CTO's say it carries integration risk. Every one of them received content "custom-made for them" — and now the three of them are not picturing the same solution at all. Consensus is further away than before.

The correct approach is the opposite: unify the narrative first, then layer in role-specific detail. Everyone is clear that the core value proposition is the same sentence — say, "cut acquisition costs by 40%" — and then the CMO gets the brand-level footnotes, the CFO gets the budget arithmetic, the technical lead gets the implementation timeline. Build relevance around the buying group, and Gartner's data shows consensus improves by 20%.

On one side, a 59% negative effect; on the other, a 20% positive one. Get the direction backwards, and the harder you work, the fewer deals you close.

This is why the playbook of account-based marketing (ABM) has been entirely rebuilt. It used to look at "contacts"; now it looks at "committees." You have to be able to say, for your key accounts' committees, who speaks up for you internally, who owns the budget, who owns security, and who uses the product every day. Then arm your internal champion with ammunition: a one-page comparison sheet, an ROI calculator, a demo video that can be forwarded straight to the skeptics.

I've come across a gut-check that stings: can you name the committee members at your top 20 key accounts? If your "personalization" is just replacing {first name} in an email, that's not ABM — that's a mail merge.

And don't get greedy. Pick 10 to 25 accounts first, get the process working and the return calculated, then expand. An "ABM" that spreads across 500 accounts out of the gate decays into ordinary ad spend before long.

3. AI Went from Advisor to Executor

The third change is about how AI gets used.

96% of marketers are already using AI. But "using AI" is forking into two paths.

One group uses AI as an advisor: ask for a number, draft a piece, produce a table.

The other group has started using AI as an executor — in industry parlance, an AI agent. It doesn't just answer questions; it watches the signals itself: who's on the pricing page, who downloaded the whitepaper, who's been active in the CRM lately. The moment it spots several people from the same company looking at your product at the same time, it automatically flags it: this company's buying committee is forming. Then it decides the next step itself: send a LinkedIn DM to the VP, send technical documentation to the engineer, send an ROI calculator to the CFO. When an ad stops running well, it pauses it itself; when budget should shift, it shifts it itself.

Impressive. Products on the market have already made this real. A marketing data platform called Improvado built an AI agent wired into over a thousand data sources. The marketing manager asks in plain language: this quarter, which ad channels are showing positive returns across Google Ads, LinkedIn, and our own event registrations? It automatically translates that into SQL, queries the data warehouse, produces the charts — and takes follow-up questions, drilling deeper on request.

Sounds wonderful. But I have to be the one to pour some cold water on it.

Forrester has a prediction: ungoverned generative AI will cost B2B companies more than $10 billion — stock price drops, legal settlements, and regulatory fines all included.

How does that $10 billion get lost? AI, working from wrongly attributed data, doubles the budget poured into a channel that "looks profitable"; AI learns from your history of closed deals, and absorbs the biases inside that history along with it, forever missing new markets; AI-generated email goes out built on data the customer never authorized, and crashes into GDPR.

So remember this: AI doesn't clean dirty data. AI only amplifies dirty data.

AI is an amplifier. A good strategy, it speeds up. A bad one, it speeds up too.

So when can you bring in agents? Pass three checks first: Is your data connected end to end? Is data quality stable — is the quarterly degradation rate within 10%? Is there someone who can verify the conclusions the AI produces? Pass all three, then deploy. Fail one, fix that one first.

4. The Yardstick for Measuring Results Has Changed

The fourth change hurts the most, because it reaches into marketers' own KPIs.

What's an MQL? A marketing qualified lead — someone who downloaded a whitepaper and filled in a form. For the past decade and more, marketing departments turned it in as homework.

Where's the problem? The person who downloaded the whitepaper may never take a sales call; and a referral from an existing customer — the MQL system doesn't record that at all — closed in 14 days. The thing you spent all that effort measuring and the thing that actually produces revenue are, for all practical purposes, two different things.

More awkward still: this year 56% of marketers expect budgets to rise, yet 90% of teams can't work out attribution, and 25% can't measure ROI at all. The average sales cycle is 10.1 months, one deal has dozens of touchpoints, and the data is still scattered across ad platforms, CRMs, and event tools. That's exactly where my friend's opening line — "I can't prove it in a report" — comes from.

So the leading teams have swapped the yardstick. They no longer count MQLs; they watch harder numbers: pipeline velocity — from first touch to a real sales opportunity, how many days; brand-assisted deal size — customers who touched your brand content versus customers who only saw the ads, whose deals are bigger, whose cycles are shorter; and the time lag from intent peak to sales follow-up — in the few days the customer was hottest, did sales catch them?

By the way, marketing and sales should stop running as two separate teams, too. 44% of B2B marketing leaders say their biggest challenge is precisely that these two teams are out of alignment. The fix is wonderfully unglamorous: sit down once a week and reconcile the numbers — which new accounts entered the pipeline this week, which deals won and why they won, which leads are junk, what battle to fight next month. Sharing one yardstick beats ten pep rallies.

But before you swap the yardstick, fix the foundation. Here's a simple self-check: go into the CRM and look — how many deals have lain untouched in the "initial contact" stage for 90 days? How many closed deals have no recorded source at all? If the list runs long, your problem is not that your strategy isn't advanced; it's that your data isn't clean. Scratch beneath most strategy problems and you'll find a data problem.

One more piece of the foundation: first-party data. Third-party cookies are dead, and the old playbook of "renting traffic" died with them. From here on, what counts is the data you accumulate yourself: form fills, login behavior, event check-ins, product usage records. And this ledger doesn't wait: email lists decay naturally at 22% a year — skip quarterly cleaning, and your targeting precision collapses within a year.

5. Content's Winning Move Has Shifted from Creation to Distribution

The fifth change concerns content itself.

One prediction says 82% of all web traffic this year will be video. The number sounds scary, but the conclusion is counterintuitive: precisely because everyone has video now, the ordinary product demo video has become background noise. What actually separates the winners is interactive product demos, one-to-one screen videos recorded for key accounts, and proprietary research turned into short data-visualization films. A rotten video hurts the brand more than no video: grainy visuals, muddy audio, a script that fits every company and none — viewers won't say it out loud, but in their heads they've already graded you down half a tier.

Distribution is the same. A $10,000 deep research report with zero promotion budget loses to $4,000 of content plus $6,000 of distribution. 60% to distribution, 40% to production — not the other way around.

Then there are chatbots. 97% of B2B companies have deployed them, yet most are still that same "Hello, how can I help you today?" A genuinely useful bot recognizes which company the visitor is from and where they are in their journey: enterprise customers get routed straight to sales, smaller ones to self-serve trials. But discretion is the bottom line: if the person never volunteered who they are, then your "I see you're with Such-and-Such Inc." isn't attentive — it's creepy.

6. The Source of Trust Has Moved Back to "People"

The sixth change is the most interesting. At the very moment everyone is talking about AI, the most "primitive" things have come back.

The first: the voices of employees and experts. Have you ever wondered — when a buyer goes to ask the AI, who does the answer end up citing? Analysts, frontline practitioners, industry experts — anything but the copy on your homepage. In the AI era, your visibility depends on whether anyone outside your company speaks for you. 75% of large B2B enterprises are adding expert-relations budget this year.

But this road has a hidden trap: 75% of employee advocacy programs die in compliance. An employee speaking for the company must disclose who they are, cannot casually promise product results, and finance and healthcare carry additional regulatory gates. So this must have a dedicated owner — 0.5 to 1 full-time hire. Hand it to a growth manager as a "side project," and it's basically dead on arrival.

The second: the small, offline dinner table. 49% of B2B companies are adding offline event budget this year. Note that the math has to be redone. The same $150,000: one big booth can scan two to three hundred business cards, of which maybe 10 to 15 are real conversations with decision-makers, ending in 2 to 3 follow-up meetings. The same money, spent on ten 15-person executive dinners: 150 hand-picked guests, all qualified to be in the room, 90 minutes of conversation over one meal, and follow-up meeting conversion rates reaching 60% to 70%.

The big booth is a transaction — scan a badge, walk away with a pen. The small dinner table is a relationship — after one meal, you have a shared experience.

The third: community. Slack, Notion, and Figma never poured money into ads early on — they snowballed on their user communities. Communities have a 90/10 rule: 90% of the content is users helping each other and exchanging tactics; only 10% is left for your own product. Treat the community as an ad slot and it dies overnight. And don't overestimate the starting point: if active members can't reach 500, the community can't warm up.

Technology amplifies scale, but what closes the deal is still trust.

7. Transparency Has Become the New Ticket In

The seventh change is one many people haven't yet registered: AI transparency is becoming the ticket into procurement.

Enterprise procurement used to ask about features, pricing, SLAs. Now, procurement teams at large companies are adding questions directly into their RFPs (request-for-proposal documents): What data was your model trained on? Please provide data lineage documentation. What is your bias testing methodology? If you can't answer, you never make it to technical evaluation.

The procurement teams ask for a reason. The EU AI Act takes effect from 2026, classifying AI systems by risk level; high-risk applications must run risk assessments, keep human oversight in place, and have data governance documentation ready. US state rules are fragmented, but they're converging on the same transparency demands.

So here is a practical suggestion for all B2B vendors: prepare an "AI transparency document" early, something sales can produce on the spot during procurement. It should contain: model cards — what this AI does, what data it uses, what it doesn't do; training data sources; bias testing results; where the human review step sits; a data flow diagram; certifications like SOC 2 Type II. Platforms like Improvado keep SOC 2, HIPAA, GDPR, and other certifications on file year-round, ready to hand over whenever procurement asks — they've made this step table stakes.

Building this set costs 3 to 6 months up front. Worth it?

Worth it. In the short term it's a cost; in the long run it's a moat. Competitors who skip the paperwork today ship faster; tomorrow, procurement asks one question — and one question knocks them out.

8. Knowing When Not to Follow Is Worth More than Following

The last one, and the one I most want to remind you of: everything above is a "does this fit us" question, not a "must do" item.

Account-based marketing suits businesses with average annual contract value above $50,000 and sales cycles above 30 days. Under $50,000 in contract value, running an ABM operation costs more than the gross profit it brings; under a 30-day sales cycle, the committee never has time to form.

Agentic AI suits teams whose data passes inspection and who have analytical talent. For a company that can't sort out its own data, deploying agents is bolting a turbocharger onto a dirty machine. And under a $100,000 budget, don't rush — implementation plus platform costs will eat up six months of returns before you see any.

Personalization suits a scale of at least 500 contacts per segment. With fewer than 500, the model you build learns noise.

Community suits businesses whose product itself has network effects. Force a community onto a product whose users have no need to talk to each other, and all you'll harvest is a dead group chat with a few dozen members.

Have you noticed — every one of these reasons to "not follow" is an honest look in the mirror at your own stage and size. Knowing when to say "this one doesn't apply to us" is worth far more than the enthusiasm to learn everything.

Back at That Table

At the end of that day, I told my friend: don't panic. The table has changed, but everyone is still drawing the new cards, and nobody has completed a winning hand yet.

He went back and did two things: he pulled the data scattered across eight systems into a single warehouse, and he mapped out the committee seating chart for his 25 key accounts, one account at a time. Three months later he told me: inbound inquiries hadn't exploded, but his sales team said that now, every call they made actually got picked up.

This change, in the end, boils down to two lines: buyers ask the AI more often than they ask your salespeople, and the people making the call have gone from one person to a crowd. But take the long way around and you'll find that the moment a deal finally closes still happens in the trust between people.

The tools change. The yardstick changes. People don't.

Here's to you: may you not only read the new table, but take the seat at it that's yours.