What Is Actually Changing in B2B Marketing

Forces, Not Predictions, And Why That Matters

Five forces are genuinely reshaping business-to-business marketing, and a numbered list of predictions is the wrong container for any of them. Forces persist, change speed, and can be reasoned about while still in motion, which is what a marketing director needs when deciding what to fund next year.

So this page is organised as analysis, and it grades its own claims. Three categories are used and sorted explicitly at the end: well-evidenced, meaning traceable to primary documentation from the company or regulator responsible; contested, meaning sources disagree or the only available figures are vendor estimates measuring different things; and speculation, meaning the argument runs from first principles and no dataset settles it.

That grading is the practical difference between the two kinds of change a marketing team faces. Documented platform and legal changes have deadlines, and missing them breaks something. Directional shifts in buyer behaviour have no deadline and cannot be verified from primary sources, and they are where most budget gets wasted on the strength of a statistic nobody can trace.

AI In Search: What Google Documents, And What No One Can Measure

Google documents two generative features in Search. AI Overviews are documented as a way to get the gist of a complicated question quickly and as a jumping-off point to explore links; AI Mode is documented for queries needing exploration, reasoning or comparison. Both are grounded in Google's core Search ranking systems, using retrieval-augmented generation against the Search index plus query fan-out. Fan-out is the genuinely new mechanic: a page can surface for a question the user never typed.

Google's documentation also states that you do not need to create new machine readable files, AI text files, markup or Markdown to appear in generative AI search, that structured data is not required, and that there are no additional requirements to appear in AI Overviews or AI Mode. Controls are the existing ones: nosnippet, data-nosnippet, max-snippet, noindex, robots.txt, plus Google-Extended for training and grounding elsewhere. There is no documented way to stay in classic Search while being excluded from AI Overviews without losing normal snippets.

What nobody can tell you is the traffic effect. No primary source publishes a figure for AI Overview traffic impact or click-through decline. Vendor estimates run from AI referrals being about one percent of all web traffic to a claimed 92 percent share within AI referrals, and they measure different things.

Content Abundance And The Premium On What A Machine Cannot Make

Adequate prose is now effectively free. When producing a competent summary of public information costs almost nothing, the market value of that summary falls with it, and most B2B content marketing has been producing exactly that for fifteen years.

What retains value is what a model cannot generate:

• Proprietary data. Numbers from your own installed base, transactions and support tickets.
• Named expertise. A position taken by a person with a traceable record, who can be challenged and who bears the cost of being wrong.
• Original testing. You ran the thing, measured it, and published the method with the result, including where it failed.
• Judgment on a specific situation. Not what the options are, but which is wrong for a company of a given size, and why.

The corollary is not to write for machines; Google documents that no AI-specific formatting is required. It is to have something a machine could not have written, and to stop paying for things it could. Grade this one honestly: speculation, argued from first principles. No B2B dataset establishes that buyers have re-weighted how they value content.

The Identifier Retreat, And The Reversal Nobody Planned For

The industry spent three years planning a migration that did not happen. Chrome is not deprecating third-party cookies. On 22 April 2025 Google stated it would not roll out a new standalone prompt for third-party cookies and would maintain its current approach, and it reaffirmed that on 17 October 2025. There is no deprecation date.

Google's October 2025 update instead lists Attribution Reporting, Topics, Protected Audience, Private Aggregation and Related Website Sets among the technologies being wound down. What continues is the privacy plumbing: CHIPS, FedCM, Storage Access, Fenced Frames and Private State Tokens. This is the largest body of incorrect published B2B marketing advice in circulation, and anything telling you to prepare for Topics or Protected Audience describes a cancelled plan.

Cookie-based retargeting still works on Chrome traffic. It does not work on Safari, which has fully blocked cross-site cookies by default since Safari 13.1 and iOS 13.4, announced 24 March 2020, and it is constrained on Firefox, which partitions cookies per site. Because the advertising APIs were retired, there is no Google-provided successor. What Google documents as durable is first-party and server-side: Enhanced Conversions with hashed data, enhanced conversions for leads, offline conversion import, and consent mode with modelling.

Privacy Law Expanding One State At A Time

Twenty US states had comprehensive consumer privacy statutes in force as of mid-2026. The direction of travel is more states, and more obligations attached to tags rather than to business models.

Three features matter to a B2B company. First, the volume thresholds most firms assume exempt them do not exist everywhere: Texas and Nebraska apply to any business operating in the state that processes or sells personal data and is not a small business, so a mid-sized manufacturer running a remarketing pixel can be in scope in Texas while out of scope in Virginia.

Second, California's temporary exemption for business-to-business contact data expired on 1 January 2023 and was not renewed. A California business contact's name, work email and job title is ordinary personal information carrying access, deletion, correction and opt-out rights, and notice at collection is owed at the point of collection, including on gated forms. The accurate framing is that a carve-out lapsed, not that California created a new B2B rule.

Third, regulators have treated advertising pixels that transmit identifiers to third parties for cross-context behavioural advertising as a sale or sharing of personal information. Colorado lists Global Privacy Control as its only valid universal opt-out mechanism, required since 1 July 2024.

Measurement Is Getting Harder, Not Easier

Every year brings more analytics features and less observed data. The features exist because the data is disappearing, and the loss lands harder on business sites than consumer ones.

• Modelling has a floor most B2B sites never reach. GA4's behavioural modelling requires at least 1,000 events a day with analytics storage denied for at least seven days, plus at least 1,000 daily users sending events with it granted for seven of the previous twenty-eight days. A site with a few thousand monthly sessions absorbs the full loss and receives none of the recovery.
• Thresholding suppresses rows. GA4 data thresholds are system-defined, unadjustable and unpublished.
• Retention is shorter than the sales cycle. The standard-property maximum for user-level and key-event data is fourteen months, the default two. That limit affects Explorations and funnel reports rather than standard aggregated reports, a distinction most published advice gets wrong.
• History was deleted, not archived. Universal Analytics data was deleted starting the week of 1 July 2024.

Label which of your numbers are observed, modelled and inferred, because the alternative is defending a modelled figure as though you measured it.

What Has Not Changed Since 2003

This site has published on corporate website marketing since 2003, and a short list has survived every platform change since.

• A committee buys, and it needs different things at different moments. Gartner's 2025 survey of 632 B2B buyers found buying teams of five to sixteen people across as many as four functions.
• The site still has to say what you sell and who it is for, in language a buyer would use. The most common serious problem on corporate websites in 2026 is the same one it was in 2006.
• Speed, crawlability and indexability are preconditions, not tactics. A page that cannot be retrieved cannot be ranked, summarised or cited by anything.
• A form that routes to a person who answers quickly beats a cleverer campaign.
• Reputation and referral originate more B2B revenue than any channel you can buy.

This is not an argument for ignoring what is changing. It is an argument for funding the fundamentals first, because a company failing at them will not be rescued by early adoption of anything.

Which Of This Is Evidence And Which Is Guesswork

Well-evidenced, from primary documentation. Google's statements that no special files, markup or formatting are required for its AI features. Chrome's reversal on third-party cookies and the retirement of most Privacy Sandbox advertising APIs. The June 2026 move of conversion uploads to the Data Manager API. GA4's modelling thresholds, retention limits and unadjustable data thresholds. Act on all of these; several carry deadlines.

Contested or unmeasured. The size of any AI-search traffic effect, in either direction. Whether structured data influences inclusion in AI answers; Google documents it as not required, which is not the same as saying it does nothing. The share of referrals arriving from assistants, where every figure is a vendor estimate of a different denominator.

Speculation. That content abundance is repricing proprietary data and named expertise.

The discipline that follows is unglamorous and works: read primary documentation rather than commentary about it, re-check anything carrying a date, keep a list of the platform deadlines affecting your stack, and decline any service whose business case rests on a statistic you cannot open the source for.

Frequently Asked Questions