Somewhere in most 2027 marketing plans being finalized this month there is a website target carried forward from the previous year: organic sessions up some percentage, new users up some percentage, a conversion rate held flat. For a large number of B2B companies that target was missed in 2026, sometimes by a wide margin, and the explanation offered in the quarterly reviews was the same each time. Search engines now answer the question on the results page, AI assistants answer it without a results page at all, and fewer people click through.
That explanation is largely correct. The mistake is what usually follows it. Either the team sets the same target again and hopes for recovery, or it quietly drops website metrics from the plan because they have become embarrassing. Neither is a good outcome. The first guarantees another year of missed numbers and defensive reporting. The second leaves one of the largest fixed investments in the marketing budget without any agreed measure of whether it is working.
The website has not become less important. What has changed is which visitors arrive, why they come, and what they need when they get there. The 2027 targets should reflect that, and setting them properly requires some analysis most teams have not yet done.
Find Out Which Traffic You Actually Lost
The headline figure, organic sessions down year over year, hides most of what matters. When I have gone through this exercise with teams, the loss is almost never spread evenly across the site. It is concentrated, and the concentration tells you whether to worry.
Start by splitting organic landing pages into groups by the kind of query that used to bring people to them. The categories do not need to be elaborate. Four usually suffice.
Informational and educational content. Definitions, explainers, “what is” articles, how-to guides on general topics, glossary pages. This is where most of the decline sits for most B2B sites, because these are exactly the questions an answer engine handles well. Much of this traffic was never commercially valuable; a good share came from students, job seekers, and people in unrelated industries.
Problem and solution research. Pages that address a specific business problem your product category solves, comparison and alternatives pages, buyer’s guides, and evaluation frameworks. Traffic here has usually declined too, but by less, and the visitors who still arrive tend to be further along.
Brand and product. Your homepage, product pages, pricing, integrations, and anything people reach by searching for your company or product name. In most of the analyses I have seen, this traffic is flat or growing. Buyers who learned about you in an AI assistant still come to verify what they were told.
Customer and technical. Documentation, support content, API references, login pages, release notes. This is often a large share of total sessions and is frequently mixed into marketing’s numbers without anyone noticing.
Once the traffic is split this way, the conversation changes. A site that lost forty percent of its organic sessions, almost all of it from a glossary and a blog archive that never produced pipeline, is in a very different position from a site whose pricing and product pages are losing visitors. The first is a reporting problem. The second is a demand problem, and it should be treated as one.
Do the same split for conversions. In most cases you will find that the pages losing traffic were contributing a small share of qualified conversions, and that the pages holding steady were contributing most of them. That is the evidence to take to the CFO before the plan is approved.
Separate the Traffic Problem From the Visibility Problem
There is a second distinction that matters for planning. Losing a click is not the same as losing the buyer’s attention.
When an AI assistant summarizes the evaluation criteria for your category and names three vendors, including you, the buyer has encountered your company without visiting your site. When it names three vendors and you are not among them, you have lost something considerably more important than a session. Traditional web analytics cannot see either case. Both show up as an absence.
For 2027 the plan should include some measure of visibility in AI answers alongside the website metrics, owned by the same person, because they are now two halves of the same function. The tracking tools are immature: most run a defined set of prompts across the major assistants on a schedule and record whether and how you are mentioned, and the results are noisy. I would not put a hard numerical target on them yet. But a consistent set of thirty or fifty prompts that reflect how your buyers actually research, run monthly, will show the direction of travel and expose the gaps. If you are absent from answers about the problem your product was built to solve, that is a content and positioning issue, and no amount of website optimization will fix it.
The other useful signal is direct and branded traffic. If buyers increasingly meet you in an assistant and then come to the site by typing your name, branded search and direct visits should rise even as non-branded organic falls. Watch the two together. A fall in non-branded traffic with a rise in branded traffic is the expected pattern of the transition. A fall in both is a warning.
Decide What the Site Is For Now
For fifteen years, the default model of a B2B website was a funnel with a wide top. Publish a large amount of content, attract as many visitors as possible, and convert a small percentage through gated assets and forms. The economics of that model depended on cheap, plentiful organic traffic at the top. That supply has contracted, and it will not come back.
The visitors who still arrive have, on the whole, already done much of their research elsewhere. They come to the site to confirm, to compare, to find what the assistant could not tell them, and to take a next step. That changes what the site needs to do well.
It needs to answer specific, verifiable questions quickly. What does it cost? Does it integrate with the systems we already run? Where is the data stored, and what certifications do you hold? How long does implementation take, and who does it? What does the contract look like? Buyers who come to verify want these answers on the page, not behind a form or in a sales call. Sites that hide pricing and technical detail are losing evaluators who have plenty of alternatives one click away.
It needs to be the authoritative source. The assistants that summarize your category draw on your site, among other sources. Pages that state clearly and specifically what your product does, for whom, and with what limitations, are more likely to be represented accurately. If your product page says you “empower organizations to unlock value,” the model will fill the gap with whatever a competitor or review site has said more clearly.
It needs to support the buying committee, not just the first visitor. The person who found you is rarely the only person who evaluates you. Security reviewers, procurement, finance, and the technical team will each visit with their own questions. Pages written for those roles, a security and compliance hub, an implementation overview, a total cost explanation, tend to be undervalued because they do not attract much traffic. They matter far more than their session counts suggest.
It needs to make the next step easy and appropriate. The form that asks for twelve fields before a buyer can see a product tour was always a poor trade. It is a worse one when the visitor arrives already informed and impatient. Self-guided demos, published pricing for standard tiers, and a direct route to a person for those who want one are worth more than another gated report.
None of this is new. What has changed is that the traffic economics that let sites avoid it have gone.
Retire the Content That No Longer Earns Its Place
A practical consequence of this analysis is that most B2B sites carry a large amount of content that no longer does anything useful. Blog archives running to thousands of posts, many outdated, some contradicting current positioning. Glossary pages that were written for search engines. Thin landing pages left over from campaigns that ended years ago.
This content was always a maintenance burden. It is now a liability, because answer engines may still draw on it. An article from 2019 describing your product’s old architecture, or a pricing explanation from three packaging changes ago, can surface in an assistant’s answer about your company. The buyer has no way of knowing it is out of date.
The fourth quarter is a reasonable time to deal with this, before the new year’s content production starts. The process does not need to be sophisticated. Sort pages by traffic, conversions, and inbound links over the past year. Anything with negligible traffic, no conversions, and no meaningful links is a candidate for removal or consolidation. Anything that makes a factual claim about your product, pricing, or company should be checked for accuracy regardless of its traffic. Redirect what you remove to the most relevant current page.
What to Put in the 2027 Plan
With the analysis done, the website section of the plan can be rebuilt around measures that reflect how the site now earns its keep.
First, replace the total organic sessions target with segmented targets. Hold or grow traffic to brand, product, pricing, and evaluation pages. Accept, explicitly and in writing, that informational traffic will continue to decline, and stop reporting it as a headline figure. Remove customer and documentation traffic from marketing’s numbers entirely, or report it separately.
Second, measure outcomes per visit rather than visit volume. Qualified conversions from the site, pipeline from opportunities whose contacts visited high-intent pages, and the proportion of opportunities in which buying committee members other than the first contact visited the site. They are harder to produce than sessions, but they measure what the site is for.
Third, add AI answer visibility as a tracked measure, with a defined prompt set and a monthly review, without a numerical target for the first year. Pair it with branded and direct traffic so the two trends can be read together.
Fourth, put a content retirement and accuracy review into the calendar, quarterly rather than as a single project. Assign an owner. Treat factual pages about the product and company as content that requires review at every significant release, the same way product documentation does.
Fifth, look hard at the website budget itself. Many teams are planning a redesign for 2027 because the current site feels outdated and the numbers are falling. Before approving one, check whether the problem is the design or the content. A redesign that moves the same vague product copy and gated assets into a new template will not change the outcome. Spending the money on rewriting the twenty or thirty pages that buyers actually rely on, and on making pricing, technical, and security information available without a form, will usually do more.
Finally, prepare the explanation for the leadership team before the plan is presented, not after the first quarter’s numbers come in. A CFO who is told in January that website traffic will decline by design, and is shown why that decline does not reduce pipeline, will accept it. A CFO who discovers it in April, from a dashboard, will reasonably wonder what else marketing has not mentioned.
The organic traffic that B2B marketing relied on for a decade and a half was in part an accident of how search engines worked. That arrangement is ending. The buyers have not gone anywhere, and they still need a place to verify claims, compare options, and find out what it will actually cost to do business with you. A site built for that purpose can do well in 2027 with considerably fewer visitors, provided it is measured on the right things.