A small company can now look like a large one in a matter of days.
The website, the product video, the deck, the case studies, and the page of customer quotes used to take a team, a budget, and months of work. They now take a few days and almost no money. Some of what gets produced is true, and the buyer has no way to tell which part.
Buyers know this. They have less reason to give any of it credit.
The cost of appearing credible is approaching zero. The cost of being credible has not moved. That is the change this paper is about. Presentation used to be evidence. A company that looked established usually was, because looking established was expensive. That link is broken, and buyers are adjusting faster than the companies selling to them.
They decide earlier. They lean harder on the vendors and people they already know. When the stakes are real, they want a person who can be held to what was said.
I wrote in April that distribution in HR tech is a trust problem first and a sales problem second. HR tech has been bought through peers, platforms, and advisors for as long as I have worked in it. The research below shows large B2B purchases across technology, services, and manufacturing being made the same way.
I want to say one thing about cause before the evidence. Many people want to pin this on AI. I do not think it needs to be pinned on any one thing. Trust in institutions has been falling for years, budgets are tight, and every buyer is buried in content. AI adds to all of that by putting a polished first impression within anyone's reach. What matters to an operator is the behavior, and the behavior has changed.
Getting a meeting has never been harder
Ask any seller what changed in the last two years and you will hear the same answer: nobody picks up.
The data backs them up. In The Bridge Group's 2025 survey of sales development teams, 60% of reps hit quota, the lowest share in the study's history. Companies have cut the monthly meeting quotas for those reps by 40% since 2018. Belkins tracked 7.5 million cold emails sent in 2025 and got a reply to fewer than one in two hundred, with the rate falling from 0.50% in the first half of the year to 0.40% in the second. Its calling data from the same year works out to roughly one booked meeting for every 370 dials. And in Gartner's 2024 buyer survey, 73% of B2B buyers said they actively avoid suppliers who send irrelevant outreach.
Some of that is volume. Inboxes are fuller, spam filters are stricter, and plenty of outreach is badly aimed. The rest is the trust problem showing up at the front door. A stranger's email looks like every other stranger's email, so buyers have stopped answering strangers. The meetings still happen. They come through someone the buyer already trusts.
We lost a deal at one of our portfolio companies that came down to exactly that. The competitor's original introduction came from someone at the buyer's private equity investor. From then on, nothing we could put in a deck was going to outweigh a recommendation from the firm that owns the company.
Looking good is cheap now
Economists described this problem more than fifty years ago.
In 1970, George Akerlof showed what happens when buyers cannot tell a good product from a bad one. They stop paying extra for good, and the honest sellers are the ones who lose. Three years later, Michael Spence explained what keeps a market like that working. The side that knows more sends a signal, and the signal means something only when it costs a weak candidate more to send than a strong one. His example was job applicants and education. The same logic applies to companies and their buyers.
Polish used to be that kind of signal. A serious website, a produced video, and a steady stream of content told a buyer that someone had invested real money.
First impressions of polish form fast. In one well-known study, people rated the visual appeal of web pages after seeing them for 50 milliseconds, and those ratings closely tracked the ones they gave after half a second.
That signal now costs almost nothing to send, and the imitation can beat the original. In a 2022 study, people could not reliably tell AI-generated faces from photographs of real people, and they rated the generated faces as 7.7% more trustworthy on average. In 2024 the FTC banned fake reviews and testimonials. The agency's commentary on the rule says AI tools let bad actors generate large numbers of realistic fake reviews quickly and cheaply.
That is where website polish, content volume, produced demos, and written testimonials are heading.
Buyers have noticed
In October 2025, Gartner surveyed 1,539 U.S. consumers. Of those, 68% said they frequently wonder whether the content they see is real, and 61% frequently question whether the information they use for everyday decisions is reliable. Half said they would rather give their business to brands that keep generative AI out of customer-facing content.
The mood is wider than marketing. Trust in news across 48 markets fell to 37% in 2026, the lowest the Reuters Institute has recorded since it began measuring in 2015. It was 42% in 2022 and held at 40% for the three years after. In the United States the figure is 25%.
Edelman found that 70% of people across 28 countries are hesitant or unwilling to trust someone whose values, sources of information, approach to problems, or background differ from their own. Pew found Americans split almost evenly on whether they can tell what is true in the news: 51% say it is difficult and 49% say it is easy.
None of those surveys is about buying software. I include them because the people who answer them are the same people who sit on buying committees, and I do not believe they set that skepticism aside at work.
What buyers do now
Skepticism matters when it changes what buyers do. Three behaviors show up clearly in the data.
They decide before you show up. 6sense surveyed nearly 4,000 B2B buyers in 2025, with a median purchase between $200,000 and $300,000. The winning vendor was on the buyer's shortlist on day one 95% of the time, up from 85% a year earlier. Buyers already had experience with 3.8 of the roughly five vendors they evaluated, up from 3.2. Before speaking to a single seller, 94% had ranked that shortlist, and their first choice went on to win 77% of the time.
Those numbers describe a sale that is mostly decided before the first call. The buyer arrives with a favorite, and the favorite wins about three times out of four. That share was 83% in earlier years, so sellers do change some minds. Most of the conversations that follow confirm a choice the buyer already made.
That does not make sellers irrelevant. In the same research, buyers first contacted a seller 61% of the way through their process, earlier than the 69% found in previous studies. They come to sellers sooner, and they come to check what they have already started to believe. The credibility has to be in place before that call.
They go with who they know. In the same study, 85% of buyers had prior experience with the vendor they chose, and three in four personally knew a seller at the vendors they evaluated. In a companion survey, about 70% said economic uncertainty pushed them toward more conservative choices, meaning known vendors and incumbents.
One field experiment shows what rising distrust does to behavior. The German newspaper Süddeutsche Zeitung showed a random group of its readers how hard it is to tell real images from AI-generated ones. Those readers came away trusting news less, including the newspaper that ran the test. In the days that followed they visited its site more often, and five months later more of them were still subscribers.
Less trust in news overall came with more engagement with a source those readers already rated highly. The authors read the result as trusted sources becoming more valuable as misinformation spreads. I expect the same in buying. When buyers cannot judge the unfamiliar, they keep buying, and they buy from the names they already trust.
They want someone on the hook. Gartner surveyed 645 B2B buyers in August and September 2025 and published two findings that look like a contradiction. In that survey, 67% said they prefer to buy without a sales rep, up from 61% a year earlier, and 69% said they prefer to validate AI-generated insights with a rep.
A third number from the same survey explains it. Asked where they are more likely to run into misleading information, 51% said generative AI and 49% said a sales rep. By that measure the rep and the chatbot are about equally likely to mislead.
Part of the reason is plain expertise. Buyers need product detail, technical answers, and help sorting out competing claims. My read is that accountability is the bigger part. A rep's answer comes with a name and a company attached, and the buyer can hold both to it. A seller's value now sits in standing behind what the buyer was told.
6sense saw the same pull from another direction. In its companion survey, 58% of buyers said they contacted sellers earlier than they otherwise would have, because they needed answers about the AI inside the products they were evaluating.
Four popular answers I do not buy
"Buyers will freeze." I understand why people believe this. Deals do stall. Forrester reports that 86% of B2B purchases stall at some point in the process. The research behind The JOLT Effect covered 2.5 million recorded sales calls and found that 40% to 60% of deals end in no decision.
That research began in March 2020, before generative AI reached buyers. The 6sense data from 2025 shows buying cycles about a month shorter globally and nearly unchanged in North America. What I see is caution. Buyers narrow the field early and choose the safe name. For a challenger that is worse than a freeze, because a frozen buyer can still be won.
"More content will fix it." Buyers in Gartner's 2025 survey already used seven information sources on average for a single purchase. Gartner also surveyed 1,464 B2B buyers and consumers about personalized marketing. It does sell: people who experienced it were 1.8 times more likely to pay a premium. They were also twice as likely to feel overwhelmed by the volume of information and 2.8 times as likely to feel time pressure to move forward. The 53% who had a negative experience with it were 3.2 times more likely to regret the purchase.
Committees make it worse. In Gartner's 2024 survey of B2B buyers, 74% of buying teams showed unhealthy conflict. Content tailored to individuals damaged group consensus, and content built for the whole buying group improved it. My read is that when every stakeholder arrives with a personal feed and a personal AI assistant, the group starts further apart.
More research does not rescue anyone either. In a study published in Nature, people who searched online to check a false news story became more likely to believe it. Their searches could land on low-quality sources that agreed with it.
"Just be authentic." This one is close, but it is too loose to act on. Across seven experiments, people who believed an emotional brand message was written by AI became less loyal and less likely to speak well of the company. The penalty shrank for factual messages, when AI only edited the message, and when people believed most marketing is written by AI.
That gives a workable rule. AI costs you little on a spec sheet or a comparison table. A thank-you, an apology, or a note from the founder should be written by a person.
"Show them proof." That works only when the proof is hard to fake. A case study is a document and a testimonial is a paragraph. Both cost almost nothing to produce now, and I expect buyers to discount them accordingly.
Three ways to be believed
The proof that still works is hard to fake. It reaches a buyer in one of three ways.
Someone vouches for you. A person the buyer already believes puts their own reputation behind yours. Edelman found that among people who trust an influencer, a majority would reconsider a company they distrust if that person vouched for it. In HR tech the influencer is a peer, a broker, a consultant, an investor, or a partner manager at the payroll platform. A buyer who cannot verify the vendor can still verify the person vouching, because they know that person and that person has a reputation to lose.
The buyer can check it. The claim can be tested without taking your word for it: a live test on the buyer's own data, a delivery record anyone can look up, a customer the buyer finds on their own instead of one you hand them.
You pay if you are wrong. A guarantee with money behind it, service credits, pricing tied to the outcome. This is the logic Spence described in 1973. A 2005 experiment on online merchants tested a money-back guarantee, branding, and a privacy statement as signals of a trustworthy seller. The results matched what signaling theory predicts.
| Cheap to fake | Hard to fake |
|---|---|
| Website and brand polish | A peer who vouches for you |
| Testimonial quotes | A customer the buyer finds on their own |
| Written case studies | A delivery record the buyer can check |
| A produced demo | A live test on the buyer's own data |
| Claims of scale | Pricing tied to the outcome |
| Promises in the proposal | A guarantee with money behind it |
Each of the three can be gamed. A reference can be coached. A guarantee buried in exclusions is worth nothing. Outcome pricing invites arguments about how the outcome gets measured. The test is the same for all of them: would it hurt the seller if the claim turned out to be false?
This favors the names buyers already know
Everything above works in favor of companies that are already known.
If the shortlist is set on day one and filled with familiar names, the best product in a category can lose to the third-best product with better relationships. I have watched that happen in HR tech for twenty years, and I expect it to become more common in more markets.
I think that is bad for buyers, and it is worth being honest about why. Trust can protect companies that have not earned it. An incumbent can win with a worse product. A middleman can gain influence because it already has the buyer's ear, whatever its incentives. The deal we lost to the private equity introduction may have gone to the better product. Nobody in that room had a way to know. That is why a recommendation is not enough on its own, and why proof a buyer can check matters as much as the person who vouches.
It is also the opening for challengers.
Most challengers answer a trust problem with more marketing: more content, more outbound, more polish. That is money spent on the signals buyers now discount. The companies that break through will spend it on being known by the right few hundred people, and on proof that would hurt them if it were false.
What I would do
Here is where I would put the effort if the market did not know my company yet. The moves follow the three ways to be believed.
Get vouched for
- Get known before there is a deal. You cannot sell your way onto a list that was written before the buyer called anyone. Be present where your buyers talk to each other: the peer communities, the platform partner teams, and the brokers and advisors they call first. Measure the team on how often you make the shortlist.
- Borrow trust until you have earned your own. That means a customer who will take a call, a partner the buyer already uses, or an advisor with their own name at stake. One person the buyer knows saying you are good can do more than your website will. Pair it with proof the buyer can check, because a recommendation can be wrong.
Make it easy to check
- Sell to the room. Build one story the whole committee can repeat to each other, with numbers the CFO, the IT lead, and the end user can each check. Retire the content built to tell each stakeholder something different.
- Make every claim easy to check. In Gartner's 2024 survey, 69% of buyers said the vendor's website and the vendor's sellers gave them inconsistent information. Fix that first. Then assume everything you publish will be fed to an AI and compared with your competitors, because 94% of buyers in the 6sense study already use these tools during a purchase.
- Be honest about where the machine is. Use AI for the spec sheet, the summary, and the comparison table. Write the hard emails yourself, and sign your name.
Put something on the line
- Put money behind your claims. Guarantee the outcome on terms a buyer can actually enforce, price against it, or run a pilot on the buyer's own data with the pass mark agreed up front. A buyer has little reason to bet on a product its own maker will not bet on.
- Change what your sellers are for. A rep who recites features is repeating what the buyer's AI assistant already said. A rep who can tell a buyer what will go wrong in month three, and what the company will do about it, is doing a job software cannot do.
None of this is free. A guarantee puts real money at risk. A pilot ties up your best people for weeks. Outcome-based pricing pushes revenue out and makes the forecast harder to call. Showing up in the communities your buyers trust takes years to pay back, and it never appears in a quarterly pipeline report. A company with twelve months of runway cannot do all seven at once.
If I had to sequence it, early-stage companies should start with borrowed trust and pilots, because both cost time more than cash. Guarantees and outcome pricing come once you know your numbers well enough to bet on them. Community presence starts on day one, because it is the slowest to pay back.
Where I could be wrong
There are three things I cannot prove yet.
I cannot tell you how much of this is AI and how much is years of falling trust and tight budgets. I have stopped needing to. The buying behavior is the same either way.
Nobody has measured what proof is worth. I believe a challenger with a guarantee and three customers who pick up the phone can get onto a shortlist it was not on. I have not found data that shows it, and that is the research I would most like to see.
AI may also end up fixing part of the problem. Buyers already use it to compare vendors and check claims. If checking gets cheap enough, the best product wins more often. I hope that happens, and I am planning as if it will not.
The question to ask
Most go-to-market playbooks are built to win attention. The harder job now is being believed.
The question I would put to any company, including my own, is what it has that would be expensive to fake.
If the answer is a delivery record, customers who vouch for you, and claims you are willing to bet on, the next few years will be good to you. If the answer is a great website, there is work to do.
