A friend of mine, Fraser Morrison, published an article this week that crystallised something we have been wrestling with at TailyX for months. One line in particular stopped me:
Buyers are 90 to 94 percent through their decision before engaging a seller, and the first company to make contact wins the business 83 to 94 percent of the time.
Read that twice. By the time a buyer hits your inbox, they have already done the work. They have read the content, watched the demos, asked their network, and shortlisted three vendors. The conversation that used to start with discovery now starts somewhere near the finish line. And whoever responds first, with substance, usually wins.
That single statistic is why we shut down our conversational AI product earlier this year and rebuilt the company around two things: lead scoring and enrichment. This article is about why.
When we started TailyX, the obvious play was conversational AI on the marketing side. A chatbot on the website. An AI SDR sending personalised cold outreach. An assistant qualifying tyre-kickers before they reached a human. It was the consensus product. Every funded competitor was building some version of it.
The pitch made sense on paper. Buyers want fast answers. Sales teams are expensive. Put an AI in the middle and the maths works.
Except the maths only works if the buyer actually wants to talk to your AI. And increasingly, they do not.
Fraser's number — 90 to 94 percent through the journey before engaging — is not an accident. It is the result of a structural shift that has been underway for years and accelerated sharply with generative AI. Buyers no longer need a vendor to explain the category to them. They have ChatGPT. They have peer reviews on G2. They have LinkedIn networks where they can ask anyone in their industry, in any geography, for an honest take in under an hour.
The old funnel assumed the vendor controlled the information. The buyer came to you ignorant and you educated them into a sale. That funnel is dead. The buyer now arrives informed, opinionated, and often more current on your competitors than your own SDRs are.
So ask the obvious question. Why would a buyer who has spent three weeks researching your category, vouching with their peers, and reading independent comparisons want to be slowed down by your chatbot asking them to confirm their company size?
They would not. And the data shows they do not. Conversational AI on the marketing side is, increasingly, friction dressed up as helpfulness.
Here is the part that was uncomfortable to admit. The problem conversational AI was built to solve — qualifying and educating buyers at the top of the funnel — has been quietly automated by the buyers themselves. They self-educate. They self-qualify. They arrive at your form already knowing whether you are a fit.
What has not been solved is what happens after they arrive.
Most B2B inbound funnels still treat every form fill the same way. The lead lands in a queue. An SDR works through it in roughly the order it arrived. By the time the high-intent buyer who already shortlisted you hears back, your competitor has already booked the meeting. Fraser's number again: 83 to 94 percent of the time, the first company to make contact wins.
That is not a marketing problem. It is a triage problem. And triage is not what conversational AI is good at.
Speed alone is not the whole answer. Anyone can fire off a fast email. The real question is whether the seller who responds first also shows up prepared. A response in five minutes that opens with "thanks for your interest, when's a good time to chat?" is barely better than one that takes two days. The buyer has already done their homework. If your seller has not done theirs, the conversation is dead before it starts.
So we built TailyX around two jobs, not one.
Scoring answers the first question: which of the leads in your inbox right now are ready to buy, and which can wait? It is the triage layer. Get it right and your sellers spend their time on the buyers who are actually decided.
Enrichment answers the second question: when the seller picks up the phone, what do they already know? It is the preparation layer. Company context, recent activity, role and seniority, the signals the buyer left behind on the way in, the language they used in their inbound message. By the time the seller dials, the brief is already written.
The two have to work together. Scoring without enrichment gets your best seller to the right buyer fast — and then watches them fumble the call because they walked in cold. Enrichment without scoring means your best seller arrives prepared to a meeting with a buyer who was never going to convert. Both halves matter. Neither works alone.
The shift in product thinking looks like this:
The buyer never sees our product. The seller does. That is the whole point.
There is a second reason we walked away from conversational AI, and it is more philosophical. Buyers do not trust vendor-controlled marketing the way they used to. Why would they? Every vendor's chatbot says the same things. Every vendor's AI SDR sends the same personalised email. The signal-to-noise ratio has collapsed.
What buyers trust is their own research and their own peers. Fraser makes the same point in his article when he describes the shift toward what he calls B2P2P2B — business to person to person to business. Every deal, eventually, comes down to a human relationship. The vendor's job is not to inject AI into that relationship. It is to make sure the right human shows up at the right moment, prepared.
Scoring and enrichment are the unglamorous, behind-the-scenes work that makes that moment possible. The buyer never interacts with either. What they experience is a seller who got back to them quickly and clearly knew who they were. That is the only signal that matters.
If Fraser is right that the future commercial model is converging — marketing and sales operating as one team, the buying journey viewed as lead-to-renewal, and human sellers augmented by AI agents — then the practical question for any sales leader is: where do I put the AI?
Our answer, after a year of getting it wrong and six months of getting it right, is this. Put the AI where it makes your humans faster and sharper, not where it slows the buyer down. The leverage is not in talking to more buyers. It is in making sure your best sellers are in front of the right buyers, within minutes of intent, with everything they need to win the conversation already in front of them.
Fraser left a comment on the original draft of this thinking that summed it up better than I could:
A great example of one of the use cases that makes a salesperson more effective and better prepared.
That is the bet. The buyer has already done 90 percent of the work. Your seller's job is the last 10 percent. Scoring tells them who to call. Enrichment tells them what to say. Together, that is what wins the deal that was already half-won before the phone rang.