The Data Vendor Market Is Quietly Collapsing. Here's What Replaces It.

The numbers don't lie

I've been thinking about this for a while. And I'm clearly not the only one.

ZoomInfo's valuation has dropped roughly 80%. Seamless AI is playing the freemium volume game. Lusha announced they're building outbound features to stay sticky. And every other week, there is a new "cheaper ZoomInfo" or a "cheaper Apollo" entering the market.

These are not small signals. This is an entire category trying to figure out what it becomes next.

ZoomInfo stock price. The market is telling us something.

I've been tracking this since we started Evergrowth. Six years ago, you couldn't run a sales team without a data vendor seat for every rep. Today, the smartest teams are canceling those seats and replacing them with something fundamentally different.

How data became a commodity

What used to be a moat for sales teams is now table stakes. Contact info, employee counts, hiring signals, technographics. Every data vendor sells roughly the same thing. And now there are dozens of them competing on price.

Think about it. When I started in B2B sales 14 years ago, if you had a clever way to scrape or source your company data, you were ahead of your competition. That was a real edge. You could build targeted lists that nobody else had.

That edge is gone. Everybody is working with the same data that you are. The same 600 million contacts. The same job titles. The same company sizes. The same industry filters. Your competitor buys from the same vendor and runs the same search. That's not a moat. That's a commodity.

It's no longer about having access to millions of companies. It's about knowing what 200 people want right now and why.

And here's the thing that makes it worse: most of the time, reps are spending hours filtering those millions of records only to chase low-fit leads. Because filtering on company size, region, and job title doesn't tell you anything about whether a company actually has the problem you solve.

Why more data won't save your pipeline

We don't need more data. We need more judgment about what data matters.

The entire data vendor model was designed for a world where salespeople were the information holders. You needed a database to find who to call. You needed data points to fill variables in your email templates. And then you automated all of that at scale.

That model worked when channels weren't crowded. It worked when email was still exciting and people actually read what landed in their inbox. It worked when you could do three filters and end up with a decent list.

It doesn't work anymore. And adding more data to a broken model just gives you a bigger pile of noise.

Here's what that looks like in practice: your reps open a data vendor, filter by company size, industry, and job title. They get 2,000 contacts. They push them into a sequence with some variable-driven personalization. "Hey {first_name}, I noticed {company_name} is in {industry}." And then they wonder why nobody replies.

The problem isn't the data. The data is accurate. The problem is that data points can't tell you why a company should care about you right now.

Context is the new moat

All the software we used to buy for go-to-market was designed to record data. CRMs, data vendors, enrichment tools. They're systems of records. They store fields. They're really good at answering "what is the employee count of this company" or "what is this person's job title."

But AI doesn't work like that. AI needs context to be useful. And context is fundamentally different from data.

Context is time-sensitive. It's built on dynamic signals, not static fields. It aligns with your ICP criteria and your playbook. It gives a rep something real to talk about. Something relevant to your value proposition. Something that shows the prospect you understand their world.

No vendor can package this for you. No database can store it. You need the infrastructure to build it, measure it, and deploy it to your team continuously.

What replaces the data vendor stack

My take is that data vendors won't disappear entirely. But they'll stop selling seats. They'll become utility layers or API-first infrastructure that AI agents rely on. The shift is from buying seats so reps can filter databases to accessing APIs so agents can collect context.

Which means the whole economics change. Instead of paying $15K per year for 5 seats that your reps use to run the same filters your competitors run, you're paying usage-based for the specific data points that your agents need to complete their research. You only pay when enrichment succeeds. You only look for contacts after agents have already qualified the account.

In the long term, we'll replace data tools with agentic workflows that do the full job:

1. Qualification before everything

Before you enrich a single contact, agents verify that the company actually has the problem you solve. They browse the website, read the careers page, check if there's a dedicated sales team, look at the product, and make a judgment call. If the company doesn't qualify, you never spend a credit on it.

2. Research that produces signals, not fields

Instead of returning "Series C, 500 employees," agents read job ads (the actual descriptions, not just the titles), annual reports, conference appearances, social media activity. They come back with context: "Hiring 3 SDRs in London, CRO is 60 days in, evaluating alternatives to Outreach." That's a signal you can act on.

3. Contact discovery based on personas, not job titles

Instead of filtering 600 million contacts by "VP Sales," agents search for people that match your buyer persona. They understand that a CPO with a supply chain background is procurement, and a CPO with a product background is not. They find people who aren't on LinkedIn. They check company staff pages, university directories, company registries. And then they run waterfall enrichment across 20+ APIs to find email and phone.

4. Personalized plays, not variable-filled templates

The research feeds directly into play copywriting agents that produce personalized talk tracks and outreach for every single contact. Not "Hey {first_name}, I noticed {company_name} is in {industry}." Instead: a cold call opener that references a specific initiative, ties it to a specific pain, and gives the rep confidence to have a two-minute conversation that actually goes somewhere.

5. Prioritization based on relevance, not recency

Agents score accounts based on how much of their research came back positive. Not based on a formula that weighs employee count and funding round. Based on whether the company actually matches your criteria, has active signals, and has reachable contacts that fit your personas.

The race is on

The future of pipeline isn't scraping faster. It's qualifying more intelligently and working the accounts that actually matter.

I know a lot of companies have been building automations around this idea of "buying signals" for at least two decades without real success. The difference now is that agents can do the research at scale. They can open websites, read PDFs, watch for changes, and make conclusions. They remember 100% of their training. They don't get tired. And they cost a fraction of what a data vendor seat costs.

The teams that are building context-driven systems today are going to crush the teams that are still buying seats to filter databases tomorrow. That's not a prediction. That's already happening.

The race to the building side is on. Don't get crushed by teams who are building systems that think for them.

And that's why Evergrowth exists.