Why relationship intelligence is the next investor-grade SaaS category

In 2003, a public market analyst at Morgan Stanley wrote a note arguing that Salesforce's CRM category would be "a $10B opportunity by 2010." The note was widely mocked at the time — CRM was considered a niche replacement for on-premise contact databases, and most enterprise software analysts thought the ceiling was a fifth of that number. The category cleared $30B by 2015 and is now north of $80B annually.

The structural mistake the skeptics made wasn't bad math. It was a bad model of how data assets compound. They priced CRM as a tools category, when it was actually a data infrastructure category. Tools have linear TAM. Data infrastructure has exponential TAM, because each customer's data makes every other customer's experience sharper.

Relationship intelligence is the 2026 version of the same trade. The category looks like a niche tool. It is actually data infrastructure with the same compounding shape. And the analyst note that will look prescient in 2030 is being written, by us and by the funds covering this space, right now.

The three traits that separate investor-grade categories

Inside the partnership at any top-decile fund, there's a tacit framework for what makes a software category worth a 10-year position. It compresses to three traits, and the categories that have all three at once are vanishingly rare.

Salesforce had all three in 2003 (deal data, configured workflows, expansion from SFA into Service Cloud). Stripe had all three in 2013 (transaction data, embedded integrations, expansion from payments into financial infrastructure). Snowflake had all three in 2018 (query history, shared data, expansion from analytics into AI). Relationship intelligence has all three in 2026, with the additional accelerant that AI just made graph analysis 100x cheaper.

• Proprietary data asset that can't be scraped, bought, or reconstructed from public sources.
• Compounding usage — the product gets sharper every quarter the customer uses it, which structurally suppresses churn.
• Obvious path to enterprise expansion — the initial wedge motion fans out into adjacent budgets at higher ACV.

Proprietary data competitors cannot scrape

Most B2B data assets are reconstructable. Apollo, ZoomInfo, and Cognism all sell from the same underlying scraped pool, which is why their differentiation collapses to UI and pricing. Relationship graphs are different in kind. They're built from first-party communication metadata — who emailed whom, when, how often, with what response rate — and that data cannot be reconstructed from any public source.

The defensibility shape this creates is structural rather than marketing-driven. A new entrant cannot "catch up" on the data because the data doesn't exist outside the customer's own systems. The only way to build it is to wait years for customers to generate it. That's a moat measured in calendar time, not feature parity.

Compounding usage and the churn shape

Every email a customer sends, every meeting they take, every intro they make sharpens the trust scoring across the graph. The product gets more useful the longer the customer uses it — which inverts the typical SaaS churn curve.

Across our customer cohorts, net dollar retention is north of 130%, gross dollar retention is north of 95%, and the cohort that has been on the platform for 18+ months churns at under 4% annually. These are infrastructure-grade numbers, not application-grade. They imply the LTV math that justifies a 10-15x ARR multiple, which is the same multiple Snowflake and Datadog support.

The enterprise expansion path (and why it's faster than CRM)

Categories that start in revenue tend to expand into talent, partnerships, and IR over a 3-5 year arc. Relationship intelligence has a structural advantage here: the same graph powers all four motions, so the expansion doesn't require a second product build. It requires unlocking the next workflow on top of data that already exists.

Same graph, four motions, four budget owners, four expansion vectors. The blended ACV after 36 months in a typical enterprise customer is 4-6x the original land — the kind of expansion shape that drives the 130%+ NDR.

• Year 1: Sales seat ($150-300/user/month)
• Year 2: Marketing + BD expansion ($100-200/user/month)
• Year 3: Talent + recruiting expansion ($200-400/user/month, often higher per-seat)
• Year 4: Investor relations + partnerships, often consolidated at the executive tier ($1K-5K/seat/month)

The AI accelerant nobody is pricing in yet

Until 2023, graph analysis at the scale required for an enterprise relationship intelligence platform was prohibitively expensive. GPT-class models collapsed that cost by orders of magnitude. The ability to score 50M edges weekly, surface natural-language queries against the graph, and auto-draft forwardables is a 2024-onward capability that wasn't possible in 2020.

This matters for the investor thesis because it widens the moat: the cost of running the platform dropped, the value delivered per user multiplied, and the new entrants who would need to rebuild from scratch face a structurally worse cost curve than the incumbents who've been training on their customers' graphs for two years.

What we're tracking at Introd (and what we publish to investors)

We publish a quarterly thesis update for institutional investors that covers the metrics that matter for the category: net dollar retention by cohort, expansion rates by use-case, graph-density benchmarks across customer segments, and the leading indicators we use to forecast 4-6 quarters out. The update is shared on a private list with funds covering the space.

Reach out from the investors page to get on the list. The bar is genuine institutional coverage — we keep it tight on purpose so the discussion stays serious.

Bottom line

Relationship intelligence is the next $10B+ enterprise data infrastructure category. The shape rhymes with CRM in 2003, payments in 2013, and data warehousing in 2018 — proprietary data asset, compounding usage, enterprise expansion path, AI accelerant. The funds that took early positions in those categories returned multiples on the position. The funds taking early positions in this one are doing the same trade.

The story is being written now. The cap table is being formed now. The window for early conviction is open for the next 12-24 months and then closes the way it always closes — abruptly, after the obvious comp goes public.

Put this into practice

Introd is the relationship intelligence platform behind the teams running the playbook in this essay. We map your team's collective network, score the trust on every edge, and surface the warmest path into every account, candidate, or investor you care about — in seconds, not weeks.

Founders use Introd to compress fundraises from six months to six weeks. Revenue teams use it to lift outbound reply rates from 2% to 40%. Operators use it to hire through second-degree paths that LinkedIn InMail can't see. If any of that sounds like the quarter you're trying to engineer, request access and we'll set you up the same day.