MARS strategy note · 2026-07-23 · For Elise
The question asked; the honest answer. Public pensions fragment into three buyer segments with very different economics. One of them is worth pursuing today with what we have. One will be worth pursuing when we ship lp_data/. One is somebody else's business.
Public pensions are a real buyer segment — but "public pensions" as one bucket is misleading. The three tiers behave like three different customer types.
The top ~50 pensions ($10B+, CalPERS class) are worth pursuing today — they buy multiple data subscriptions and want complementary intelligence we can uniquely deliver (real-time Form D, GP person mobility, MCP surface). The middle ~400 pensions ($500M–$10B, Mendocino class) are Buyouts Insider's moat and hard to displace head-on. The bottom ~4,000 aren't worth chasing.
A note on the specific example: D1 most likely means D1 Capital, which is a hedge fund, not a pension. Worth clarifying with Elise. This note assumes she meant tier-1 large pensions; if she meant D1 Capital as the buyer type, that's a separate conversation and the answer is also yes.
Grouping "public pensions" as one segment is the trap. Real pensions look like three buyer types with very different sales motions.
| Tier | Count | AUM range | Buys today | ARPU range | MARS fit |
|---|---|---|---|---|---|
| Top 50 — sovereign class | ~50 | $10B – $500B | Preqin, PitchBook, Bloomberg, Burgiss, in-house research | $50K – $150K/yr | Pursue now |
| Middle — county / state class | ~300–500 | $500M – $10B | Buyouts Insider, PitchBook light, consultant reports | $5K – $15K/yr | Wait for LP data |
| Small — city / municipal | ~4,000 | < $500M | Consultant reports only; no dedicated data budget | $0 – $5K/yr | Pass |
ARPU ranges reflect what these tiers actually pay for data today, not what we'd charge. The Mendocino pension in the Buyouts Insider article sits in the middle tier ($1B AUM).
The pitch is complementary, not substitute. Top-50 pensions already buy Preqin. We win where they can't.
Top-tier pensions already have LP-side data. They need additional intelligence on the GP-side that Preqin doesn't do well:
| MARS capability | What it gives a top-tier pension analyst | Substrate today |
|---|---|---|
| Real-time Form D signal | New GP fundraise 4–72 hours after filing, weeks before Preqin publishes | 354K filings, hourly ETL |
| GP person mobility | "The partner who left KKR in 2023 for Ares is raising Fund V" — real career-graph, not just headshots | 30K persons, 164K work-history rows |
| ADV firm profiles | Every SEC-registered manager with AUM, CRD, disciplinary flags, redflag counts | 100K investors, 49K RIA firms |
| Canonical deal graph | Cross-vehicle roll-up: every Blackstone fund's deals across ~10 registered entities | 91K canonical deals |
| MCP + AI-native surface | Answers to natural-language questions instead of dashboards; workflow lives in Claude / ChatGPT | Live, sub-50ms p95 |
The middle tier belongs to Buyouts Insider because of one specific dataset we don't have yet.
Every Buyouts Insider article opens with a datapoint we can't source today: "Mendocino County committed $50M to Blackstone Real Estate Fund X." That LP↔fund mapping is not in Form D, ADV, or any SEC substrate we ingest.
Buyouts Insider gets it by scraping public pension CAFRs (annual reports) and consultant board packets. It's public data but nobody's normalized it.
Our lp_data/ vertical was scaffolded specifically for this. Not yet built. Estimated 30–40% coverage of the Preqin universe at ~$0 marginal cost once running.
Fund performance metrics. Genuinely private for most vintages; leaks through pension CAFRs (which do disclose returns per commitment) and quarterly manager reports.
Same source path as commitments — public pension disclosures — so the two problems solve together via lp_data/.
Coverage would be partial (~40% of the fund universe at best) because private funds with only private LPs never surface publicly.
Direct competitive read on the Mendocino article we saw today.
Buyouts Insider (PEI Media) sells four database products — LP Search, GP Search, Fund Search, Commitments — plus editorial coverage. Their moat is the LP↔commitment mapping, built from years of CAFR scraping and consultant relationships. Their weakness is on the GP side: they cover the fundraising narrative but don't have SEC-native substrate, don't have person mobility, don't ship an MCP surface.
| Capability | Buyouts Insider | MARS today | MARS after lp_data/ |
|---|---|---|---|
| LP↔fund commitments | Strong | — | Partial (30–40%) |
| Fund IRR/TVPI/DPI | Partial | — | Partial (~40%) |
| GP fundraise signal | Weekly | Hourly | Hourly |
| Entity graph (person + firm) | — | 30K persons | 30K persons |
| Real-time SEC substrate | — | 354K Form Ds | 354K Form Ds |
| MCP / AI-native workflow | — | Live | Live |
Elise's second question — worth answering here because the two segments overlap in what they need.
Callan is an investment consultant. They advise pensions on manager selection and portfolio construction. They pay for data at ~10× the middle-tier pension rate — think $200K – $1M per year for institutional Preqin / PitchBook / Burgiss subscriptions — because their business model rests on informed manager recommendations.
We already have Callan in our substrate: three SEC-registered entities in investors_v2. Callan LLC ($41.9B AUM), Callan Family Office ($10.2B), Callan Capital ($2.3B). All CRD-anchored.
The advisor tier is ~30 firms total (Callan, Wilshire, Mercer, Meketa, Aksia, Cambridge Associates, NEPC, RVK, Verus, etc.). Higher ARPU per relationship, fewer relationships to build, more sophisticated buyers who evaluate data quality rigorously. The MARS pitch to Callan looks nearly identical to the top-tier pension pitch — same substrate, same real-time signal, same MCP surface — but at multiple times the price point.
If we want to turn this into action.
| Segment | Action | Rationale |
|---|---|---|
| Top-50 pensions (CalPERS class) | Pursue now — pitch as complementary to their Preqin sub | Real-time Form D + person mobility + MCP is a genuinely differentiated product. Preqin doesn't cover it. |
| Middle 300–500 pensions | Wait for lp_data/ | Buyouts Insider's moat. Attacking head-on before we have LP commitments means selling half a pitch. |
| Bottom 4,000 pensions | Pass | No dedicated data budget. Consultant-driven decisions. |
| Advisors (Callan tier) | Pursue now — this is likely the highest-value slice | Same product as top-tier pensions, 10× the ARPU per firm, fewer firms to reach. And Callan is already in our substrate as an anchor demo. |
Pull 20 recent Buyouts Insider LP News stories. For each, measure what MARS surfaces on the same event today. Concrete coverage percentage; concrete missing-data list. Basis for the pitch to top-tier pensions and advisors.
Estimate: ~2h.
One-page memo: what a MARS pitch to Callan looks like. Real-time Form D demo + person mobility demo + MCP demo, using our existing Callan substrate as the live example. Delivery format: staticpipe-hosted HTML brief plus a live MCP session.
Estimate: ~1h. Reusable for Wilshire, Mercer, Meketa, etc.
Public pension CAFR ingest — free public data. First-cut coverage of LP commitments (~30–40% of Preqin's universe). Ships the missing piece of the middle-tier pension pitch.
Estimate: 4–6 weeks scaffold to first data. Currently unowned; needs a call on priority.
Two interpretations: D1 Capital (Dan Sundheim's hedge fund, ~$30B AUM) or top-tier pension ("D1" as in top division). Very different sales pitches. Worth a two-sentence check-in before we design either one.
Estimate: one message.