Commercial Due Diligence & Primary Research for Private Equity: A 2026 Practitioner's Guide

Why commercial due diligence matters more than ever in 2026

Private equity entered 2026 with a problem that won't go away: too much capital chasing too few defensible deals. Global dry powder sits at roughly $2.59 trillion, with the U.S. alone holding a record $1.1 trillion of unallocated capital. Deal value rose 57% year-over-year in 2025, and the median purchase multiple has crept to 11.8x EBITDA, up from 11.3x a year earlier.

At those multiples, getting the commercial thesis wrong is catastrophic — and you'll be living with the mistake for a while. Average buyout hold periods now exceed 6.4 years, with over $1 trillion of NAV trapped in older vintages. Multiple arbitrage is dead. Cheap leverage is dead. Growth is now the only meaningful value-creation lever, and growth lives or dies on the strength of the commercial thesis you underwrote at signing.

Meanwhile, the work is harder. Deal teams are running confirmatory diligence in 4 to 8 weeks from NDA to binding offer. They evaluate roughly 80 opportunities for every one investment. Buyout volume rose 10% in 2025 but average deal size jumped to $849 million — fewer, larger, higher-stakes transactions where the IC memo has to hold up to real scrutiny.

This guide is for PE deal teams at mid-market and upper-mid-market firms who need to do CDD better, faster, and with more defensibility than they did last cycle. It's opinionated. It assumes you already know what a CIM is and what an LBO model does. The focus is on the part of CDD that actually moves the needle: primary research.

Where primary research fits in the PE deal funnel

Most firms still treat primary research as a confirmatory-stage exercise. The best firms have stopped doing that. Here's how to think about it across the funnel:

• Sourcing & screening: 2–3 targeted expert calls to validate or kill the sector thesis before you spend partner time. Cheap insurance against pursuing the wrong deal.
• Pre-IOI: A handful of customer and ex-operator conversations to stress-test management's growth narrative before you anchor on a price.
• Pre-LOI / IC prep: The bulk of primary research — 10–20+ expert interviews, off-list customer references, and ideally a quantitative survey if the customer base supports it.
• Confirmatory / post-LOI: Fill remaining gaps, pressure-test specific risks flagged earlier, lock in the IC narrative with named quotes.
• Post-close / 100-day plan: Convert findings into the operating roadmap — customer pain points become product priorities, adjacency signals become commercial bets.

Calls executed too late are the single most common failure mode we see. By the time you're in confirmatory, walking away costs you broken-deal fees, reputation with the banker, and a sunk-cost bias that's almost impossible to overcome at IC.

What CDD actually has to answer

Strip away the consulting jargon and a good commercial due diligence answers seven questions, each with a binary, defensible position:

1. Is the market real, growing, and addressable? Not "is the TAM big" — is the SAM growing at the rate the seller claims, and why?
2. Is the target's position defensible? What are the actual switching costs, and would customers leave for a 10% price cut?
3. Is the customer base healthy? Concentration, NPS, gross retention, net retention — validated by customers, not the data room.
4. Is pricing power real? Can the company pass through 5–10% price in the next cycle without volume loss?
5. Who actually competes, and how? Not a logo slide — a view on how competitors go to market, where they're winning, and what they say about the target.
6. What does the growth plan require, and is it credible? New products, new geos, new segments — pressure-tested against operator reality.
7. What's the AI / disruption exposure? Is this a "revolution" target (rare), a "transformation" target (most), or an "augmentation" target?

If your CDD output doesn't take a clear position on each, it's theatre — and a lot of CDD is theatre. It validates rather than challenges, because the team running it was hired to validate.

The five primary research methods that matter

1. Expert interviews

30–60 minute conversations with former executives of the target or competitors, industry specialists, channel partners, and ex-operators. The default workhorse of CDD. For a confirmatory workstream on a $500M–$2B deal, plan for 10–20+ calls across customer, competitor, and supplier perspectives.

2. Customer reference calls — on-list

The names management gave you. Useful, but assume they've been cherry-picked. You'll get a relentlessly positive view of the target. Run them, but never weight them heavily on their own.

3. Customer reference calls — off-list

Customers you sourced independently. Off-list references are, in our experience, 10x more valuable than on-list. They're the only way to hear about churn risk, pricing complaints, and product gaps from people who weren't pre-briefed. If you do nothing else differently on your next deal, do this.

4. B2B surveys

Quantitative validation across a larger N — typically 50–300+ respondents — to pressure-test claims that can't be confirmed with 12 calls. Best for: NPS, vendor selection criteria, share-of-wallet, churn drivers, willingness to pay. Critical for B2B software, healthcare services, and any consumer-adjacent target.

5. Channel checks and "secret shopper" calls

Calls to distributors, resellers, and suppliers to triangulate volumes, win rates, and pricing. Secret-shopper calls to competitors — posing as a prospective buyer — to see how they actually pitch against the target. Underused, and often the cheapest source of genuinely new information.

Choosing your research model

There are essentially four ways to get primary research done. Each has a real cost — usually not the one on the invoice.

Traditional expert networks (GLG, AlphaSights, Guidepoint, Third Bridge)

You request expert profiles, the network matches you, you schedule and run the calls yourself. Expert rates average $950/hour, with premium experts at $1,150–$1,350. The cash cost is only half the picture. The real cost is your associate's time — writing discussion guides, scheduling, conducting 15 calls, and synthesising transcripts. For a typical CDD that's 80–120 hours of deal-team capacity you don't have.

Custom-sourced expert networks (Apex, Infoquest, Mosaic, Nexus)

Experts recruited per project rather than pulled from a pre-built database. Materially better fit for niche sub-sectors, emerging markets, and any deal where the database networks return generic "former VP of sales" profiles. Pricing typically 20–30% below the big four. Still DIY on the call execution and synthesis.

Full-service CDD consultancies (Bain, L.E.K., Kearney, A&M)

Soup-to-nuts engagement with a branded report at the end. Strong for IC optics. Weak on speed and cost — typical engagements run 6–10 weeks and several hundred thousand dollars, and the report is rarely sharp enough to actually change a deal decision. Best reserved for platform deals where the brand matters or for sectors where you have no internal expertise.

Done-for-you primary research providers (our category)

You brief the research need; the provider designs the discussion guide, sources experts (including off-list customers), runs the calls, fields the survey, and delivers a synthesised output. The deal team gets back the 80–120 hours and reads transcripts and findings instead of writing guides and chasing schedules. The honest tradeoff: you give up some control over the day-to-day execution. The honest benefit: lean deal teams stop being the bottleneck.

AI-enabled research platforms

Useful adjuncts — Brightwave, Brownloop, Dynamiq, transcript libraries like Third Bridge and Tegus. Good for CIM parsing, transcript synthesis, and pulling existing context fast. Not a replacement for deal-specific primary research, and dangerous when junior analysts treat AI summaries as primary evidence (more on this below).

Designing a discussion guide that pressure-tests the thesis

The single biggest difference between useful and useless expert calls is the discussion guide. Generic guides produce generic answers. Hyper-specific guides produce insights.

Bad question: "Tell me about the EV charging market."

Good question: "How do commercial fleet operators in Texas justify Level 3 charger ROI under current utility tariffs, and what's the payback period they actually accept?"

A guide that actually works has four parts:

1. Thesis pillars at the top. Three to five investment thesis statements the call has to test. Every question should ladder up to one of these.
2. Open framing questions. Let the expert tell you the world as they see it before you anchor them.
3. Specific, falsifiable questions. Numbers, timeframes, named competitors, named customers. "What share of your renewal conversations include a price increase ask?" — not "How is pricing trending?"
4. Disconfirming questions. Explicitly ask what would make the bull case wrong. The best research kills bad deals; it doesn't validate good ones.

Sourcing the right experts and customers

Two things matter here, and both are underweighted in most processes:

Custom-sourcing beats database sourcing for niche deals. If your target sells specialty industrial coatings into Tier-2 European auto OEMs, you don't want the 14 "former auto industry executives" the database surfaces. You want the three people who actually ran procurement at those specific OEMs in the last five years. That's a sourcing job, not a search job.

Off-list customer references are non-negotiable. Management's references will tell you the target is wonderful. They were selected because they'll tell you that. The customers worth talking to are the ones management didn't put forward — including ex-customers. Sourcing them takes work (LinkedIn, industry associations, channel partners, our own networks), but it's where churn risk, pricing fragility, and product gaps actually surface.

Synthesis: what good IC output looks like

• A binary position on each thesis pillar — yes / no / yes-with-conditions — with the underlying evidence linked.
• Direct quotes (named or anonymised) from customers, ex-employees, and competitors. Quotes are what the IC remembers.
• Quantitative survey results that either reinforce or contradict management's forecast.
• A competitive view that goes beyond a logo slide — how each competitor wins, where they're investing, and what they say about the target when asked directly.
• Explicit disconfirming evidence. If you didn't find any, you didn't look hard enough.

Bridging diligence to value creation

The best deal teams treat primary research as an asset that pays out twice: once at IC, and again at the 100-day plan. The customer who said "they keep promising an API integration that never ships" is your first product priority. The ex-VP of sales who said "they've never tried selling into adjacent vertical X" is your first commercial expansion bet. The competitor who said "we're winning on implementation speed" is your operational target.

If your CDD findings don't directly inform the value creation plan, you've done the work twice and learned half as much.

AI in the workflow: where it helps and where it hurts

Roughly 95% of PE and VC firms now use AI somewhere in the investment process, and nearly two-thirds use it in due diligence. The realistic picture:

Where AI earns its keep: CIM parsing, contract extraction, financial document tagging, transcript summarisation, cross-deal pattern recognition, initial market sizing drafts. Hours saved, not days.

Where AI is dangerous: Generic, open-web models produce confident outputs you cannot trace to a source. Hallucinated conclusions in an IC memo are a career risk. Lack of provenance is an audit and compliance risk. Junior analysts using ChatGPT to "summarise the competitive landscape" is the single most common failure mode we see now — and the resulting slide always looks plausible enough that it doesn't get challenged.

The rule: AI for document work and synthesis; humans for thesis design, expert selection, and IC narrative. Every AI-generated claim in an IC memo should be traceable to a primary source.

Compliance and MNPI risk

Expert calls carry real compliance exposure if run carelessly. The 2009 FrontPoint case — where formal expert-network calls drifted into direct contact and then into MNPI — is still the cautionary tale. Hard rules:

• All expert engagement runs through a documented process with screening for current employment, equity holdings, and confidentiality obligations.
• No direct contact with experts outside the formal channel.
• Audit trail for every call — written guide, recorded or transcribed call, retention policy.
• Same applies to AI tools: clear data provenance and permissioning, especially for anything that touches deal-specific materials.

A 6-week confirmatory CDD checklist

Week 1: Lock the investment thesis in writing. Translate it into a discussion guide with 4–5 thesis pillars. Identify required expert types (customer, competitor, ex-employee, channel, specialist). Begin sourcing — both database and custom.

Weeks 2–3: Run the first wave of expert calls (5–8). Iterate the guide based on what you learn. Launch the B2B survey if applicable. Begin off-list customer sourcing.

Week 4: Second wave of calls (8–12), now including off-list customers and ex-operators. Channel checks. Secret-shopper calls to top 2–3 competitors.

Week 5: Survey closes. Synthesis begins. Identify remaining gaps and run targeted follow-up calls. Draft IC findings around the thesis pillars.

Week 6: IC memo finalised with named quotes, survey charts, and a clear position per pillar. Findings handed to the value-creation team for 100-day planning.

That cadence is achievable — but only if the deal team isn't simultaneously trying to schedule 20 calls and write 20 transcript summaries themselves.

The honest takeaway

Most CDD in the market is too late, too generic, and too reliant on management's references. At 11.8x EBITDA multiples and 6+ year holds, that's not a margin for error any IC can afford. The firms winning right now are the ones running primary research earlier in the funnel, sourcing customers off-list, designing thesis-specific discussion guides, and treating diligence findings as the first draft of the value creation plan.

Whether you run that work in-house, through an expert network, through a consultancy, or through a done-for-you research partner is less important than running it well. What's no longer optional is running it at all.

If you'd like to talk through how we run primary research for PE deal teams on live processes — including off-list customer sourcing and B2B surveys on compressed timelines — get in touch with our team.