Building origination
Deal origination function: how to build one.
Most firms do not have a deal origination function. They have a deal team that also does origination when it has time, which means origination happens in bursts, stops the moment a live deal needs attention, and never compounds. A deal origination function is a standing capability with its own engine, not a task the deal team squeezes in between closings, and the difference between the two is the difference between proprietary flow that is reliable and proprietary flow that is lucky.
I run origination at DealSource Systems, and the same question comes up on almost every first call: should we build this in-house, outsource it, or do something in between. This guide is the honest version of that decision. It covers why ad hoc origination quietly fails, the three operating models and how they compare, what the function actually needs to work, and the test to apply before you commit to any of them. It is written for private equity firms, but the same mechanics apply to M&A advisors and corporate acquirers running buy-side mandates.
Why does ad hoc origination quietly fail?
Ad hoc origination fails because it cannot do the three things a real function exists to do: cover a market, run continuously, and compound. The default setup looks reasonable. Partners work their networks, an associate runs some outreach, a data subscription sits in the background. It produces deals, so it feels like it works. The problem is what it cannot do.
It cannot cover a market. A few people working relationships will always miss most of the owners who fit your thesis, simply because those owners were never in anyone's network. It cannot run continuously, because the moment a deal goes live the same people stop sourcing. And it cannot compound, because nothing is being built. Every quarter starts close to zero.
Origination done in bursts will always miss the one window it exists to catch. Owners decide to sell on their timeline, not on your pipeline review.
This matters more now than it used to. Private equity is sitting on more than $1 trillion of buyout dry powder, according to S&P Global, and that capital has to find a home. When everyone has money, the edge is no longer the cheque, it is who reaches the right owner first. A real function fixes coverage, continuity, and compounding. The question is how to staff and run one.
What are the three deal origination models?
There are three ways to build a deal origination function: in-house, fully outsourced, or hybrid infrastructure. They trade off control, cost, and time-to-coverage very differently, and the right choice depends on your scale and your timeline more than on any principle.
In-house team
You hire a sourcing lead, maybe an analyst or two, buy the data tools, and build the capability internally. This gives you full control and keeps the relationships inside the firm.
It is also expensive and slow to stand up. A genuinely good origination operator is hard to find and costly to keep, the data and tooling stack runs into real money before it produces anything, and you carry the management overhead. For a larger fund with the budget and the patience to build over a year, this can be the right answer. For most firms it is more cost and time than the result justifies.
Fully outsourced
You hand origination to an agency that sends you opportunities. Fast to start and light on management, but with two common failure modes. Either the provider optimises for volume and floods you with companies that do not fit the thesis, or it treats your mandate as one of many and the work is generic. You end up with a full calendar that converts to nothing, which is worse than no calendar at all.
Outsourcing works only when the provider runs a real system, scores against your specific thesis, and puts experienced people on the founder-facing work. Most do not.
Hybrid: origination as infrastructure
The model that works for most firms is a hybrid. The heavy, scalable machinery, the data layer, the thesis scoring, the signal detection, the deliverability, runs as infrastructure you do not have to build or staff. Your team stays in control of the thesis and owns the relationships once a founder is in conversation. You get the coverage and continuity of a built-out function without the cost and lead time of building it from scratch. The firm rents the engine and keeps the relationships.
Which deal origination model is right for you?
The right model depends on scale and timeline, and the table below makes the trade-offs explicit. Most firms that need coverage this year, not next, land on the hybrid model.
| Model | Control | Time to coverage | Cost profile | Best for |
|---|---|---|---|---|
| In-house team | Full | 9-12 months to build | High fixed cost; salaries plus tooling | Large funds with budget and patience |
| Fully outsourced | Low | Fast to start | Variable; cheap until it underperforms | Firms testing a single narrow thesis |
| Hybrid infrastructure | Thesis and relationships in-house | Weeks, not quarters | Predictable; engine rented, team retained | Most PE firms and M&A advisors |
The honest test for any of the three is the same. Can it cover the market, run every week without stopping, and improve over time? In-house can, if you fund and manage it long enough. Outsourcing rarely does. Hybrid infrastructure is built for exactly those three properties, which is why it is where most firms end up.
What does a deal origination function actually need?
A deal origination function needs six things, and missing any one is why most setups underperform regardless of which model runs them. Whichever route you choose, hold the build against this list.
- 1. A thesis precise enough to score against. Not a sentence in someone's inbox, a set of criteria a system can rate every company on.
- 2. A market map broad enough to be honest. Built from many data sources plus custom research, not one stale subscription that everyone else also rents.
- 3. Continuous signal detection. So you reach owners around the moments that signal readiness, succession, an ownership change, a senior hire, rather than at random.
- 4. Founder-grade outreach, checked by people. Volume the machine handles, judgement the operators handle, with a human reading every message before a founder does.
- 5. Deliverability infrastructure. The invisible layer that decides whether your messages arrive at all, and the one most firms forget exists.
- 6. Weekly measurement. Conversations, fit, and progression tracked, so the function improves instead of drifting back to zero.
AI does the work that scales. Operators do the work that matters. A good deal origination function is built around that line, whoever runs it.
The reason this list matters is the supply coming to market. McKinsey estimates that around 6 million US businesses, worth up to $5 trillion in enterprise value, will change ownership by 2035 as a generation of founders retires. Roughly half of small-business owners are already 55 or older and most have no succession plan, reports CNBC. Most of that supply is off-market and reachable only through a function that maps, monitors, and reaches owners directly. Add-on acquisitions already make up roughly three-quarters of all buyout deals, according to Cherry Bekaert, and every add-on starts with sourcing a target that fits a platform you already own.
How do you know the function is working?
You know a deal origination function is working when it produces thesis-fit conversations every week, not when it produces activity. Activity is easy to fake and easy to feel good about. Conversations with owners who actually fit are the only leading indicator worth watching.
- Thesis-fit conversations per week. Real dialogues with owners who match the criteria, the closest signal of future deals.
- Coverage of the mapped market. What share of the fitting universe you have actually reached, not just touched.
- Reply and progression rates. Whether outreach lands and moves toward a process, not whether it was merely sent.
- Time to first conversation. How quickly a new thesis turns into live dialogue.
For a sense of what good looks like, one healthcare investment bank we run origination for reached 14 owner conversations in three weeks and 133 within 90 days. That is what a function looks like when coverage, continuity, and compounding are all in place at once, and it is the bar to hold any model against.
Conclusion: stop running origination in bursts
A deal origination function is the part of the deal lifecycle that compounds the most and gets engineered the least. You do not need to choose the perfect model on day one. You need to stop running origination in bursts. Write the thesis down as scoreable criteria, get an honest map of who actually fits, put continuous signal detection behind it, and hold a human standard on everything a founder sees. Run it every week and let it compound. Whether you build it, rent it, or split the difference, the firms that treat origination as a standing function are the ones reaching owners before the auction ever opens. The work is learnable, and it is the highest-leverage thing most firms are not doing on purpose.
Key Terms Glossary
Frequently asked questions
What is a deal origination function?
A deal origination function is a standing capability that finds, prioritises, and reaches acquisition or investment targets continuously, with its own engine and measurement. It is distinct from a deal team that does origination occasionally, because it is designed to cover a market, run every week, and compound rather than reset each quarter.
Should we build a deal origination function in-house or outsource it?
It depends on scale and timeline. Large funds with budget and patience can build in-house over roughly a year. Most firms get coverage faster and cheaper by running the machinery as infrastructure and keeping the relationships in-house. Pure outsourcing only works if the provider runs a real system with experienced operators on the founder-facing work, which is the exact test to apply before signing anything.
What does a deal origination function need to work?
Six things: a thesis precise enough to score against, a market map broad enough to be honest, continuous signal detection, founder-grade outreach checked by people, deliverability infrastructure, and weekly measurement. Missing any one is the usual reason a setup underperforms, whatever model runs it.
How long does it take to build a deal origination function?
An in-house team typically takes nine to twelve months to hire, tool up, and produce its first reliable flow. A hybrid infrastructure model produces founder conversations in weeks, because the heavy machinery already exists and only the thesis and market map are bespoke. One bank we run origination for reached 14 conversations in three weeks and 133 within 90 days.
What is the difference between deal sourcing and a deal origination function?
Sourcing is the activity of finding and reaching targets. A deal origination function is the standing structure that does that activity continuously and measurably. We unpack the wider distinction in deal sourcing versus deal origination and in the complete guide to deal sourcing for private equity.
How is AI used in a deal origination function?
AI handles the work that scales: building the market map, scoring every company against the thesis, and watching for readiness signals. Experienced operators handle the work that matters, which is checking every founder-facing message. There is a fuller account in the honest version of AI in deal sourcing and at how it works.
How do we measure whether the function is working?
Track thesis-fit conversations per week, coverage of the mapped market, reply and progression rates, and time to first conversation. Measure conversation quality and coverage rather than raw activity, because activity is easy to manufacture and tells you nothing about future deals.
Why does proprietary deal flow matter more than intermediated deal flow?
Intermediated deals arrive in competitive processes where you bid against everyone else and win on price. A deal origination function reaches owners before that, so you are often the only firm in the conversation. We explain the contrast in proprietary versus intermediated deal flow.