Startup scouting is how a company finds young firms that can solve a problem it has, and turns the best of them into a pilot. There are two directions. You can go looking: databases, events, accelerators, research. Or you can ask: publish an open call with a clear problem and let startups come to you. Most companies that do this well use both, and the open call is where most of them go wrong.
This guide covers the ways to scout, when an open call beats a search, how to run one from brief to pilot, and what venture clienting adds. For the bigger picture, start with what open innovation is.
What is startup scouting?
Startup scouting means systematically looking for startups whose product or technology fits a need in your business, checking whether they are real, and deciding which to work with. It is different from investing: a scout is looking for a supplier, a partner or a pilot, not a stake.
There are five common ways to do it:
| Way | How it works | Good for | Watch out for |
|---|---|---|---|
| Databases and platforms | Search startup databases by technology, market or funding stage | A long list, fast | Profiles are often out of date; you see who is listed, not who is best |
| Events and demo days | Meet startups at trade fairs, pitch events and accelerator demo days | Seeing the team, quick first filter | Random: you meet who happens to be there |
| Accelerators and VCs | Ask their portfolio teams for introductions | Pre-checked startups | You see one network’s portfolio |
| Research scouting | Search patents, papers and technology reports for who is working on the problem | Deep technical questions, early technology | Finds research groups as well as startups; needs expertise to read |
| Open call | Publish one problem and let startups apply | Specific problems, comparable answers, startups you did not know | Needs a good brief, a jury and fast answers, or it backfires |
The first four are searching; the fifth is asking. Searching gives you a list, while asking gives you answers to your question, written by the startups themselves, in a form you can compare.
When does an open call work better than searching?
An open call is the right tool when:
- the problem is specific. “Cut packaging weight by 20 percent without changing the filling line” gets useful answers. “Sustainability ideas” gets brochures.
- you do not know who the right startups are. A search only finds what is labelled the way you expect. A good call reaches the startup that solved the problem in another industry.
- you need to compare like with like. Every applicant answers the same questions, so the jury compares answers, not sales decks.
- you can promise something real. A paid pilot, access to a site or data, a named owner. Startups have little time; they answer calls that can turn into business.
Searching is better when the field is small and known, when you need one specific technology fast, or when confidentiality means you cannot describe the problem in public. In practice, scouting and calls feed each other: the long list from a search tells you who to invite to the call.
How do you run an open call to startups?
These ten steps take a call from idea to pilot. A first call usually runs eight to twelve weeks end to end.
- Pick one problem with an owner and a budget. The owner is the person whose team will run the pilot. If nobody owns the result, nobody will act on it, and the call becomes innovation theatre.
- Write the brief. Problem, constraints, what a good answer looks like, what you offer and the timeline. See the template below.
- Decide the scoring criteria and the jury before you publish. Three to five jurors: the problem owner, a technical expert, someone from procurement or legal. Agreeing criteria first stops the loudest juror from deciding.
- Keep the application short. Eight to ten questions at most. Ask for what you will actually score; anything else costs good startups time and loses you applicants.
- Publish where startups are. Accelerators and incubators, university innovation offices, industry associations, LinkedIn, and direct invitations to the startups your search turned up. A page on your own website mainly reaches people who already know you.
- Keep it open for four to six weeks and answer questions in public. Publish the questions you get and your answers, so every applicant has the same information.
- Screen, then score. A quick screen removes the clearly off-brief answers. Then each juror scores the rest on their own, against the agreed criteria, before the jury meets.
- Invite a shortlist to pitch. Three to six startups, 30 minutes each, with the problem owner in the room. Ask how they would run the pilot, not just what the product does.
- Agree a pilot, not a programme. Small, paid, time-boxed and with success criteria written down before it starts. More on this below.
- Answer everyone. Every applicant gets a decision and a short reason, including the no’s. Startups talk to each other, and the next call depends on this one’s reputation.
Our guides to writing a challenge brief and running a first challenge in ten days cover the same mechanics for internal challenges; most of it carries over.
What goes in the brief?
A brief that startups can answer in an evening, and that your jury can score, has seven parts:
| Part | What to write | Example |
|---|---|---|
| The problem | One paragraph, in plain language, with numbers | Our secondary packaging weighs 340 g per case. We want to cut it by 20 percent. |
| Why it matters | The business reason, so startups can judge the size of the opportunity | Packaging is our second-largest material cost and a target in our 2028 climate plan. |
| Constraints | What cannot change | Same filling line, same pallet pattern, food-contact safe |
| What good looks like | How you will judge success in a pilot | Tested on one line for 8 weeks, no increase in damage rate |
| What you offer | Pilot budget, access, data, a named contact, possible contract | A paid pilot at one plant, with a packaging engineer as contact |
| Timeline | Deadline, shortlist date, pitch date, pilot start | Open until 15 March, pitches in April, pilot from June |
| What you will not ask | IP, exclusivity and confidentiality terms at this stage | No confidential information in the application; no IP transfer |
The last row matters more than most companies think. Asking for confidential details or IP rights in the application is the quickest way to lose the startups you want most. Ask for a public-level description first; sign an NDA with the shortlist.
How should the jury score startup applications?
Score each answer on the same few criteria, from 1 to 10 or 1 to 5, with a short note on the reason. A common set:
| Criterion | The question behind it |
|---|---|
| Fit with the brief | Does this solve the problem we described, within our constraints? |
| Technology readiness | How far along is it? Lab prototype, tested in a real environment, or already in use? |
| Time to pilot | Could it run on our site within three months? |
| Track record | Has the team delivered something similar, with a customer we can call? |
| Commercial fit | Could we buy this at scale if the pilot works? Price, support, size of the company |
For technology readiness, many companies use the technology readiness levels (TRL) the European Commission uses in its research programmes: from TRL 1, basic principles observed, to TRL 9, a proven system in operation. A pilot usually needs TRL 6 or higher, a prototype shown in a relevant environment.
Score on your own first and discuss afterwards. Our guides on scorecards and scoring models and prioritising ideas go into weighting and calibration.
What is venture clienting, and how does it fit?
Venture clienting is a way of working with startups as a customer rather than an investor: the company buys the startup’s product for a pilot, tests it in a real business unit, and, if it works, buys more. Gregor Gimmy developed the model at BMW and set up the BMW Startup Garage in 2015 to run it. It has since spread to other large companies, especially in Germany, and banks and consultancies there now pitch it to the Mittelstand as well.
| Venture clienting | Corporate venture capital | Accelerator | |
|---|---|---|---|
| What the company does | Buys a product and pilots it | Buys a stake in the startup | Runs a programme for early teams |
| What the startup gets | A paying customer and a reference | Money | Coaching, network, sometimes money |
| Time to a business result | Months | Years | Years, if ever |
| Who needs to be involved | A business unit with a problem | An investment team | A programme team |
Venture clienting and open calls fit together. The open call finds and compares the startups; venture clienting is how you work with the winner. The rules that make venture clienting work are the same ones that make a call worth answering: a real problem, a business unit that owns it, a paid pilot and a fast purchasing process.
How do you set up a pilot that leads somewhere?
Many startup collaborations stall between a successful pilot and a real contract. Startups call it pilot purgatory. Five rules help:
- Pay for the pilot. A paid pilot gets the startup’s best people and shows your organisation is serious. Free pilots signal neither.
- Fix the scope and length. One site or one process, usually 8 to 12 weeks.
- Write the success criteria down before you start. The same criteria from the brief, with numbers. If they are met, what happens next should already be agreed.
- Fast-track purchasing. Agree a simplified procurement route for pilots under a set amount. A startup cannot wait six months for a supplier number.
- Give it an owner in the business, not only in the innovation team. The owner decides whether to roll out, so they should be the ones running it.
Which tools do you need?
For a first call, an online form, a shared inbox and a spreadsheet will do. They stop working when several calls run at once, when a jury of five scores sixty answers, or when you need to show what happened to last year’s applicants.
Tools fall into groups by the job they do. Startup databases and dealflow platforms help you search. Solver networks bring in experts you have never met. Some idea management platforms let you run calls to partners you invite, with a jury that scores in one place. Hives does the last: each call is a challenge of its own, startups answer one short form without creating an account, guests see their challenge and nothing else, and the jury scores every answer on the same criteria. On Pro and Enterprise you can connect Findest, which needs its own subscription, to search research and startups already working on the problem. Hives does not include a startup database or investment tracking. Our open innovation software guide compares the options by job.
Frequently asked questions
What is startup scouting?
Startup scouting is the systematic search for startups that can solve a specific problem in your business, followed by checking them and choosing which to pilot. It is about finding suppliers and partners, not investments.
What is the difference between startup scouting and an open call?
Scouting usually means searching: databases, events, networks and research. An open call means asking: you publish one problem and startups apply with their answer. Searching gives a list of names; a call gives comparable answers. Many companies search first and then invite the best names to the call.
What is venture clienting?
Venture clienting means a company becomes a startup’s customer instead of its investor. It buys the product for a pilot in a real business unit and buys more if the pilot works. The model was developed at BMW, which set up its Startup Garage in 2015 to run it.
How long does an open call to startups take?
A typical first call takes eight to twelve weeks from publishing the brief to choosing a pilot: four to six weeks open, two to three weeks for screening and scoring, and a week or two for pitches and the decision. The pilot itself usually runs another two to three months.
Should we pay startups for a pilot?
Yes, in most cases. A paid pilot gets the startup’s attention and best people, and it tests whether your organisation can actually buy from a startup. Free pilots often end without a decision.
Do startups need to sign an NDA to apply?
They should not need to. Ask for a description they are comfortable sharing in public, and sign an NDA only with the shortlist, when you start discussing details. An NDA up front puts off many of the startups you want most.
Sources
- Global Venturing, Why the venture client model is “king” for startups (Gregor Gimmy on the origin of the venture client model at BMW)
- INSEAD, case study BMW Startup Garage
- Startbase, Interview with Gregor Gimmy
- European Commission, Horizon 2020 Work Programme 2014–2015, General Annexes, Annex G: Technology readiness levels