Big companies can too easily become victims of what I call “innovation tourism”: They partner with startups and VCs in an effort to innovate, but they wind up merely sampling new technologies and practices and failing to drive innovation and change throughout their organizations.

There are many reasons these partnerships fail to result in real, lasting change, but a lot of them have to do with how the partnerships are structured. There’s no single model that will work for all companies. Instead, each organization must find the approach that fits best and that maps to its own strengths.

Here’s a look at four different partnership models that can be effective:

1. An executive takes it on as a priority

In this model, an executive starts a program from (or near) the C-suite because his or her traditional vendors or partners aren’t providing a path forward for the company. In this setup, top company executives stay intimately involved in building out partnerships and trying to integrate new technologies into the company.

This model is often effective because there is buy-in from the top of the company. I have seen CEOs of major national brands make pilgrimages to Silicon Valley to educate themselves about new technologies and meet with specific startups. Often, this can accelerate the rate at which partnerships are forged.

At other companies, it’s the chief technology officer who leads the way. At Coca Cola, CTO Alan Boehme holds meetups on Sand Hill Road with venture capitalists and tech CEOs. He also sponsors an Israeli technology incubator called the Bridge. As part of the program, selected software startups join a six-month program to receive marketing training, access to mentors and a chance to pilot their technology inside Coca Cola, among other perks. Startups usually focus on products related to consumer engagement, consumer retail, supply chain, marketing or health and wellness—obviously, all areas of interest to Coca Cola.

2. A rank-and-file employee goes rogue — in a good way

At the other end of the spectrum is the “rogue” model, usually kicked off quietly by a rank-and-file employee. These in-the-trenches employees often have a more-informed view about specific technologies a company actually needs to improve its products or services. They also may be subject matter experts, capable of evaluating and recommending technology. So, harnessing these employees’ enthusiasm can be a good idea.

But there’s a big downside: When a program begins without high-level, executive support, projects often wind up without sponsors and, therefore, no budget.

Sometimes, lower-level employees also aren’t as savvy in negotiating business partnerships. I once witnessed a situation in which an employee at a big company approached an open-source startup for a proof-of-concept — but, he wanted significant help on implementation and training from the startup’s services team, free of charge. Satisfying this request likely would have cost the startup any shot at revenue from the partnership. So the startup scrapped it, and the larger corporation missed out on an opportunity.

3. Sending out a scouting team

Technology “scouting teams” are also popular, especially with big banks and technology providers who are interested in a broad set of technologies, from back-end IT infrastructure to new types of marketing technologies.

Here’s how it works: A small set of people in the company, in collaboration with business leaders, create a shopping list of unmet needs; then they try to find vendors to satisfy them. This structure is particularly practical for foreign companies that don’t have a strong pipeline of tech innovation inside their respective countries. In the U.S., however, most of the major Wall Street banks have people based in the Valley and other technology hotbeds. They continually network with their shopping lists in mind, and these people often have direct access to top bank decision-makers once they find promising tech.

The challenge of this model for practitioners is the ability of the scouting team to clearly articulate the requirements for specific projects and new technologies. Also, this team-based approach — similar to the “rogue” setup described above — can often mean the scouts lack authority to implement a proof-of-concept and pull the trigger to buy.

4. Parking a semi-autonomous engineering team in a tech hub

In this model, the company puts engineers on the ground in a technology hub, like Silicon Valley, and charges them with rapidly prototyping new technologies and experimenting with new vendors. This team is doing more than shopping – they are engaged with the business. The members have GitHub accounts, are writing code and white-boarding new solutions. They actively integrate technology into their product from outside providers; also, some failure is expected. It is a clean-slate approach that many companies, ranging from Ford to Capital One, have deployed to leverage talent in the Valley.

These teams often compete directly with local startups for talent and try to lure top engineers through the payoff of having their final products commercialized quickly inside a large company. Having so many engineers on the ground also means they’re closer to new technologies and can find solutions to specific challenges more quickly.

For example, Capital One currently has hundreds of engineers working in San Francisco’s SOMA neighborhood. My firm recently worked with the group to find a startup that could determine the eligibility of an individual for specific bank product offerings. Because of the Capital One team’s charter to discover, test, and implement new technologies quickly, they dedicated an engineer to work with the startup’s API and implement the technology inside Capital One. In cases like this, projects can get off the ground in days, instead of months.

Each of these structures has trade-offs. My advice is to set up your organizations to balance risk, cost, and positive outcomes. Big companies should stay fluid, building flexibility into their organizations so that they can adapt and change at the same rate as new technology. And by being strategic in setting up a corporate innovation program, they can steer themselves in the direction of smart technological evolution. Their future, after all, may depend on it.

Scott Goering is the vice president for business development at Battery Ventures. He was previously a business development executive at Intel Capital.