Value comes from what you achieve together – not the price of the initial proposal

🇫🇮 This blog is originally written in Finnish. You can read the Finnish version by using the language switcher on the top right corner.

How to buy data and AI consulting services the right way

Too many have experienced a tender where price decides everything and a project where the cheapest option ultimately turns out to be the most expensive. I understand that price is the easiest thing to measure. But I’d encourage you to also measure the value created and to give it more weight in how you run procurement processes.

Data and AI are becoming such strategic assets for many companies that the related capabilities must exist in-house. At the same time, access to talent can be challenging. That’s why experienced consultants are valuable: they bring best practices from the market and provide temporary support for larger initiatives. Over the past six months, we’ve seen more of these larger projects being tendered than in the previous two years combined. This is a positive development; companies should absolutely be leveraging data and AI to drive their business forward.

procurement should not be done based on outdated principles

A lot has changed over the years. The quality of tender materials and the way the requested solution is defined often vary greatly, which directly impacts the quality of the proposals received. For example, it can be difficult to clearly describe the current state if the data environment has evolved into “spaghetti” over the years and suffers from heavy vendor lock-in. At the same time, too little time is often spent defining future business use cases and vision. Vendors are sent a mixed collection of materials, with the hope that their responses will somehow clarify the situation internally as well. Of course, there are exceptions, but this pattern is common. There’s also often a desire to reduce risk by favoring fixed-price deliveries. However, this doesn’t necessarily support the value of the end result, where open communication and collaboration play a critical role.

Throughout the years, vendors have learned how to play the game when only price is measured. A traditional way to win a tender is to offer a fixed price that’s too low; and then, early in the project, start discussing “unexpected findings” and how the price needs to change. This approach is problematic in many ways and can ultimately lead to a more expensive outcome for the customer. The client may be under time pressure and unable to restart the procurement process. More importantly, it breaks the foundation of a strategic partnership, which depends on trust and continuous dialogue. Fixed pricing also tends to turn the vendor’s focus inward: how to deliver the project within the agreed time and budget. This can lead to unpleasant surprises after delivery, when the result doesn’t actually meet the client’s real needs.

Two years ago, we proposed a project that would have cost several million. Today, we would offer the same project at a fraction of the price. How?

Data and AI sit at the core of strategy. In my view, companies should build strategic partnerships around them rather than optimize individual projects in isolation. I understand the need to manage risk and that investment decisions often require a clear price tag. But the way data and AI initiatives are bought should evolve alongside their strategic importance. During a project, continuous dialogue is essential to ensure direction, alignment, and knowledge transfer. The procurement model should support this. Data and AI projects are not “just” IT projects; they are enablers of strategy execution.

At Recordly, we sometimes act “naively” in tenders. Based on our experience, we provide a price that reflects what we believe the project will actually cost with an experienced team, including surprises. Even in purely technical projects, we invest time in understanding the underlying business needs. Because of this, we’ve lost deals; either on price or because we’ve asked too many questions. Still, we believe in our approach in the long run. Interestingly, losing a deal has often been the beginning of a great partnership. When a project has started with another vendor and eventually has run into trouble, we got the call.

AI-assisted development, AI agents, and specialized tools have rapidly evolved in recent years. Yet too often, they’re still underutilized. There are barriers on both sides.

Vendors may hesitate to disrupt their business models; after all, junior staff need billable work. At the same time, client organizations may not yet be ready to demand more automated, AI-driven solutions. I can relate to this. If you’ve spent years in the same company, it’s not easy to stay on top of a rapidly changing market after a full day of meetings. As a result, the scale of the change and the opportunities may not be fully understood.

The pace of change has surprised us as well. We’re not worried about AI taking our jobs, but we are concerned that not using AI might cost someone else theirs. From this perspective, we sometimes see middle management unintentionally slowing down AI adoption in our client organizations. We see AI as an enabler, something that strengthens capabilities and shifts work toward higher-value tasks.

Our strategy is to lean into this transformation, and I’d recommend the same to business leaders. Disruption is happening, and the rewards are being distributed now. With our team size, we see mostly upside. But for organizations with large junior teams or heavy reliance on nearshore models, the impact can be significant. Overall, the demand for data work is growing and there’s room for many types of delivery models. A project we might have delivered cost-effectively with nearshore teams a few years ago would today be done by a small senior team, supported by virtual data engineers and AI agents. This shift is not always fully understood and questions related to it are often missing from tenders. A company may choose the cheapest proposal without realizing the project could have been delivered far more efficiently with the right approach.

Do you want to create value or just get the project done?

We’re rarely chosen based on hourly rates. But we are chosen when the value of the overall solution is understood.

So when you’re running a procurement process, consider this:

• Prioritize long-term collaboration and continuous development over optimizing a single project
• Don’t just ask for cost; ask what value will be created and how it will be measured
• Consider how the procurement model supports (or limits) effective collaboration
• Evaluate the team’s expertise, references, culture, and how well they align with yours
• Look at total cost of ownership (TCO), not just project cost hidden costs often exist across departments
• Ensure that knowledge created during the project is transferred to your organization
• Make sure you’re fully leveraging the potential of AI in the project
• Ensure you have the capability to assess the substance of proposals (e.g. AI agents). 

The cheapest option is not always the best, but the best option is often the most cost-effective in the long run.

If you’d like to discuss how to build value together, feel free to reach out.

More reading 

• An interview with Pete Nieminen on how to find a reliable Data & AI partner

• My own thoughts on whether it’s time to rethink how data work is done.