Data science, big data and AI are current buzzwords, and many companies are rebranding their business intelligence products with these definitions to get a more significant market share. Although some of these companies have indeed created valuable products and service bundles, others are just coating their old offerings in a shiny new package to answer clients’ demands.
>See also: The UK’s top 50 data leaders 2017
With such a broad offering available it has become difficult to choose the right vendor, and an appropriate structure needs to be put in place as an algorithm to follow during the selection process.
What do you want to achieve?
It is recommend that you start any data science acquisition process with an inwards analysis. Why do you need these tools? Which gaps are you trying to fill? Do you already have a defined scheme and just need the answers? Give accurate definitions of what you are trying to achieve, like better segmentation, less churn.
It is a strategic mistake to buy a data science solution just to keep up with trends and annihilate the fear of missing out on this opportunity before having a clear understanding of what success would mean for this project.
Minimal standards
Before contacting any vendors, it is wise to create a baseline through an initial report of the existing state. Based on this first assessment, you should create a list of minimal must-haves for the selected vendor which will help you rule out a lot of undesirable offers.
>See also: 10 UK-based data entrepreneurs you should know about in 2017
Don’t settle for anything less than your minimum list and come up with an evaluation system for the additional features or ‘nice to haves’. By creating an algorithm, you are protecting yourself from emotional decisions based on the vendor’s marketing strategy.
What cost structure can you afford?
Budget is the cornerstone of any project, and there is no way around this. An expert from InData Labs explained the alternatives in this case, which are summarised as:
• Hiring data science experts. This is the costliest option since it means going through the hiring process and adding more people to your payroll. Only advisable if you intend to make data science one of your revenue streams.
• Staff augmentation. Raise the skills of your existing staff by bringing in experts to help them leverage some time-sensitive issues.
• Project-based. A cost and time effective approach which is highly suitable for clearly defined projects or to kickstart a collaboration with a vendor, during a trial period.
>See also: Data governance vendors running hard but skipping first mile
• Hybrid centre of excellence. This partnership between in-house teams and external specialists can be done through an ongoing process inside a CoE. An excellent choice for companies that are considering the opportunity of having an in-house team at some point in the future.
• Outcome-based. Paying based on outcomes requires a great collaboration and clear metrics to measure success, to validate the project and release funds.
Cloud-based or on-premises?
Your data security and compliance regulations could influence this decision or force you to opt-in for on-premises. If you have a choice, cloud-based solutions are usually more cost-effective in the short term but can add up over time. An in-house solution requires consistent initial investment, as well as maintenance and upgrades.
If your data vendor proposes a cloud arrangement, be sure to ask about encryption, security and data ownership. Read the fine print to be sure that you are not giving up any rights and that the uptime is guaranteed.
How to evaluate the vendor?
Start by researching a comprehensive list of potential vendors and run them through the minimal standards. You should be left with about 10-15% of your initial selection. Contact these companies with an RFP, stating your needs, describing the project and asking for a specific solution.
>See also: 8 global leaders inducted into the Data 50
Portfolio/reliability
If not provided explicitly on their websites, be sure to ask about previous similar projects. Ask for contacts and references from previous clients. A strong vendor will be proud to offer you these, while one with little or no experience will be reluctant.
Don’t just evaluate their technical skills, but also the way they deal with clients. Do they answer promptly? Is the information offered on point and presented professionally? You need to look out for these early signs since the quality of the relationship will be influenced heavily by such aspects.
MVP (Minimum Viable Product)
A high-quality vendor should provide you with a realistic solution, describing the way they will approach the situation, the steps they intend to take and setting some KPIs.
The more details, the better, since this only proves that the vendor took enough time to consider the project and they didn’t just recycle an old proposal or just offer a canned answer. Look for mentions of the budget and a timeline.
Room for growth
Make sure you select a vendor who has the skill and resources to grow with your project if needed. Ask about other options they have, except for the ones included in the offer to ensure they would be able to handle more work. Not all projects will evolve, and not all tasks require the involvement of a large vendor, but you must look beyond your immediate needs and think strategically.
>See also: Targeting the enterprise: 4 things SaaS vendors need to get right
Making the right decision
Much like any other business partner, a data science vendor must be able to help you achieve your business goals by offering you support in the areas where you lack the necessary skills. Don’t be afraid to ask questions related to their capabilities, management team, unique abilities, pricing structure, and more.
It is essential that businesses acknowledge there is no tried and tested recipe to select the right vendor, it is a particular process that they will have to design through successive iterations. In the best case scenario, you will be able to use it again for new projects.
Sourced by Emilia Marius, healthcare business analyst
