Reading Time: 9 minutes

by: Zank, CEO of Bennett Data Science

Introduction

Choosing the right firm to bring data science to life at your company is a difficult decision. In this article, I provide a framework for executives looking to hire a data science firm you can trust. This methodology will equip you to minimize your risk, maximize your revenue, and move forward with confidence.

This decision framework includes the following seven questions you must ask yourself before hiring a data science firm.

  1. What could you achieve with data science? – Start by understanding the possibilities. Then, gain a basic understanding of what types of problems can be solved and what results can be achieved using data science.
  2. Is it the right time to invest? – Avoid shiny new thing syndrome by asking hard questions about whether your organization is prepared to invest in data science.
  3. What are the risks when hiring an outside firm? – Know how to spot red flags and address possible issues well before you sign a contract.
  4. What are the rewards when hiring an outside firm? – Consider the clear advantages of an outside firm over hiring in-house, including cost of services, speed of delivery and quality of work.
  5. How do I find and interview a firm? – Understand the best ways to find a firm and to navigate the tricky process of assessing whether their skills match your needs.
  6. How will I ensure measurable success? – Work together to articulate what the firm will deliver in terms of measurable results. This is why you hired them.
  7. Is it really over? – Define your endpoint clearly. Address the possibility for additional engagements by communicating early and often.

Question I – What could you achieve with data science?

Before you consider hiring a data science firm, it’s helpful to understand what data scientists deliver. Data scientists build statistical models that make predictions. These predictions help your company to increase revenue.

Common types of predictions include recommenders, dynamic pricing, churn modeling, fraud detection, and customer segmentation. Let’s look at a few examples to bring it to life.

Amazon has models that predict what to show each shopper based on his or her tastes, which increase average customer spend. Netflix personalizes your media recommendations, making the product stickier and increasing customer retention. Lyft predicts how likely a customer is to ride based on supply and demand and adjust their prices (hello 2x surge charge) accordingly.

While the nuts and bolts vary from one business to the next, data science is all about using data to make smart predictions about how people will behave in order to maximize revenue. If your company collects data, then data science can likely help you grow faster and more efficiently.

Question II – Is it the right time to invest?

Once you’ve gained an understanding of what types of problems can be solved using data science, you’re ready to evaluate whether it is the right time for your organization to invest. Data science is undeniably trendy right now, but that does not mean every organization is well suited for it yet.

In the world of data science, the ‘right time’ means that you have a stable technical platform, you have collected reliable data, and you have enough runway (money in the bank) to support the effort.

Here are a list of good questions to ask your data engineering and/or IT teams:

  • Do we have a consistent audience and a consistent way of collecting data?
  • What type of data do we currently collect from our product/audience/website?
  • How far back does our data go? How reliable is it?
  • Is our platform stable?
  • Is our IT team prepared to support the requests that a data science firm will have or have you considered bringing on a data engineer?

If you’re struggling to get clear answers to the questions above or simply aren’t sure if you’re ready to invest in data science yet, consider our Machine Learning Readiness Assessment, a one day intensive onsite where we answer each of these questions and position you to use your data to grow your business.

Question III – What are the risks when hiring an outside firm?

This might be the most important question to ask yourself before bringing in an outside firm. The following conditions are red flags to be aware of.

Organizational

  • Data difficulties
    • An external data science firm is going to need access to your data. If your team doesn’t have the time to pull the data, or provide instructions to do so, bringing them onboard is probably not a good idea.
  • Perceived threat
    • If your internal team feels challenged by an outside team, bringing in a firm can stifle growth. The key here is communication. Talk openly with your team about what you expect from the firm and how your team feels about the collaboration.
  • Improper buy-in
    • If you’re only considering this because your board told you data science is the future, think again. Commitment to data science within the organization is imperative. Without it, data science initiatives sit like unpicked grapes on a winter vine.

Technical

  • Limited IT resources
    • Ultimately, any data science firm will need to deploy their work, which means to embed what they’ve created in your website or product. The firm will either deploy the models themselves, typically with some infrastructure an IT team has provisioned for them, or the IT department will handle the deployment. Either option will require time and involvement from IT. This must be budgeted for, otherwise all of the work of the data science team has been in vain.

Question IV – What are the rewards of hiring an outside firm?

Now that we’ve considered the risks of engaging an outside firm, let’s consider the rewards. There are generally three conditions under which a company hires outside data science help. The company has:

  • no data science team or analytics professionals
  • a data science team, but they are overwhelmed or underperforming
  • a data science team but needs organizational realignment

For each of these conditions, bringing in an outside firm promises efficiencies in speed, people management, cost and quality.

Speed

Outside data science experts with skills specific to your needs can deliver quality products in a fraction of the time required to find, hire and train your own team. Data science firms shorten the product delivery timeframe while leaving you with a world-class system to build upon.

The output you’ll receive from a six-month engagement with seasoned experts could take upwards of 18 months with a junior team that is still learning the technical ropes, as well as how to work together. Naturally, junior data scientists (those who have been working in the space for less than five years) require a lot more time than their senior counterparts.

People Management

It’s tough to manage data scientists, due to the advanced math and complexity required to perform the job. Nearly the entire field is jargon! This typically requires the oversight of managers and the involvement of project managers to translate between teams and stakeholders.

External firms tend to be more self-sufficient and versed in communicating with executive audiences. After all, firms won’t land many contracts if they can’t make their work simple for companies to understand.

Cost

External firms are brought in for their value and expertise. The best firms provide value-based pricing with established deliverables. The cost of the data science team should always be a fraction of the value delivered. This makes costs defensible and manageable. Additionally, costs are often less than full-time employees, as firms don’t require benefits, specific timeframes and protection if things need to pivot.

Quality

The quality of predictive models varies widely. Like DevOps, data science done wrong is brittle and prone to breaking down upon changes to data or parameters. Be sure the firm you choose has deep experience delivering production code and models. Ask for real-world examples of the work they’ve done. Nothing is more important than verifying that the firm you’re interviewing has actually produced models, in production, and at scale.

Outside expertise also brings in a few sets of fresh eyes. Sometimes a new, professional perspective is all it takes to see the forest for the trees.

Question V – How do I get, find and interview a firm?

Now that you’ve assessed your readiness, as well as the risks and rewards of bringing in an outside data science firm, it’s time to begin your search. There are plenty of firms out there, but a simple Google search won’t necessarily reveal the one essential quality you are looking for: trust.
Above all, you’ll need to find someone you trust.

“There are three things a client needs to trust you: certainty that you will deliver; the belief that you’re worth more than your bill rate and the ability to articulate his or her needs to you.” – from Forbes

Here’s how to get started:

  • Reach out to your network to find a reputable firm with experience in your sector.
  • Use Google to find articles addressing the type of data science you need (e.g. product recommender or facial recognition) and write the author directly.
  • Ask a data scientist who works for a firm you know and respect if they have any ideas or connections.

Do not be afraid to reach out to several firms for an exploratory interview. An experienced firm will quickly qualify you as a company they can help or not, and in the latter case, may be able to refer you.

When interviewing firms, look for the following:

Technical Experience

  • Domain expertise
    • Find a firm that has done something similar to what you need. Someone versed in personalization for apparel would likely be a great fit for personalization of food items, but may not be a great fit for text-based personalization.
  • Infrastructure familiarity
    • Make sure the firm you’re interviewing is familiar with the infrastructure and cloud or local environment you’re using. If you’re using Google Cloud and a team only has experience in AWS, it may not be a big deal, but know that your IT team will likely be asked to help out more than usual.
  • Code compatibility
    • If you need all your code written in C#, that’s another thing to figure out early on. Generally, data science is done in Python. Even at scale, Python is used to drive other packages that achieve big data prowess.

Organizational Fit

  • Soft skills
    • Data science isn’t all pocket protectors and taped glasses. We’re nerds, yes, but you can, and should, demand top-notch communication from the firm you hire. Ask the firm how they have handled communication effectively in the past. If they can’t describe what they did before, chances are, you’ll have little idea what they’ll do when they’re in your company.
  • Compatibility
    • Find out who will be managing your project and talk with her or him. You’ll be spending a lot of time receiving results from this person, so making sure you jive is important.

Question VI – How will I ensure measurable success?

As I mentioned above, firms typically provide value-based pricing with clear deliverables, which makes it easy to track progress.

Examples of deliverables might include:

  • Increased revenue through increased user engagement driven by intelligent personalization
  • Decreased churn or fraud resulting from user behavior modeling and prediction
  • Customer segmentation, helping multiple teams understand users and their desires

For any of these deliverables, you’ll want to define specific KPIs and agree how success will be measured at the beginning of the engagement. What data will be collected? What is the baseline for comparison?

Maintain open communication to ensure that you are working towards your goals, but also be willing to pivot if the firm discovers opportunities that align better with your overall business objectives.

Question VII – Is it really over?

There’s no hiding this one: the consulting firm you hire would probably love to extend their contract with you. It’s so much easier to continue working with a current client than to find another client.

This is advantageous for both sides: you both know each other and there’s likely been some good rapport built over the time of the engagement. The truth is, with a good firm that delivers high-value, revenue-driving work, you’re probably going to want them to stick around too.

This is the sweet spot. Now you’ve got a person or team that knows your products, your data, and your workflow. It’s a no-brainer to keep them around. If it didn’t work out, that’s unfortunate, but at least it’s a short term commitment and it’s easy if you don’t want to re-engage.

In order to make sure that both parties are comfortable with the end or extension, be sure to define your expectations early and communicate throughout the engagement about what is going well and where there are opportunities for improvement.

Closing Thoughts

In this article, I’ve covered the seven questions executives must consider when looking to hire a data science firm. You’ve learned what data science can achieve, when and how to invest, and what you need to know to find and engage a firm.

If you’re ready to bring what you read to life, write us at info@bennettdatascience.com or ring us at (858) 945-2303. The first time we talk, we’ll show you exactly how we can increase your revenue. On principle, our first consultation is always on us. We’d be honored to work with you.

Zank Bennett is CEO of Bennett Data Science, a group that works with companies from early-stage startups to the Fortune 500. BDS specializes in working with large volumes of data to solve complex business problems, finding novel ways for companies to grow their products and revenue using data, and maximizing the effectiveness of existing data science personnel. https://bennettdatascience.com