First of all, you may say that the writer of this article is a consultant himself.  With my experiences in the industry, I have been worked at “in-house” for risk management in a subsidiary of a bank and working as a consultant in most of my life including SAS Institute (HK) and other different consulting teams.  Thus, I would like to share my viewpoints & experiences from different angles including vendor, consultant company and user.

Before considering to hire a consultant / consultant company, the management of a business should fully understand the exact reasons for outsourcing.

There are lots of reasons to hire a consulting company (outsider) and some common ones are:

1. Internal staffs are lack of / with gap on the skills, knowledge gap with a product or technology to build up their data science solutions.
2. Internal conflicts and/or politics between different departments are the barrier for carrying data science implementations.
3. It is too risky for the project to be 100% handled internally. It is expected to have some valuable advice provided by the consultants.
4. Staffs within the organization should have the best knowledge in the business but they may not able to start their first analytics without proper skills and/or tools.

There are a number of facts to be considered:

1. Understand your pain-points (business problems) and define your goal

The well-defined problem is critical for which pain-point is going to solve with the support of data.  The goal with valuable result is only coming from the well-defined problem from the first day.

2. Future-proofing Solution Design

The landscape for data science is so broad.  There are 10 different data sources today and it could be doubled in the next year.  So, it is important to validate whether the solution suggested by the consultant is a scalable one with considerations on future development & growth.

3. Check the consulting team’s experience

There are lots of people claiming themselves as “data scientists” or experts.  However, it is important to check with the key consulting team members’ project experiences.  For Data Science consulting companies, they are in general with shorter history but it is vital to check on the experiences of the key team members.  If you are seeking for solutions with new technology or something without experienced internally, it is better to request a proof-of-concept or a pilot project in order to ensure consultants are qualified.

Further, the word “experience” could be tricky and it is more important to see their experience in the “business” area rather than types of customers.  To take an example, the credit card department of a bank should seek for a consultant with experiences in “Customer Profile 360” rather than the one with experiences in Finance & Banking.

4. Compare & contrast the investment amount on consultancy services

There are lots of different consulting companies providing data science services.  For instance, there are Information Technology company, Big Four Accounting firms, data science consulting company, etc.  You need to check with the quotation and try to ask them to break down their activities.  Some companies are billing extremely cheap and it is expected to see a team of young people for the implementation.  Big Four Accounting firms are charging high rates and they are more likely focused on strategy planning.   For a most proper team, it is important to have some experienced people for solution design, understanding requirements and providing advice; meanwhile, there are some less experienced people working on the data integration, programming and other implementation tasks for lower costs.  You should find the average man-day rate in your country for similar job.

5. Reputation for the company

It is important to ask for client reference if available.  To take an example, it is better to see repeated customers for a consultant company rather than 100 customer listed for once-off service only.  For Data analytics, it is never a once-off project because business environment is always changing and continuous development is expected to fulfil the business needs.  Further, it is possible to check with the social media profile like LinkedIn page and personal profile for the key project implementation member.

6. Reliability, Ethics and Data Security Consideration

For data science consultant, it is not only required to have technical skills but also other qualities like ethics and their data security policies.  It is better to see whether your suppliers have created codes of ethics.

7. Training / Skill Transfer to the stakeholder (in-house)

As data analytic is a journey, it is better for the consultants to provide guidance service to allow client users to grow their skills up to “citizen data scientist” level.  Moreover, it is important to hand over the consultant’s work as “project mode” to “operational mode” being maintained locally.  Also, it is important to have all corresponding materials like whitepaper, application user guide, data dictionary, etc.

8. Master Data Management

For choosing a consultant, they should be able to provide a data strategy implementation roadmap and corresponding recommendations to enable Master Data Management.  There are  number of elements in the MDM and it is vital for the consultant developing a data strategy for meeting the defined goal.

To sum up, there are lots of different factors affecting the selection of data science consultant team.  It is suggested to ask for Proof-of-concept (POC) before any commitment of service contract.  There are lots of people claiming themselves data scientists.  It is worth to invest some more time to verify their ability before hiring them.

Samuel Sum

Data Science Evangelist (CDS, SDi)

Vice President (AS)