A data science consultant is an external person with expertise and experience in data science and AI, who assists companies and organisations to get the best value out of their data. It’s an interesting job with a mix of the technical side (such as data analysis, statistics, and software engineering) and the business side (understanding problems in a particular industry).
A data science consultant is usually hired by an organisation because they needed an external expert with skills that may be hard to source in-house, or they may not have capacity for the data science projects that they want to complete. For example, an in-house data science team may not have enough experience in natural language processing or another niche area of data science and so they may look for a consultancy which specialises in that field, like Fast Data Science, which focuses on NLP.
Many companies are not necessarily tech-focused. In particular medium-sized corporations are increasingly turning to data science consultants to fulfil their needs in this area.
Above: some of the results a data science consultant can achieve for a business
A data science consultant’s job is varied and no two days are the same. Here are some real examples of data science projects, from Fast Data Science’s case studies:
Data science consultant - London
Below you can see a small demo of a clinical trial cost prediction model which Fast Data Science has been working on.
| Trial is for condition | |
| Phase | |
| Vaccine trial? | |
| Country | |
| Sample size | |
| Number of interventions |
Check out our case studies page if you’re feeling inspired!
We can help you get value from your unstructured data, such as PDFs that may be floating around your business. We can analyse your data and present it in a jargon-free way using graphical techniques, and move on to building explainable predictive models.
For example, if you want to understand the causes of customer churn or employee turnover, Fast Data Science can identify key correlations and even find causal mechanisms. We can recommend strategic actions (e.g. “send a coupon to customer #2123 on X date”). We can even train and deploy a machine learning model which can highlight any employee or customer at risk.
Our main area of focus at Fast Data Science is natural language processing (NLP), which means that we do a lot of work with unstructured text data (documents in English or other natural languages).
A data science consultant should ideally have qualifications and experience both in machine learning and in the industry that they’re consulting in. A minimum of a Masters degree in data science or a data science adjacent field is a must.
For example, the director of Fast Data Science, Thomas Wood, studied Physics as his undergraduate degree at Durham in the UK, and then completed a Masters in 2008 at Cambridge University in Computer Speech, Text and Internet Technology. Since then, he has been working exclusively in machine learning and data science, with a large number of engagements in particular in the pharmaceutical industry. He has also completed other professional certifications such as the Bond Solon Expert Witness program and the Microsoft Azure Data Scientist Associate certification.
Examples of what a data science consultant can do for a business include:
Fast Data Science’s data science consultants have built machine learning models and consulted for companies across industries and of all sizes. Below you can see a selection of some of our past consulting engagements, and you can find out more by reading our case studies.
An open-source, NLP and generative AI-driven tool for harmonising questionnaire items, even across different languages, developed for the Wellcome Data Prize in Mental Health. Try Harmony online.
Learn More
Foreflight provides a powerful mobile app used by commercial pilots for flight planning and in-flight naviation. It is an electronic flight bag which runs on iOS and iPadOS devices. We analysed Foreflight's customer database to identify any users likely to churn (cancel their subscription).
Learn More
The Office of Rail and Road (ORR) is the British national rail regulator, responsible for health and safety on mainline rail, the London Underground, light rail, and trams.
Fast Data Science built a drag-and-drop dashboard which allowed internal users at the ORR to analyse data collected on train delays, flooding, accidents, construction, etc, and build machine learning models on the fly.
Learn MoreA Natural Language Processing-powered tool developed for the Bill and Melinda Gates Foundation, assisting experts in estimating the risk of a clinical trial ending uninformatively. The Clinical Trial Risk Tool also reads a clinical trial protocol and creates an itemised per-subject budget from the schedule of events, looking up costs in hospital charge masters. Try the Clinical Trial Risk Tool.
Learn More
For a number of large pharma multinationals, we have been developing natural language processing models to redact sensitive data (PHI) or commercially sensitive information from clinical trial narrative reports, protocols, statistical analysis plans, synopses, ICFs, and other documents.
Learn More
Tesco is the UK's largest grocery retailer with over 15 million regular shoppers. We developed machine learning models to predict the amount in kilos that a shopper will put in their basket, as well as the turnaround time of delivery vehicles, in order to improve route optimisation and driver rostering.
Learn MoreIf a data science consultant is hired for a foundational data science project such as proof of concept, data exploration and data governance, or AI strategy consulting, you should agree on a measure of success which reflects the value for the business. Often a proof of concept may not lead to immediate profit, but it will lead to a more powerful production system further down the line, and so it should be viewed as an investment. However, a proof of concept should still work and be of use. The measure of success for an early stage system could be simply that it works with a reasonable level of accuracy, precision, recall, or some other metric, within an acceptable time.
For a project that involves implementing a production system, such as a live customer analytics model, the measure of success could be something as explicit as change in customer lifetime value, increased customer retention, or even impact on turnover. Fully mature data science projects usually lead to a cost saving (for example, fewer staff needed to do the task that is automated), better forecasting, or risk reduction.
Once a mature data science project has been completed, you can calculate the ROI as:
ROI = net benefit from data science project / total cost of project * 100
Since the cost of a data science project for a company with 10,000 customers is not too far from the same project for a company with 100,000 customers, there’s a huge leverage effect when data science is done at scale, and a data science project can pay for itself many times over.
Above: a video detailing how natural language processing consulting could benefit a business. This is one area of data science consulting which we do at Fast Data Science.
We always begin with the question you want to answer:
Following a rigorous and numbers-based approach, we proceed to find answers to the question using statistics and machine learning algorithms. At all stages we keep you on-board and provide easily interpretable visualisations of our findings, complete documentation, and interactive handover workshops. We work with both traditional machine learning techniques as well as the state of the art such as neural networks. We normally use Python for our data science consulting work, but can adapt to your technology stack. We often develop front ends for any models that we deploy.
We will not use unnecessarily complex or over-hyped models. If another data science consultant tells you they will solve your problem using an LLM connected to Blockchain, come to us instead and we might be able to give you a model you can work with on pen and paper! We prioritise simple models (Occam’s Razor) and make sure that the client is onboard with all the technology used.
We recognise that generative AI is a tool in the toolkit and we will use it for some of our projects with the client’s permission. At present, organisational policies on generative AI vary and there may be restrictions because of either the opaque (non-explainable) way that it generates predictions, as well as objections to the idea of sending data to a third party provider. However, times are changing. But it should be noted that simple problems that require only a regression model can be solved more cheaply and reliably with that regression model. Generative AI is exciting but it isn’t the answer to everything!
In some industries, such as healthcare, insurance, pharmaceuticals and legal, it is common for large amounts of the data to be stored in an unstructured format, for example PDFs, Excels, and other documents, which may be stored on the cloud, on on-premise servers, on employees’ laptops, or even physically. These kinds of documents are particularly difficult to analyse.
Fast Data Science has a proven track record of extracting useful information from gold-mines such as this for a series of high-profile clients. Our area of focus is natural language processing, which allows us to analyse data in free text, or even in a multitude of languages. Imagine you are an executive in an insurance company and you have a collection of a thousand ship inspection reports, which are written in plain text, submitted in Word or PDF form, and by people of different nationalities, in highly technical language. You want to know how many vessels failed the fire safety inspection. Do you:
Data science is being increasingly adopted across the healthcare sector. This technology is sometimes called healthtech or MedTech.
As an example, machine learning models can predict doctors’ workforce attrition, patient outcomes, or A&E or emergency room demand. Natural language processing can be used to analyse clinical data in text form and categorise (code) information by MeSH terms.
We have worked on a number of healthcare data science consulting projects, such as:
The technology we use for your project depends on you and your organisation’s preferences, but all other things being equal, we will choose the simplest tool to answer your questions in an efficient and explainable way.
You’ll be pleased to find out that data science is one of a company’s operations whose impact is most easily quantified – after sales, of course!
Taking an example from one of Fast Data Science’s case studies, the data science consultants identified a key pain point in the client’s customer sign-up form which was causing a percentage of the customers to drop out. The consultants could estimate from UX research and prior knowledge that an improvement would result in a 7% increase in sign-ups.
The CEO of the company was sceptical, so the consultants ran an A/B test for two weeks. Half of potential customers were presented with the old form, and half with the new form. After two weeks, the expected increase of 7% did indeed occur. The client company promptly switched the website over completely to the new design.
You may wish to measure the ROI of a data science project in terms of effect on turnover, customer acquisition, customer retention, or EBITDA and profit and loss. Non-monetary metrics include things like customer or employee satisfaction.
How much would your company benefit from a data science consultant identifying quick wins like this?
Our data science consultants have worked on a number of projects for major companies recently, including:
Please visit our Portfolio for more information, or simply contact us for a chat.
Yes, you can enter into a retainer agreement and have a data science consultancy like Fast Data Science on call. If you have need for C-level expertise in data science in your organisation, you can also hire a fractional chief data scientist or head of AI.
What we can do for you
