Data and analytics have such a central role in creating value, delivering core strategies and supporting business-as-usual that they might seem obvious candidates for a permanent internal department. Even so, no business is able to cover off every aspect of these, especially where specialisms are concerned. From web analytics to data science, resource constraints force nearly all firms to bring in third-parties at some point or to package up projects or project elements for creation and delivery by a business partner.
A DataIQ Leaders roundtable in May 2021 brought together members from broadcasting, infrastructure, insurance, law and retail to discuss how to work with outsources. This briefing brings together their views on when projects should and should not be handed off and how to ensure success when they are.
Drivers of outsourcing
Choosing to partner in order to realise a project is generally the result of a number of common challenges:
Lack of capability – Even the largest and best resourced data department will have gaps in its capabilities, while some skillsets are hard to hire or too expensive to operate in-house, such as data science. In this scenario, an external outsourcer is a rational option to plug that gap.
Funding – The business case for using an outsourcers is different to that for build a capability in-house. Contractors can be paid as capital expenditure, rather than operating expenditure, which means they are covered out of the data department’s discretionary spending, rather than the organisation’s fixed overhead. Sign-off by the finance department is easier to get as a result.
Employee turnover – Analysts and data scientists can get bored, especially if overly-focused on business as usual, rather than innovation. This erosion of headcount can impact directly on business-critical projects and processes, which means finding cover wherever it is available.
Core competency v specialism – An important driver of outsourcing is where it is just not worth investing into a specialism as in-house headcount. Some of these competencies are relatively low-value, yet essential, meaning there is little consequence of having them delivered externally.
New ideas – External consultants bring fresh thinking or a different perspective which it can be hard for incumbent data and analytics teams to have. This can be valuable at the strategic level, where management consultancies are routinely brought in to advise on new directions of travel, but also around emerging technologies which the organisation may not have the time to spot and consider.
Barriers to outsourcing
Even where the drivers noted above have been recognised, that does not automatically result in a decision to outsource a project or process. Inhibiting factors may prevent this and force the data department to achieve an internal workaround:
Day rates – Demand has pushed these upwards which creates a significant drain on capital resources that may lead to resistance by the finance department, or simply put outsourcing as an option beyond the affordability of the business.
Knowledge transfer – Clients worry about outsourcers using a project to gain insight that they will then apply to a different organisation. (Although this is obviously one of the reasons why an organisation uses an outsourcer in the first place.) Outsourcers can also use projects as an opportunity to “land and expand” by learning about the business, identifying gaps in its capabilities, then pitching its services against these. As one data leader bluntly put it, “they can feel like a parasite”.
Internal resistance – Some departments, leaders or even teams may have a sense of ownership or a belief that any project can be delivered with in-house resources or is so essential to the business that it can not possibly be outsourced. Given the use of external partners for sensitive or vital activities, such as legal or financial advice, this is not rational and usually reflects an emotional investment that needs to be addressed.
Data access – Care needs to be taken from a compliance perspective when using a third-party to develop authoring using data, but also to preserve the integrity of data and the insight it provides to the client. Controls over access and limits on processing should be in place. Also, an important management task is to ensure external analytics teams do not go down “rabbit holes” out of curiosity – they need to remain focused on the specified task.
Project management arrangements
Balancing the drivers for outsourcing and the obstacles that will be encountered is an important task for any data leader. By putting the right arrangements in place from the outset, the chances of success are increased. These include:
Time-defined projects – A major element in the good running of a project is having clear start/stop points with milestones and KPIs along the way. Deadlines focus minds as do metrics.
Extending the team – Where outsourcing is being used to cover for capability gaps, this may involve creating a virtual partnership that embeds the outsourcer into the in-house team. This can be effective as it ensures the right questions get asked about the business and its operating environment. When a new project comes to deployment, IT obstacles are often a cause of failure, such as a retailer who worked with a business partner on a voucher code project only to discover the proposed solution would not allow for online redemption, only in-store. Aim to create a properly integrated team and ensure common goals and milestones. Ensure skills match the roles – this is not always easy especially in a complex domain like data science. But be aware that certain roles prefer to work in isolation, but this is not optimal for project outcomes – data scientist and data engineer were called out in this respect.
Roll-on, roll-off – The purpose of key milestones in the project is to ensure the team profile matches what is required for the next step, not for the step just completed. This is achieved by rolling-off outsourced resources that have completed their phase and bringing in the next set of skills.
Project management issues
All outsourcing projects will run into issues that have not been foreseen. Recognising what these might be in advance can help to ensure mitigating measures are available when they arise:
Is the project tightly specified? Mission creep is a concern for data leaders. Specific requirements need to be set out, such as programming languages or preferred tech environments that any solution needs to be delivered against. Transferring between these can be very difficult after the event.
Is the partner suited to the specification? Different types of outsourcers are good at delivering directly against a simple specification, while others will challenge and push the boundaries. This is not always a bad thing, especially if the driver of outsourcing is a need for new ideas.
Does the data leader have the capacity to manage the project closely? Onboarding and roll-off are time-consuming. Co-ordinating calendars for multiple project stakeholders and contractors is particularly difficult. Offshore contractors mean the working day is extended for the project leader. In advance of starting the outsourcing project, the data leader needs to consider the impact this will have on their own capability during its course.
Is there a direct face-off between all stakeholders and the contractor? IT and operations may not have been involved in the project plan and specification – they can be disruptive and hard to keep onside.
Do both sides have a translator? Outsourcers will not understand the business and need a stakeholder to translate its processes and culture; internal communication within the business partner also needs to align with this culture. On the client-side, this can become onerous for subject matter experts who get overloaded with demands from the outsourcer.
Have the risks been clearly identified and mitigated? IT cost over-runs, model failures, contractor errors are all possible and need to be called out in a risk register in advance.
What is the migration or exit strategy? Organisations need to be able to lift-and-shift a solution into a new environment, including on-premise or cloud, as they transform their technology stacks. An outsourced solution should not be an end-point or technological cul-de-sac.
Conclusion
Outsourcing can be effective and deliver a new solution rapidly. Or it can become challenging with scope creep and missed deadlines. Much of the success lies in the hands of the data leader and the extent to which a project is carefully specified and managed, along with the choice of the right business partner and their adherence to that specification.
Top Tips
- Spend time setting clear parameters for the project, even if it is urgent.
- Ensure coding language has been included in the technical specification.
- Ensure the incumbent tech stack can accept developed solutions.
- Include training in the specification (and costing) if new tech forms part of the solution.
- Don’t over-promise – outsourcers will habitually say they can deliver against any specification, but there is usually a stretch involved.
- Don’t over-complicate – the simplest solution is less likely to fail and more likely to deliver value, even if this is below what a more complex approach might yield.
- Respect the client’s authority – outsourcers can sometimes be following a playbook which may conflict with the client’s view on next steps. Remember, who pays the piper calls the tune.
- Be clear who owns the IP – knowledge transfer is inevitable, but where a new model or process is created, clarity is essential over ownership between the client and the outsourcer.
- Set milestones and deadlines with a clear exit data for the outsourcer.
