From pilot to public: why alignment makes or breaks AI‍-‍driven customer experience

For the Australian public sector, the experience of everyday people as they interact with government entities is paramount. Artificial intelligence (AI) has the chance to change the game in how government organisations can improve their customer experience, but without proper internal alignment, AI-based CX projects may struggle to reach the public at all.

The rollout of production-ready, AI-powered CX products and features depends heavily on getting the right people to come together and agree on key points. Given the sensitive and sovereign nature of much government-held data, there can be a tension between stakeholders in charge of data access and data governance, and this can hinder the success of AI projects.

That’s why getting key stakeholders aligned early gives public entities a better chance to work with compliance mechanisms across departments and disciplines, striking the right balance between compliance and risk that is needed to get AI projects over the line from proof-of-value pilots to production. Broadly, it’s less of a technology problem and more of a business readiness matter.

Below are three key things for government organisations to consider when setting out to bring AI-powered CX products and improvements into production.

1. Gain clarity: align early on the problem to be solved, not the technology platform

The first step to success is deceptively simple: be clear about the problem you are trying to solve and intimately understand the ‘why’. The details may add complexity to the conversation, but the simple act of clearly articulating the business problem and why it should be solved provides a foundation for determining the best solution and accurately measuring its outcomes.

AI initiatives that start out as a proof of concept have a higher propensity to failure because they begin with the solution, not the problem to be solved. So, as part of the process behind aligning key stakeholders early on, be sure to ask: what is the problem we’re solving, for whom and why? This approach helps to avoid stakeholders interpreting the key objectives differently.

Articulating the problem clearly can facilitate early engagement and alignment across stakeholder groups, which is a critical step in getting AI projects off on the right foot. It is especially important to involve CX leaders, data owners, security and legal teams at this point in the process. Early buy-in from these stakeholders will help ensure success in production.

2. Define outcomes: replace product promises with internal proof points

Once the problem to be solved is agreed upon by all stakeholders and articulated clearly, it’s time to think about what success really looks like. This is where it becomes important to properly define the key outcomes of an AI-based CX initiative. Identifying the problem already puts teams on the path to settling on the most important outcomes for all stakeholders involved.

In certain sectors, scepticism around AI is prominent. This is particularly true in the public sector. Promises of certain public-facing outcomes for AI-based CX solutions are generally treated with suspicion — and rightly so. But if an AI capability can solve an identified problem in a measurable way, buy-in will be easier to secure.

What will change? By how much? What will it save us? What will we gain? Such questions, and the data points attached to them, can define meaningful outcomes upon which to measure the ROI of an AI initiative. Well-defined success metrics do more than justify investment, they help align stakeholders with very different priorities, reinforcing stakeholder support and cooperation.

3. Pathways to production: start small, but start with something that matters

With clear success metrics agreed upon across stakeholders and aligned across organisational departments, there is a shared baseline that can be used as a launch pad from which to kick off proof-of-value or pilot projects designed to demonstrate some tangible benefits. This will help to shape much of the initiative. But there may still be questions about the best pathway to broad adoption. This is where many AI initiatives quietly fail.

A proof of value, by its very nature, usually starts small. Traditionally, this is a sound approach for trials, especially when they involve AI solutions that will eventually be customer-facing. However, it’s important not to mistake a small project with low-stakes outcomes. Yes, there are good reasons to start small, but that shouldn’t mean you start with something insignificant.

The more insignificant the trial is, the easier it is to dismiss the outcomes, even if they’ve been clearly defined. Instead, a more effective approach is to begin with a use case that is genuinely important to the business. In the public sector, that may be something associated with access to government services, ‘ask me once’ experiences, or scenarios where accuracy is critical.

This approach forces teams to think meaningfully about what the project will look like in production. While citizens demand high quality and accurate AI-based CX innovation from the public sector, security and data privacy are non-negotiables. Getting teams aligned and governance sorted out early will set projects up for lasting success.

Image credit: iStock.com/Supatman