Successful AI project launch: Sigli’s insights

Artificial intelligence continues to change the way businesses operate. When applied sensibly, this technology opens up a wide range of new possibilities for efficiency, innovation, and profitability. 

You can get the full webinar recording by following this link.

At Sigli, we strongly believe that we should use our deep business and tech knowledge to help individuals and companies uncover the practical value of cutting-edge AI-powered tools. One of our latest initiatives in this space was a free educational webinar for SMEs titled “The AI Profit Toolkit”. The event was hosted by Sigli’s Chief Business Development Officer Max Golikov.

This article includes the key takeaways from the webinar and offers actionable guidance for anyone looking to launch AI projects that deliver real value without excessive risks.

What you should know about AI before you even start

One of the most important principles to understand before diving into AI is that it is not always the answer.

AI is a powerful tool. But it doesn’t mean it should be applied universally. In many situations, simpler, more cost-effective technologies may yield better and faster results. The key is to find out why you want to use AI and whether a non-AI solution might accomplish the same goal with less complexity. If there are other approaches to addressing your needs, it may turn out to be much more sensible to use them instead of investing in an AI project.

Nevertheless, we by no means deny the business value of AI. It’s absolutely measurable and wide-ranging. According to industry insights shared during the webinar, AI’s contributions to business value break down into several categories:

• 57% comes from boosting productivity, primarily through automation.
• 17% stems from automating routine daily tasks.
• 16% is linked to enhancing customer experience.
• The remaining 10% covers a range of other use cases.

The transformative power of AI is especially tangible when applied thoughtfully to pain points that drain time and resources.

AI value matrix

To evaluate whether your future AI project is worth your time and money, we recommend you use an AI value matrix. This strategic tool helps find the right balance between the technical feasibility and business value of your ideas.

According to this approach, on the technical side, you will need to assess:

• Data quality. The foundation of any successful AI project is high-quality data. Projects with clean, abundant data move faster and require fewer resources at every stage of development and maintenance.
• Technological complexity. Before committing to an AI project, you need to consider the resources needed to move from idea to impact. Low-complexity solutions (like plug-and-play SaaS and low-code components) tend to score best on the AI value matrix.
• Availability of talent and skills. Even if your problem seems to be solvable with AI, you also should carefully analyze whether you have experts with the required skills in-house or whether you are ready (and have the budget) to hire reliable partners. Moreover, if you choose the second path, make sure that you have a hand-off plan.

On the business side, it will be necessary to focus on such aspects as:

• Alignment with business goals. AI should be tightly aligned with your strategic objectives or profit and loss (P&L) metrics. This ensures you are not chasing tech trends for their own sake but solving real, impactful business problems.
• Sponsor support. Every AI project needs a committed sponsor. In the ideal scenario, it should be someone who owns the budget and the performance metrics that the project aims to improve. For example, projects that have a C-level sponsor are likely to move much faster than those supported by people without the authority to sign cheques or make decisions independently.
• Measurable KPIs. Your AI use case should be built around a small set of clear, quantifiable KPIs. For instance, these metrics might include time saved, cost reduction, revenue growth, or user engagement. It makes sense to choose the KPIs that are already tracked and understood in your organization. This will help you avoid debate over measurement methods and simplify the “before vs. after” comparisons.

The projects that score highest across both dimensions are those with the highest likelihood of success and return on investment.

Choosing the right AI pilot: A real-world example

During the webinar, Max Golikov shared a couple of powerful case studies to demonstrate what you should pay attention to while considering different ideas for your AI projects. One of the examples was Sigli’s cooperation with a mid-sized UK apparel e-commerce retailer.

The company had an annual revenue of £48 million and 220K monthly orders. 12 fulfillment staff processed orders, returns, and refunds via three legacy tools (Magento, Zendesk, and homemade Excel macros).

When the company turned to Sigli, they had serious issues with the time needed for different processes. The returns and exchange queue averaged 5000 orders, which caused 7-day refund delays and seriously hurt the company’s rankings on Trustpilot. Apart from that, manual triage was pretty expensive as it cost £3.40 per refund.

Another problem was the fact that the customer support team had to spend 30% of time copy-pasting return merchandise authorization (RMA) info instead of applying upsell techniques and driving sales.

To solve the described issues, it was offered to consider different ideas, including an LLM-powered returns-triage bot, a demand-forecast bot, and a personalized upsell recommender.

As the AI value matrix shows, the first idea won.

For such a solution, the company had rich training data. The team had access to 95,000 labeled Zendesk tickets, which made it easy and fast to train the AI model.

From the integration perspective, there were no serious pitfalls. The solution only needed to connect with the Zendesk API and an existing RMA. It means that the implementation of this bot didn’t require a full system overhaul.

Moreover, it was quite simple to detect clear success metrics. Such KPIs as cost per return and refund turnaround time were already being tracked daily. Thanks to this, it was possible to measure the immediate project impact.

The project also could receive strong business sponsorship. The Fulfillment COO both owned the P&L and sponsored the pilot.

All this led to a successful AI implementation and tangible outcomes:

• Cut cost per return to £1.90 (a 44% reduction);
• Reduced refund delays to 2 days (a 72% improvement);
• Boosted Trustpilot score by 0.3 stars.

Key things to keep in mind to maximize the ROI of your AI project

Let us briefly summarize what we’ve discussed above. Based on our practical experience, we recommend focusing on the following points to ensure your AI initiatives deliver real value:

• Quick-win quadrant. Want to achieve quick wins? Select projects that score at least 7+ on technical feasibility and 8+ on business value on the AI value matrix.
• Clean and accessible data. Well-organized data always saves time and reduces rework.
• One clear KPI. Choose a metric that you have been already tracking for a while. Thanks to this, performance can be easily compared before and after implementation.
• A single, accountable sponsor. A budget-owning stakeholder accelerates decision-making and ensures follow-through.
• Measurable deliverables. Projects should be much more than just flashy tech demos. They should yield clear financial results.

Closing thoughts

Artificial intelligence offers incredible opportunities to streamline operations, reduce costs, and deliver standout customer experiences. However, all benefits become available only when this technology is used deliberately. It means that AI should be guided by clear goals, supported by strong data, and evaluated by hard metrics.

At Sigli, we can provide support for your business through every phase of the AI journey: from ideation and pilot planning to full-scale deployment. If you are ready to turn AI potential into business value, we are here to help.