Forecasting is the engine that powers workforce planning in every contact centre. Without accurate forecasts, schedules fall apart, service levels dip and costs spiral. As technology advances, we’re seeing more and more vendors market their forecasting solutions with phrases like “AI-powered,” “machine learning,” and “predictive modelling.” It sounds impressive, but is it always necessary?

In this post, we’ll unpack the real value of AI forecasting, where traditional methods can still outperform and what to think about before jumping into a new tool that promises intelligence on autopilot.

What We Mean by “AI Forecasting”

Let’s start with some context and purpose: a working definition. AI forecasting generally refers to the use of machine learning algorithms that analyse past data to detect patterns, predict trends and adjust forecasts automatically. This can include:

• Recognising seasonality or recurring volume patterns
• Adapting forecasts based on anomalies (e.g. unexpected spikes)
• Continuously learning from new data to fine-tune predictions

This represents a shift from traditional statistical models, such as weighted moving averages or exponential smoothing, which rely on preset logic defined by a planner.

But here’s the key: more intelligence doesn’t always mean more accuracy.

Traditional Forecasting: Simple Yes but Still Smart

Traditional methods aren’t just “manual” or outdated. In many environments, they’re ideal. Especially when historical data is consistent and business conditions are relatively stable.

Methods like:

• Time series forecasting (e.g. ARIMA)
• Interval-level pattern analysis
• Historical overlays

…can all yield highly accurate results without the overhead and technical debt of AI. Time series forecasting methods, such as those found in Aspect Workforce, enable your planners to actively engage in the historical update process, which helps uncover insights from data that even AI might overlook, allowing them to build more accurate and informed forecasts. In some cases, taking a hands-on approach to analysis can be more beneficial than relying on automated tools.

At Call Design, we frequently see contact centres achieving impressive forecast accuracies of within +/- 5% without the use of AI. This success often hinges on the use of well-maintained historical data, reliable traditional methods and practical hands-on training, such as our WFM Essentials or Aspect Inbound Forecasting Training. In fact, for many organisations, this level of accuracy proves to be more than sufficient.

When AI Forecasting Shines

That said, there are real-world use cases where AI genuinely makes a difference:

• Highly variable environments where historical patterns don’t repeat cleanly
• Multi-channel operations where volume is influenced by less predictable drivers (e.g. social media mentions)
• External data modelling, such as using marketing campaign schedules, weather events, or competitor actions

In these scenarios, machine learning can alleviate some of the workload on planners by providing insights more quickly and reducing the need to manually adjust models each time the volume changes manually. The question you need to keep in mind though: Does this approach improve the practicality of your forecasts compared to traditional methods?

The Hidden Trade-offs of AI Forecasting

AI-based forecasting tools aren’t magic. In fact, many of them require:

• Clean and consistent data (garbage in = garbage out)
• Time to train the model, meaning early predictions might not be any better than traditional ones
• Transparency trade-offs, where users don’t know why the AI chose a particular forecast

And even in the best cases, many planners still want to apply judgment based on business context, something that AI will never fully replace.

The User Experience Gap

Another practical issue is that many vendors who advertise “AI-powered” forecasting still require a lot of manual configuration. It’s common for customers to buy into the promise of automation, only to discover that setting up the model, maintaining it and interpreting the results require as much work, if not more, than their previous method.

On some platforms, changing a forecast requires multiple screens, manual recalculations and thorough quality assurance (QA) before publishing it against schedules.

In our experience, we often observe a scenario like this: the customer uses the AI forecasting, it produces wildly unreasonable volume forecasts, and they manually adjust the numbers anyway. Or worse still, the only insight they can provide for variances in the forecast is “that’s what AI predicted.”

What to Consider Before You Invest

So, should you choose a WFM system with AI forecasting? Here are five things to consider first:

1. How predictable is your contact volume? If your call patterns follow a stable, repeating pattern, AI may not improve much.
2. Do you need external drivers? If factors such as sales, marketing campaigns or public sentiment have a significant impact on your volume, AI may be beneficial. Keep in mind that a good forecasting tool, such as Aspect Workforce, should enable you to factor in external drivers into your forecast (with features like campaign sets and holiday factors), without the unnecessary AI complexity.
3. What’s your team’s skill level? If your WFM team is comfortable interpreting models and managing data flows, you’ll likely benefit from AI. However, you should consider how you expect your WFM team to defend their forecast if the methods are locked away in an AI black box.
4. Can you explain your forecast to stakeholders? When it comes to forecasts, transparency is crucial; AI forecasts must be both explainable and accurate. Your WFM team works hard to achieve the best outcomes for your customers and employees, and all that effort can diminish over time if they can’t explain the ‘why’ behind the ‘what’.
5. Do you have the data maturity to feed AI properly? Clean data history, tagging and event tracking are essential.

Where Call Design Stands

We’ve worked with contact centres around the world for over 20 years. What we’ve seen time and again is that the best results come from choosing the right tool for the environment, not just the newest one.

Our customers using Aspect Workforce Management often achieve outstanding results using advanced traditional forecasting methods. The platform enables detailed control, unlimited what-if modelling and precise manual overrides, without locking planners into a forecasting black box or forcing them to create models outside the system (read: Excel) just to import them back in again.

That said, we’re also closely monitoring the development of AI forecasting. In the proper use cases, it can reduce manual work and support faster, more agile planning.

But if you’re being sold AI as a one-click solution that fixes forecasting forever, it’s time to ask more questions. 

The Real Intelligence is Knowing When to Use it

While machine learning offers benefits in terms of efficiency and speed, it may lead to over-reliance on algorithms that could misinterpret complex planning scenarios. Planners might become less skilled at manual adjustments and critical thinking if they depend too heavily on these automated insights. Additionally, machine learning models require significant data inputs and continuous updates to remain effective; without proper data management, the quality of insights can diminish, potentially leading to misguided planning decisions.

AI forecasting isn’t a must-have for every contact centre. It’s a powerful tool when applied correctly, with the correct data and in the right environment. But traditional models aren’t obsolete: they’re tested, transparent and in many cases, easier to maintain and explain.

The smart move? Don’t buy into the hype. Make your forecasting choices based on what improves accuracy, usability and confidence for your planning team.

We might get there one day, but for now, AI forecasting promises more than it practices for many contact centres.

Do you need help reviewing your current forecasting setup? Contact the Call Design team to see how we can help.

Written by Jamie Powderly, WFO Consultant Team Leader