Insights is designed to help you detect performance shifts early and respond with confidence. When used consistently, it enables you to do a structured health-check routine for your merchandising strategy.

This article walks you through a practical workflow using multiple performance metrics on the Insights main dashboard and the ranking performance report.

Establish overall health

Every health-check begins on the Insights main dashboard. At this stage, the objective is not deep analysis but orientation. You are assessing whether overall performance behaves as expected.

Focus first on the Revenue and CTR tiles. Revenue reflects overall business impact, while CTR serves as an early signal of merchandising effectiveness.

Before interpreting the data, ensure that the selected date range is meaningful. For recurring checks, reviewing at least one month of data is recommended. This reduces the influence of daily fluctuations and reveals broader trends.

Note: If you are reviewing the impact of a specific recent change, such as a ranking adjustment or campaign launch, set the date range such that you can properly investigate the potential change cause, e.g. one month before until one month after the event. Always choose a time frame that reflects what you are trying to evaluate.

Pay attention to CTR

Although revenue is the ultimate business metric, it is influenced by many downstream factors: checkout experience, payment friction, stock availability, and basket abandonment all affect final purchase numbers.

CTR (Click-through-rate), by contrast, reacts immediately to merchandising decisions. When product visibility, ranking logic, or assortment alignment improves, CTR typically responds within hours. Revenue may follow later, and often to a lesser degree. For this reason, CTR functions as an early diagnostic indicator. A significant change in CTR, even if revenue remains stable, is worth investigating.

Take a close look at the CTR tile on your Insights main dashboard and identify possible abnormalities.

Interpret spikes and drops in context

Not every fluctuation represents a problem. Performance must be interpreted in context.

A spike may result from a promotion, product launch, or seasonal peak. A subsequent decline may simply represent normalization. However, if performance drops below historical baseline levels, especially outside expected seasonal patterns, further investigation is appropriate.

Ask questions such as:

• Was there a recent promotion?

• Has marketing launched a new campaign?

• Was there a product launch?

• Has a peak period (e.g. seasonal sale) just ended?

• Is this a normalization after an unusually strong period?

More generally: Is this expected behavior, or does it require deeper analysis?

Narrow the scope: search vs navigation

If you determine that performance has shifted unexpectedly, the next step is localization. Compare search and navigation performance to understand where the change originates.

When both search and navigation decline simultaneously, the cause is often broader in scope. This may reflect changes in traffic quality, seasonal shifts, marketing campaigns, or even technical tracking issues. While these root causes may lie outside of Fredhopper, they can still be addressed within the tool (see Address cause).

Warning: A sudden and dramatic drop in total events (traffic, clicks, revenue) should prompt validation of tracking and site stability before adjusting ranking logic.

If, however, search remains stable while navigation declines, the issue is more likely related to category-level merchandising, ranking configuration, stock concentration, or seasonal assortment alignment. In this case, proceed to ranking-level analysis.

Investigate further using the ranking performance report

The ranking performance report allows you to isolate which rules contribute to the change. By comparing multiple ranking rules, you can determine whether the issue is concentrated in a specific department, strategy, or rule configuration.

If you already suspect a cause, select the rules most closely related to that hypothesis and include a comparable rule as a reference point. If no clear hypothesis exists, begin with the rules affecting high traffic numbers. 

Switch between metrics such as view conversions (CTR), basket conversions, and purchase conversions to observe how performance behaves across the funnel.

Evaluate funnel behavior

When interpreting funnel metrics, consider statistical reliability. For example, daily purchase conversion rates can fluctuate significantly due to small sample sizes. If patterns appear inconsistent, aggregate data weekly to reduce volatility and identify meaningful trends.

A decline in CTR suggests reduced product relevance or visibility, e.g. there is a mismatch between what merchandisers are incentivized to advertise and what shoppers are looking for. A stable CTR combined with declining basket or purchase conversion may indicate that customers are engaging but not converting, possibly due to pricing, stock availability, or product alignment, e.g. decreased stock availability at the end of a season.

Note: Numbers alone do not provide context. After identifying where performance changed, evaluate what operational or marketing changes may have occurred during the same period.

Address cause

Even when root causes originate outside Fredhopper merchandising, such as seasonal transitions or changes in marketing traffic quality, adjustments within the platform remain possible.

For example, if seasonal demand shifts, weighting recent performance more heavily may align results with current shopper intent. If stock levels decline in key categories, ranking adjustments can emphasize in-stock products.

Insights not only highlights changes but it enables a structured response to those changes.

Check health frequently

Over time, this process becomes repeatable:

Begin with the dashboard and review Revenue and CTR over a meaningful time frame. If performance deviates from expectations, determine whether the change is contextual or requires investigation. Compare Search and Navigation to localize the issue, then use the Ranking Performance report to isolate contributing rules. Evaluate funnel metrics carefully, accounting for statistical noise. Finally, form a hypothesis grounded in business context and implement adjustments where appropriate.

Ensure to do regular health updates as described above on a weekly basis. However, if you want to track how deliberate changes to your merchandising strategy effect performance metrics, check Insights more frequently after the change has been made.