Similarweb - Demand Analysis

Context

Demand Analysis sits inside Similarweb's Web Intelligence suite, its job is to help teams answer one core question: how is consumer demand shifting, and what should we do about it?

Customers use it to size a category, watch trends over time, surface emerging behaviors, and align product and marketing with where attention is actually moving.

The module is built around three entities:

1. Keyword Lists: user-built or shared inside an organization.
2. Topics: predefined keyword lists curated by Similarweb.
3. Search Terms: on-the-fly lists generated from related keywords.

Personas & JTBD

Two core personas drive the module.

• Market Research / Consumer Market Insights (CMI) - a supportive function. Analysts who do deep insight work for the teams who then make the calls.
• Business Strategy - owners of the larger growth and expansion decisions, especially around the digital side of the business.

Both share these jobs-to-be-done: find new opportunities, understand market and consumer trends in order to sell the right products, and report on business performance in market context.

The Problem

The funnel data made the gap clear: of users who reached the Demand Analysis homepage, only ~34% went on to analyze a trend.

For benchmarking, we looked at a sibling module in Web Intelligence — serving similar personas through a structurally similar entry point. Its homepage was converting at ~51%. Same audience, same suite, very different result.

Closing that gap was the goal - and the direct KPI we'd track after launch.

Research

User interviews (Internal)

To understand the why behind the funnel drop, I ran interviews with internal users — analysts, CSMs, and PMs working closely with our customers, who could speak both to their own use and to the patterns they were seeing from the field. The themes were consistent and painted a homepage that worked against its own users:

• Lists overloaded the surface. Saved lists were dumped alphabetically with no sense of recency, relevance, or signal. Orientation was hard, and managing them was harder.
• Topics were hidden. Trending Topics — arguably the most discovery-friendly content we had — were buried in a separate tab that only 9% of users clicked.
• Search was narrow. It didn't reach across all entities, didn't surface topics, and the country filter wasn't legible.
• List-creation was a gate. To analyze a brand-new trend, users first had to create a keyword list. The main action ended up being "create a list," not "analyze."
• Comparisons reset. Once a user navigated away from a comparison, they had to rebuild it from scratch.
• No starting point. New users landed and felt the cold-start problem hard — nothing was personalized, nothing inspired the next step.

Competitor scan

Alongside the internal research, I looked at how adjacent products framed the same surface: Google Trends, Exploding Topics (Semrush), Glimpse, and Treendly.

The recurring pattern across the strongest ones: the homepage was a trend feed, not a file manager. Search-first, discovery-second, with personal saved lists as a side rail rather than the main act.

Strategy & Vision

Elevator pitch: Easily discover, track, and analyze the trends that matter most.

Research collapsed into six opportunities, each tied to a validated pain point:

1. Find faster — less friction in getting to a result.
2. Search any trend immediately — decouple analysis from list creation.
3. Provide a starting point — inspire new users instead of dropping them at a blank surface.
4. Make Topics discoverable — bring them into the main flow, not a side tab.
5. Surface trending lists & topics — with the metrics that inform action (volume, change).
6. Personalize — use industry, competitive set, and homepage topics to tailor the surface.

These collapse into 3 concrete flows that the module is built around:

• Find — "I have a hypothesis to verify." I have a trend in mind and need a quick read on its size and direction.
• Track — "I want to keep an eye on the trends that matter most to me."
• Discover — "I want to uncover what's hot in my categories."

Solution

Home page

The new homepage is built around three zones, each mapped to one of the JTBDs.

Find — A unified search at the top of the page lets users analyze any trend instantly. No list required: type a keyword or pick a saved entity, and you're in the report. This is the move that opens the door for Or while removing a friction Marta was working around.

Discover — A trending Topics surface, suggested from the user's website category, plus access to a full Topics tree. What used to be a 9%-clicked tab is now the main visual block on the page. We display each topic with the metrics that actually inform a "should I look here?" decision — volume and change.

My Lists — User-created lists, but sorted by recents and surfaced with the metadata users were missing (volume, change, key keywords). It functions like an at-a-glance dashboard instead of a static directory.

Report

On the report side, the same redesign supports the modes that drive deeper analysis — Single / Compare, YoY / PoP, and Clusters for Topics — across every widget. The three primary widgets remain Search Volume Over Time, Top Countries, and Keywords Trends, but every mode now stacks cleanly on top of every widget, so the report scales from one quick read to a full multi-trend study without changing pages.

Outcomes

The redesign launched in March 2025. The post-launch Mixpanel data shows the homepage conversion lift the redesign was built for, plus a clear behavior shift toward the new immediate-search flow.

• Homepage conversion (Gate → Analyze): 34% → 44%. A ~10-point lift, measured on the same 2-step funnel before and after launch — closing over half the gap to the sibling module.
• Search Term became the dominant analyzed entity. Pre-launch it was a niche path (188 unique users alongside 189 for Keyword Lists). Post-launch it surged to 719 — outpacing both Keyword Lists (426) and Topics (230). Surfacing it as the default in the unified search worked.
• Discover is still picking up steam and is expected to drive further conversion gains as users get used to the surface.
• Unified Search and Library were intentionally held out of this round and remain in progress.

Reflection

This was a project where the design problem and the business problem pointed at the same thing: a homepage that was managing files instead of starting analyses. Reframing it as a trend surface rather than a list manager let us serve our existing power users better while opening the door to the lighter users adoption depended on.