How WEX shipped AI-powered embedded analytics in 90 days

Most SaaS teams budget the better part of a year to build customer-facing analytics into their product. WEX did it in 90 days. That gap is worth understanding, because the timeline wasn't luck: it came from a specific set of choices any product team can learn from.

Embedding analytics into a SaaS product is harder than it looks. Each customer has to see only their own data, so multi-tenancy and row-level security aren't optional, and dashboards have to feel native rather than bolted on. End users who aren't analysts still expect to ask their own questions without filing a request. Building it in-house means months of engineering, and buying a tool that can't handle tenancy, branding, and self-service just moves the problem.

Building from scratch gives full control but consumes engineering time most teams don't have. Embedding a traditional BI tool starts faster, but it often can't deliver multi-tenant security or a native, white-labeled feel, and self-service tends to fall short for non-technical users. WEX paired a platform built for embedding with natural language querying, so security and branding were handled by the platform while end users could ask their own questions. That combination compressed the timeline without cutting the capability.

This piece covers what embedded analytics is and why SaaS teams build it, then breaks down the decisions behind WEX's 90-day timeline and the platform capabilities worth evaluating for a build of your own.

What is embedded analytics?

Embedded analytics places charts, dashboards, and natural language querying directly inside the product a user already works in, rather than in a separate BI tool they have to log into. Access, branding, and permissions inherit from the host application, so the experience feels native instead of bolted on. That's the core distinction from standalone BI: standalone tools serve analysts inside their own interface, while embedded analytics serves customers inside the SaaS product itself.

Why SaaS companies build embedded analytics

Why do SaaS teams build embedded analytics instead of pointing customers to a separate reporting tool? The value shows up in a few consistent outcomes:

• Users get insight inside their workflow instead of exporting data to another tool.

• Analytics becomes a product differentiator and, for many companies, a paid line item.

• It deepens engagement and retention, since customers who find answers inside the product stick around.

• It reduces the report-request load on internal engineering and support teams.

The WEX story: shipping AI-powered embedded analytics in 90 days

WEX Field Service Management serves more than 35,000 contractors in HVAC, plumbing, and electrical services, who spend an average of five hours a day on the platform. When homegrown reporting couldn't keep up, WEX rebuilt its analytics using ThoughtSpot Embedded. For the full story, read how WEX built AI-powered embedded analytics in just 90 days and WEX's self-service analytics journey.

The challenge

WEX's original reports ran on homegrown, PHP-based code pulled directly from production databases. Field technicians hit five-minute timeout windows without getting data back, and engineering was buried under hundreds of one-off requests for a single field or filter. An earlier attempt with a traditional BI vendor made things worse: it shipped as what one WEX product manager called "a single iframe box," a bolted-on tool that meant nothing to the technicians using it.

What made 90 days possible

The first move wasn't choosing an analytics vendor; it was fixing the data. WEX migrated off production database queries into a Snowflake-powered warehouse and layered a semantic model on top, so business terms could be understood consistently and validated before reaching a customer.

With the data ready, WEX rolled out ThoughtSpot Embedded in three layers: familiar stock reports for continuity, interactive Liveboards for real-time KPIs, and embedded natural language querying for ad hoc questions. A team of three to four engineers and product managers ran two-week delivery cycles with releases twice a day, involving customers in design from the earliest stage. That crawl-walk-run pace, not a single big-bang launch, is what fit the work into 90 days.

The role of AI and natural language

The natural language layer, which WEX branded internally as AssistIQ, changed what end users could actually do. Instead of learning a BI tool or filing a ticket, field technicians could ask a question in plain language, get an answer, and follow up without starting over. Within 90 days of rollout, 65% of user interactions involved AI-powered queries rather than static reports, and report delivery went from five-minute timeouts to under three seconds.

What to look for in an embedded analytics platform

WEX's build worked because the platform handled requirements that would otherwise have consumed months of engineering. The same criteria apply to any team evaluating embedded analytics platforms or embedded analytics solutions:

• Multi-tenancy and row-level security, so each customer sees only their own data without custom code per account.

• White-labeling, so the analytics experience carries the product's own branding, not a separate vendor's look.

• Self-service and natural language querying, so non-technical end users can ask their own questions.

• Governed definitions, so an answer means the same thing everywhere it appears.

• Speed to deploy and maintain, since a platform that takes as long to run as it did to build defeats the purpose.

What is the best analytics for SaaS?

There's no single best embedded analytics tool for every SaaS product; the right fit depends on how well a platform handles multi-tenancy, white-labeling, and self-service for your users. A platform built for embedding from the start, rather than a BI tool with an iframe bolted on, tends to close that gap faster.

Embedded analytics examples in SaaS

WEX is a direct example: field technicians now get answers about efficiency and operational gaps inside the product they already use, in seconds instead of minutes. The pattern repeats elsewhere. A fitness app might surface a member's workout trends right on their dashboard. An e-commerce platform might show a merchant their store's sales performance without a separate reporting login.

Are SaaS tools being replaced by AI?

AI is reshaping SaaS, not replacing it. Standalone AI tools still need governed data, permissions, and an interface to sit inside, and SaaS products supply all three. Embedded AI analytics, like what WEX built, is an example of SaaS absorbing AI rather than losing ground to it: the product stays essential, and AI becomes how users interact with the data already inside it.

Embedded analytics FAQs

1. What is embedded analytics?

Embedded analytics puts dashboards, charts, and natural language querying directly inside a software product, so users get insight without leaving the app.

2. What is embedded analytics for SaaS?

For SaaS, it means multi-tenant, white-labeled analytics built into the product itself, so each customer sees only their own data with a look and feel that matches the app.

3. How long does it take to build embedded analytics?

A from-scratch build often takes the better part of a year. WEX shipped a production experience in about 90 days by pairing a data foundation move with a platform built for embedding.

4. What should you look for in an embedded analytics platform?

Multi-tenancy and row-level security, white-labeling, self-service and natural language querying, governed definitions, and speed to deploy and maintain.

5. Does embedded analytics require a data team?

It requires clean, well-modeled data, but not a large team. WEX ran its build with three to four engineers and product managers once its warehouse and semantic model were in place.