The Enrich Button: Amazon Dash for Business Memory

Well's Enrich button fills 23 fields from 6 data providers in 2.3 seconds. One click on a company record. Company name, address, revenue, employee count, tech stack, funding history. Done.

The Eleven-Minute Problem

A user spent 11 minutes populating a single company record. She opened Clearbit for the address. LinkedIn for headcount. Crunchbase for funding data. Three browser tabs, three copy-paste cycles, one record.

This wasn't unusual. This was the workflow. The average record had 6 fields populated out of 23. Not because the data didn't exist. Because nobody had time to enter it.

Waterfall Logic: Not All Providers Are Equal

The hard part isn't fetching data. It's knowing which data to trust.

Six providers feed the Enrich pipeline. Each returns data with varying reliability depending on the field type. Provider A has accurate revenue estimates but outdated addresses. Provider C nails tech stack detection but misses legal names.

Revenue data might flow Provider B, then A, then D. Address data follows a completely different order. Each value carries a confidence score. Above 0.85: accepted automatically. Between 0.50 and 0.85: flagged for review. Below 0.50: discarded.

91% of fields resolve above 0.85 on the first provider. The remaining 9% cascade through an average of 2.4 providers before resolving. Total time: 2.3 seconds for all 23 fields.

The Behavioral Shift Nobody Predicted

Amazon discovered that Dash button users didn't just reorder faster. They reordered more frequently and tried new products. Removing friction didn't optimize existing behavior. It changed behavior entirely.

Same pattern with Enrich. Users who click the button once start clicking it on every record. Then they ask what else the system can do automatically. Then they connect integrations. Then they build workflows.

73% of users who trigger Enrich in their first session become weekly active users. Without that first click, day-30 retention sits at 34%. Same product. One button press separates retained users from churned ones.

When the Machine Doubts, Humans Get Better

Confidence scoring created an effect we didn't plan for.

When the system flags a field at 0.72 confidence, users verify it. They check the source, correct when needed. When those same users type data manually, they rarely double-check their own work. The machine's uncertainty makes humans more careful than their own certainty does.

Before Enrich: 6 of 23 fields populated. After: 19 of 23. That's the difference between a contacts list and actual business memory. You can't run analytics, spot patterns, or automate outreach on 6 fields. On 19, you can do all three.

From Data Entry to Data Use

Manual data entry is a tax on attention. Not just time. Cognitive load.

Every minute copying an address between tabs is a minute not spent assessing whether a deal is real. Not noticing that a prospect's tech stack overlaps with three existing customers. Not doing the work that actually moves revenue.

Well's Enrich button processes roughly 15,000 requests per day across 239 workspaces. The metric that matters isn't volume. It's what happens after. Records with enriched data generate 3x more follow-up actions than those without.

The Gateway to Automation

Every product has a moment where the user first believes the tool is smarter than their manual process. For Well, that moment is Enrich.

One click. Low risk. The result is immediate and verifiable: the data is right or it isn't. No learning curve, no configuration, no workflow change required.

That single click shifts the mental model. The user stops thinking of the system as storage and starts seeing it as something that works for them. Every subsequent feature adoption builds on that initial proof.

The Enrich button isn't Well's most profitable feature. It trains users to trust automation. Once they believe the system knows things they don't, they stop going back to spreadsheets.