Natural language search layer for Adobe Analytics

The moment it breaks: an analyst needs a specific segment comparison before a Monday morning meeting, types something into the search bar inside their analytics tool, gets back a list of reports named with internal shorthand nobody outside one team understands, and spends 20 minutes clicking through nested menus before giving up and asking a colleague. This is a weekly occurrence, not an edge case.

The gap persists for a structural reason: Adobe and similar vendors have no incentive to fix navigation because power users have already memorized where things live, and those power users are the loudest voices in renewal conversations. The people who struggle — newer analysts, generalist marketers who dip in occasionally, anyone who joined after the last big UI change — have no concentrated leverage over the vendor's roadmap. So the interface keeps accumulating complexity while the search stays shallow.

What's actually broken, per the complaints: naming conventions inside these tools reflect how engineers organized the database, not how analysts think about questions. 'I want to see mobile conversion rate by campaign source last quarter' maps to three separate menu paths with names like 'Fallout', 'Flow', and 'Workspace' that don't suggest any of those things. Updates make it worse — 'I cannot find where things are' after a UI refresh is something analysts say out loud in Slack channels every few months.

This is a business and not a feature because the buyer — an analytics team lead or marketing ops manager — pays for it to reduce the time their team spends not doing analysis. Every hour an analyst spends navigating instead of interpreting is a direct cost. The need recurs every time someone new joins the team, every time the underlying tool updates, and every time the organization runs a new kind of question it hasn't templated before.

What to build

Build a Chrome extension that sits on top of Adobe Analytics Workspace and lets analysts type a plain-language question — 'show me bounce rate by device for paid campaigns this month' — then maps it to the correct report, segment, and date range and opens it directly, learning from how each org's instance is named.

Where to start

Start with a fixed library of the 20 most common analysis types analysts run in Adobe Analytics (campaign performance, mobile vs desktop conversion, funnel drop-off) and get those working perfectly before tackling custom variables — this wins the first customers without needing the hard per-org customization solved.

The hard part

Adobe Analytics instances are heavily customized — variable names, prop numbers, and eVar labels are different in every org, so the mapping layer has to be trained per-customer rather than once globally, which makes onboarding expensive and delays time-to-value.

How it makes money

Monthly per-seat subscription charged to the analytics team, starting around $30–50 per analyst seat; team leads pay for their whole team rather than individual analysts expensing it separately.