CEDX Search · Platform
Five hundred fifty-eight thousand searches a month in the demo estate, and the console leads with the part most search tools average away: a 9.1% zero-result rate — 51K searches that found nothing — with the failing queries named and the fixes scored.
One support query returns nothing 84.7% of the time, 5,453 times a month. The console does not just count it — the synonym that would fix it is a scored action, waiting to be published.
Runs on demo data. Acme Retail Group in the captures is the software's demo workspace, not a customer.
What it is
The Queries table carries searches, CTR, zero-result rate, conversion, average position and intent on all 320 rows. Sorting by failure is the point: a cycle-count query at 84.7% zero-results and a receipt query at 40.5% are the rows a search team should start its week with.
The Clicks screen does not celebrate position one at 48% — it prints the expected band, roughly 45–55%, next to it. Click volume by result position (8,332 clicks at positions 2–3) and the no-click queue are on the same screen, so over- and under-performance are both visible.
Catalog at 78, Help Center at 76, Docs at 82, Blog at 80 — each with records, lag, searches, zero-result rate and facet coverage of its own. The worst zero-result rate in the estate, 10.9%, belongs to the help index, which is where users go when something is already wrong.
Product tour
Not a mockup and not a concept deck. This is what opens at /app/search.
01 — Overview
Index health reads 88, and below the metric cards the console lists what to do next — scored recovery actions: publish a synonym cluster worth +2.2K searches, pin a winner on a head query, ship a boost rule. Impact plus effort, ranked.
02 — Queries
Three hundred twenty queries with their full stat line. The head is healthy — a scanner query at 8,420 searches and 45.2% CTR — and the tail is where the work is: support queries with zero-result rates over 40% and average positions past seven.
03 — Clicks
Five buckets of result positions with the volume in each, and a CTR-by-position chart with its expected band printed. A top hit at 68.9% CTR on a returns query is healthy; the no-click queue next to it is the work.
04 — Indexes
Four indexes with independent health scores, lag measurements and sync times — the catalog synced 2 minutes ago, the blog an hour. Facet coverage runs 76% to 94%, and AI answers are on for two indexes, off for two, as a visible per-index choice.
Who runs it
Roles, not references. We have no named customers yet, so nobody in these photographs is quoted, credited or claimed as one.
Owns the zero-result board: 51K failed searches a month, the synonyms and pins that fix them, and the recovery actions ranked by impact.
zero-result · 9.1%
Owns the head: the 41.4% click-through on commercial queries and the pins and boosts the recovery list keeps scoring.
recovery upside · 8.1K
Owns the help index — 2,140 articles, a 10.9% zero-result rate, and the gap between what users type and what articles are titled.
help ZRR · 10.9%The shape of it
Every figure below is legible in the captures above. Nothing here is a projection of your estate — it is the state of the demo data.
How it runs
Four indexes sync on their own clocks — catalog 2 minutes ago, blog an hour — with lag measured per index so freshness is a number.
Every query earns its stat line: searches, CTR, zero-result rate, conversion, position, intent.
Failures arrive root-caused — a CTR drop traced to a title rewrite that buried the commercial match, not just a downward line.
Synonyms, pins and boost rules ship as scored actions — the A/B typo-tolerance test is already showing +3.4 points of CTR in the alerts.
One record
The indexes are built from what the other products hold — findability is a property of the whole estate, not a bolt-on box.
Classification and staleness in the file library are ranking signals here — a 63%-labelled library searches like one.
CEDX KnowledgeThe help index is the knowledge base searched: 2,140 articles, and the zero-result queries are the articles that do not exist yet.
CEDX CopilotAI answers are a per-index switch on the Indexes screen — the assistant answers from the same corpus search ranks.
CEDX EmbeddedSearch analytics become product analytics when the dashboards live inside your own app — same figures, your chrome.
Limits
Finding this out on the third call is worse for you than reading it here, and worse for us.
Start
Open it right now
Your indexes, your queries
Search with the rest of it
Questions
Yes. Every screenshot is a capture of the running build and you can open the same build at /app/search. It runs on demo data, which the page says next to the figures rather than in a footnote.
The share of searches that return nothing — 9.1% across the demo estate, 51K searches a month. It is the single most honest search metric: averages look fine while a support query fails 84.7% of the time, which is why the failing queries are named, not smoothed.
The Do this next panel scores recovery actions by impact and effort: a synonym cluster worth +2.2K searches ranks above a pin worth +1.4K. A/B results arrive the same way — typo tolerance is already showing +3.4 points of CTR in the alerts.
Search leading to an order or a ticket, at 4.2% in the demo configuration. It sits next to CTR and position because a clicked result that goes nowhere is a different failure than a query with no results.
Yes — the four demo indexes carry independent health scores, lag, facet coverage and an AI-answers switch each. The help index's 10.9% zero-result rate is a separate problem from the catalog's 9.1%, and the console treats it that way.
No certification has been issued. What we can evidence about hosting, encryption, tenant isolation and production access is written up on the security page.
Live build, demo data, no card. Then ask what your own site's zero-result rate is — and who saw it this week.