Who is querying

The view that justifies the whole section, because it needs both planes and nothing else can produce it.

The view that justifies the whole section, because it needs both planes at once and nothing else can produce it.

Loghound already classifies every address it sees in a web log. Opensolr already records every address that queried a search index. Cross-referencing them answers two questions neither side can answer alone.

01 · The two questions
WEB LOGevery address, classifiedSEARCH LOGevery address, countedJOIN ONthe addressBOT TRAFFIC, COSTING YOUsearch work done for automationSEARCH WITH NO WEB TRAFFICdid not come through your site

Figure 1 — the correlation. Two calls total, whatever the number of addresses: one facet on the platform and one faceted query against Loghound’s own sessions index, joined in the panel’s own code.

How much of the search backend is working for bots

That turns an abstract bot problem into a concrete cost. Search capacity is the expensive part of a site; knowing that a measurable share of it is being spent on traffic already classified as automation is a number you can act on, whether that means blocking, caching or simply budgeting honestly.

Which search traffic has no matching web traffic at all

An address that queries your index but never appears in your web log did not come through your site

That is a leaked API key, a scraper that found the endpoint, or an integration somebody set up and forgot about. It is invisible from either side alone, and it is the finding that most often surprises people the first time they look at this page.

02 · Two calls, not sixty

The busiest addresses and handlers come from one facet on the platform, bounded at sixty terms. What those addresses did on the website comes from one faceted query against Loghound’s own sessions index. They are joined in memory.

A lookup per address would be the obvious implementation and the wrong one — sixty round trips to answer what one facet answers.

03 · When the correlation cannot be computed

Holding Opensolr credentials but not running on the machine that serves the website is a perfectly reasonable way to install Loghound — and in that case there is no web plane to join against.

The card then says what it would have told you, in terms of the numbers already on screen. It never renders a broken table and never blames a technical problem that does not exist.

The same applies with no Opensolr credentials at all: this whole section renders an explanation of what it is for and where the two values go. Settings is where they are entered.

Loghound is open source and MIT licensed. Questions about the Opensolr half — the account, the indexes, the plan — go to opensolr.com/contact; questions about the software itself belong on GitHub.

Loghound Documentation