Organic Revenue Up 145% for a National Legal Nonprofit

THE CHALLENGE:

The organization’s content answered legal questions well for human readers and badly for the systems that now answer those questions first. People with an urgent legal problem increasingly ask an assistant before they ask a search engine, and assistants do not rank pages — they extract passages. Content written as long, unbroken prose gives them nothing clean to lift. Heading structure that varies from page to page gives them no reliable map of what a document contains. The organization had years of substantive legal content and very little of it was shaped for that kind of retrieval.

Two structural faults compounded the problem. Blog and news pagination canonicalized every paginated page back to the first, which meant the overwhelming majority of the archive was signalling that it should be disregarded in favour of page one — years of material effectively invisible to anything crawling the site. And the site carried no entity-level structured data connecting practice areas, case types, and editorial content, leaving machines to infer relationships that could have been stated explicitly.

THE STRATEGY:

Moving Traffic Media rebuilt the content for extraction rather than for ranking, on the premise that citation and ranking are different mechanisms with different requirements.

A page can rank well and never be cited. Ranking rewards relevance and authority at the document level; citation requires that a specific claim be locatable, self-contained, and unambiguous at the passage level. A 1,200-word explanation of a legal process may be the best answer on the internet and still lose to a thinner competitor whose version is broken into steps a model can lift without rewriting.

That reframing set the priority order. Content restructuring came first, because it determines whether anything else is worth doing — discovery protocols that deliver a model to an unextractable page accomplish nothing. Entity schema followed, to state relationships the content implied but never declared. The pagination fault was treated as urgent rather than as routine housekeeping, because no amount of restructuring helps content that has been instructed to defer to another URL.

It also set the scope. MTM evaluated 353 priority pages and identified 59 that needed structural work — roughly one in six. The remaining pages were left alone. Rewriting an archive wholesale is expensive, risks the rankings of pages that are already performing, and mistakes volume for diagnosis. The gains here came from changing a small, specifically identified set of pages rather than from touching everything.

DETAILS:

The engagement ran across four connected workstreams.

Content restructuring for extraction. MTM audited 353 priority pages against how language models segment and retrieve text, and identified 59 for restructuring — bulleted and numbered lists in place of prose sequences, consistent and descriptive heading hierarchy, and shorter self-contained chunks that carry their own context rather than depending on the paragraph above them. The substance of the content did not change. Its retrievability did.

Entity schema. A large-scale JSON-LD implementation established explicit entity relationships across every site segment and throughout the blog, connecting practice areas, case types, locations, and editorial content into a machine-readable graph rather than a set of unrelated pages.

Pagination remediation. Blog and news pagination had canonicalized every page back to the first, suppressing the bulk of the archive. Correcting the canonical treatment restored crawler and AI-fetcher access to years of existing legal content that had already been written and paid for.

Discovery infrastructure. The site implemented the IndexNow protocol so new and updated content is submitted for discovery on publication rather than waiting to be recrawled, and deployed an llms.txt file to make the site’s structure and key content explicit to AI systems that consume it.

THE RESULTS:

Visibility in AI-generated answers rose sharply, and it converted.

Between January and July 2026, the number of priority terms for which the organization appeared in an AI Overview grew from 111 to 182, a 64% increase. It gained that ground while the opportunity was shrinking: Google served AI Overviews on 653 of the tracked priority terms in January and only 467 in July, a 28% contraction in the available surface. The organization’s share of the AI Overviews actually served therefore rose from 17% to 39%.

Both measures moving together is what makes the result unambiguous. A rising share on its own could simply reflect a shrinking denominator. A rising count on its own could reflect Google serving more AI Overviews to everyone. Here the count rose 64% while the pool fell 28%, which leaves one explanation.

Average monthly citations in AI-generated answers rose from 7,600 to 12,900 over the same period, up 70%. The organization was not merely appearing more often — it was being drawn on more heavily within the answers where it appeared.

The commercial outcomes followed. Case sign-up form submissions, the organization’s primary intake path, grew 18% year over year. Online donations grew 37%, and organic-attributed revenue increased 145%, the gap between those two figures reflecting a rise in average donation size alongside the rise in donation volume.

That revenue figure outpacing every visibility measure is what happens when the gains land on the questions people ask at the point of need. Being cited in an answer about a legal process reaches someone who has that problem right now, and that audience converts at a rate general awareness traffic does not.