How Many of Your Route Pages Should Exist

With 514 kabupaten and kota spread across 34 provinces, the theoretical number of "jasa pengiriman [kota asal] ke [kota tujuan]" page combinations for an Indonesian logistics company runs into the hundreds of thousands. Almost none of them should be built. This is not a caution against programmatic SEO in logistics; it is the entire discipline. The question was never whether to generate route pages at scale. It is which routes actually carry enough real demand and real operational data to justify a page existing at all, and what happens to the rest of the site when that question gets ignored.

Google's own John Mueller has described programmatic SEO as "often a fancy banner for spam," which sounds harsh until you notice the qualifier. Scaled page generation is not inherently penalized. The problem, consistently, across every independent research pass on this topic, is thin templated pages that carry no information gain: the same paragraph with two city names swapped, published thousands of times, indexed regardless of whether anyone is actually searching for that specific pair.

The Uniqueness Test That Actually Matters

There is a simple diagnostic that cuts through most of the debate about whether a given route page is safe to publish: remove the city-pair variable from two different route pages on the site. If what remains is substantially identical, both pages are thin, and no amount of internal linking or schema markup will fix that. This is not a theoretical test. It is the exact mechanism Google's automated content-quality systems, including SpamBrain, are tuned to detect, and detection accuracy for template-generated pages with substituted variables has become materially more reliable in recent update cycles.

Practically, a route page that survives that test needs real, route-specific data: actual transit times pulled from operational systems, actual rate bands by weight bracket, port or hub dependencies specific to that origin-destination pair, and coverage confirmation at the sub-district level, not a province-level generality. The recommended floor across the underlying research sits at roughly 800 words of genuinely unique content per indexed route, with at least three distinct H2 sections carrying route-specific information, not boilerplate padding to hit a word count.

What Happens When the Thresholds Get Ignored

The cost of getting this wrong is not hypothetical, and it is not small. One documented case involved a directory-style site that generated roughly 45,000 near-identical "[service] in [city]" pages, essentially the same architecture pattern a logistics company would use for route pages if it published every theoretical city pair without a demand filter. The result was a doorway-pages manual action, a traffic collapse of approximately 96%, and an eight-month recovery process that required deleting 42,000 of the offending pages before rankings began to return, and even then only partially.

The mechanism behind that kind of collapse matters more than the specific number. Google's Helpful Content systems evaluate the proportion of genuinely helpful content across an entire domain, not page by page. A logistics site where a large share of published URLs are programmatic near-duplicates does not just fail to rank on those specific pages. It drags down ranking potential for the rest of the domain, including the B2B procurement content, the regulatory guides, and the genuinely useful pages that took real effort to build. Publishing at scale without a quality gate does not just waste the low-value pages. It taxes everything else on the site.

There is a useful counter-example worth holding in mind, even though it comes from outside logistics: a travel booking platform's model of generating a distinct page for every city-to-city route works, at genuine scale, precisely because each page pulls from a rich underlying dataset that makes every page meaningfully different. The lesson is not that scale is dangerous. It is that scale without a real data source behind it is dangerous, and a logistics company's Transportation Management System is exactly the kind of real data source that makes safe scale possible, if it is actually connected to the content pipeline rather than treated as a separate operational tool.

Indexation Discipline Is a Publishing Decision, Not a Technical Afterthought

The single most consistent recommendation across the underlying research is to apply noindex deliberately to any city-pair combination that falls below the demand threshold, rather than publishing everything and hoping search engines sort it out. This inverts how most content teams think about indexation: instead of indexing by default and pruning later, the safer posture is to treat indexation as something a page has to earn.

For large route-page sets, that means XML sitemaps prioritized toward genuinely high-value routes, noindex applied to filtered views and sort-order variants that add no unique value, and a crawl budget strategy that does not waste Googlebot's attention on tier-3 and tier-4 city pairs nobody is searching for. Below the 30 to 50 monthly search floor, a route page should either not exist at all, or should be built noindex, serving navigation and internal linking purposes rather than competing for a ranking it structurally cannot earn.

Differentiation Through Data That Actually Changes

A route page that is technically unique on launch day but never updated again eventually becomes stale in a way that matters both to search engines and to the buyers reading it. The stronger pattern is to pull live or regularly-updated rate and transit-time data from internal operational systems, so the page content genuinely evolves over time. This does double duty: it prevents the kind of staleness that erodes rankings, and it gives repeat visitors, particularly B2B buyers checking a route before finalizing a shipment, a reason to trust that the numbers on the page reflect current reality rather than a snapshot from whenever the page was first built.

Signal	Route Page That Survives an Audit	Route Page That Gets Suppressed
Content source	Pulled from live TMS data, updated on a schedule	Written once, city name swapped for every new page
Indexation basis	Verified search demand, roughly 30 to 50 searches/month minimum	Every theoretically possible city pair, regardless of demand
Word count	800+ words of route-specific operational detail	200 to 400 words of generic marketing language
Variable test	Fails to read identically when the city name is removed	Reads identically to every sibling page once the city name is removed
Rollout pattern	Staged, starting with the 50 to 100 highest-volume routes	Bulk-published, thousands of pages at once

The Same Discipline Protects AI Citation, Not Just Rankings

The same structured, data-rich route pages that survive a Google quality audit are also the pages most likely to be cited when a buyer asks an AI system about shipping options between two cities. A route page that opens with a direct, specific answer, such as stating the exact transit range and rate band for that lane within the first 100 words, gives an AI system something concrete to extract and repeat. A generic paragraph about being "trusted nationwide" gives it nothing. The discipline required to pass a Google Helpful Content review and the discipline required to become a citable source in an AI Overview are, in practice, close to the same discipline, which means getting programmatic route pages right pays off on both fronts at once, not one or the other.

Not Every Route Query Faces the Same AI Overview Risk

Google AI Overviews now appear on a meaningful share of informational searches, and the exposure is not uniform across the query types a logistics company actually targets. A simple informational rate query like "berapa ongkir Jakarta ke Bali" is easy for an AI system to summarize directly from structured data, which means it carries high AI Overview risk and a corresponding click-through decline. A navigational query like "cek resi JNE" carries low risk, because the intent is to reach a specific tool, not to read a synthesized answer. A B2B procurement query such as "3PL provider Indonesia untuk e-commerce" sits somewhere in between: high consideration, nuanced enough that AI systems tend to summarize less aggressively, and still winnable through strong entity signals and structured evidence.

This risk profile should directly shape which route and rate pages get the deepest investment. Pages exposed to high AI Overview risk need to compete differently: leading with a concise, direct statement of the rate range or transit time within the first 100 words, using subheadings phrased as natural-language questions, implementing FAQPage schema, and building consistent entity signals across the Google Business Profile, directory citations, and structured data. The goal on these pages shifts from winning a blue-link ranking to becoming the source an AI Overview actually cites, since a page that gets cited retains visibility even when the organic click gets absorbed by the summary above it.

Crawl Budget Is Not Infinite, Even for a Site With Real Data

Large programmatic route sets, once they clear the demand and uniqueness thresholds, still create a second-order problem: crawl budget. A site publishing a thousand or more genuinely qualified route pages needs Googlebot spending its attention on the pages that matter, not on filtered views, sort-order variants, or paginated directory listings that add no unique value of their own. XML sitemaps should be structured to prioritize the highest-value routes, and secondary, non-canonical variants of the same content should carry noindex, follow directives rather than being left to compete for crawl attention against pages that actually deserve to rank.

The pattern that causes the most damage here is not usually the route pages themselves, once the quality gate is in place. It is everything generated around them: pagination, filter combinations, and sort orders that multiply the URL count without multiplying genuine content. Left unmanaged, this dilutes crawl efficiency across the whole domain, including the small set of route pages that were built correctly and deserve to rank.

---

Frequently Asked Questions

We have thousands of route combinations. How much of this can be templated safely?

Safety is determined by data richness per page, not the total number of pages published. A route page can be templated safely when it draws on genuinely distinct operational data for each origin-destination pair: real rates, real transit times, and route-specific notes. The threshold that matters most is verified search demand, roughly 30 to 50 monthly searches, below which a route should not be indexed at all.

What is the fastest way to check if a route page is thin content?

Remove the city names from two different route pages on the site and compare what remains. If the pages read as functionally identical once the variable is stripped out, both are thin content, regardless of how many words each one contains.

Does programmatic SEO always trigger a Google penalty?

No. Scaled publishing is not inherently penalized. The risk is specifically thin, templated pages generated primarily to manipulate rankings rather than to help a searcher. A programmatic system built on genuinely unique, regularly updated operational data does not carry the same risk profile as one built on find-and-replace city names.

How many route pages should a logistics company launch with?

The staged approach recommended across the underlying research starts with the 50 to 100 highest-volume, highest-confidence routes, validates indexation and ranking performance, and only expands in reviewed batches after that first cohort proves out.

Should low-demand route pages be deleted or just noindexed?

Noindexing is usually the safer first step, since it preserves the page for internal navigation and site structure without asking it to compete for a ranking it cannot earn. Deletion becomes the right call only when a page is confirmed to add no value even as a navigational or internal-linking asset.

This is the second in a six-part series on SEO and GEO for logistics, freight, and supply chain companies in Indonesia. The first article in this series covers why B2B and consumer logistics audiences need separate content architecture entirely, a distinction that also determines which route pages are actually worth building. For the technical SEO framework behind safe content scaling, Tessar Napitupulu's book Found Before They Search covers programmatic SEO for the Indonesian market in more depth.

Arfadia's logistics SEO services include route-page architecture built around exactly these thresholds, from demand verification through to staged indexation.

Written by Tessar Napitupulu, Founder & CEO of PT Arfadia Digital Indonesia, GEO pioneer since 2023.

---