SaaS
A SaaS business runs three sites on one domain: a marketing site judged on conversion, a documentation estate judged on accuracy, and an application that must not be crawlable at all. Most SaaS search problems are boundary problems between them.
What makes engineering for saas different
These are the conditions that change the shape of the problem. They are why a generic approach underperforms in this sector.
Three estates with different owners and different success measures
Marketing, documentation and the application are usually built and maintained by different teams with different priorities. Search performance depends on all three behaving consistently, which means the work is frequently coordination rather than code.
Documentation is often the largest search surface
For developer-facing products, documentation attracts a substantial share of qualified organic traffic and is usually the least deliberately engineered part of the estate — frequently built on a separate platform, with its own URL conventions, from the marketing site.
Versioning multiplies everything
Current, previous and future versions of the same documentation, plus translated variants, create duplicate and near-duplicate URL spaces at scale. Left unresolved, a versioning scheme can consume more crawl attention than the entire marketing site.
The application must be reachable and must not be indexed
Authentication flows, application routes and staging environments sit behind or alongside indexable content, and the boundary is where indexation defects and, occasionally, genuine exposure occur.
Conversion happens after search, not during it
A visitor rarely converts on the page they arrived at. Attribution spans a documentation visit, a pricing page and a trial signup, so the technical work's value is measured in pipeline rather than in sessions.
What usually goes wrong
Documentation not ranking for what it answers
Content that solves the exact problem a developer searched for, written on a platform with weak URL, canonical or linking conventions, so it does not surface.
Duplicate content across versions and locales
Version and translation schemes producing near-identical page sets competing with one another, diluting rather than extending reach.
Application and staging surfaces reachable by crawlers
Indexation of routes that should not be public, arising from a missing directive in a framework's routing configuration rather than from anything deliberate.
Programmatic pages that help nothing
Templated pages generated at scale without distinct value, consuming crawl attention and diluting quality assessment across the estate.
How we work in this sector
The work is grouped by objective rather than by service, because in this sector the objectives interact.
Treat documentation as a first-class estate
- URL and canonical conventions aligned with the marketing site
- Documentation-specific discovery and internal linking
- Structured data reflecting what the content actually is
Contain the version and locale explosion
- Canonical strategy per version and locale
- Crawl directives on superseded versions
- Translation signals that resolve rather than compete
Draw the crawlable boundary deliberately
- Directives covering authenticated and application routes
- Staging protection verified at the crawler, not assumed
- Pre-launch check integrated into the release process
Measure against the funnel
- Search entry points tied to trial and demo starts
- Attribution spanning documentation and marketing
- Segment analysis by page class rather than sitewide totals
Frequently asked
Do you have a SaaS case study?
Our relevant work is in the underlying disciplines: platform engineering, search, performance and automation. Ask us which examples are closest to your product, architecture and growth stage.
Should our documentation live on a separate domain?
Usually not, and the decision is worth revisiting if it was made early. A separate domain starts with no accumulated authority, has to earn its own separately, and splits the internal linking that would otherwise strengthen both. There are cases where it is correct — different audiences, different products — but it is often a consequence of when the docs platform was chosen rather than a deliberate strategy.
How do we stop the app being indexed?
Deliberately, at the routing level, and verified by fetching as a crawler rather than by checking in a browser. Framework defaults rarely do this for you, and the failure is silent — nobody notices an application route being indexed until it appears in results. We treat it as a release-process check because the configuration is usually one commit away from regression.
Is programmatic page generation worth it?
Only where each page has something distinct to say. Generated pages that differ only by a substituted term generally do not rank, consume crawl attention, and can affect quality assessment across the rest of the estate. Where there is genuinely distinct content per page — real data, real comparisons — it works well, and we will say so when it applies.
Related capabilities and work
Tell us what your saas platform is doing
The symptoms are usually more informative than the diagnosis. We will tell you what we think is happening.