Discussion starter: a B2B and SaaS SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Indexing
- SaaS SEO
A SaaS product has 300 integrations. Index every integration page, or only those with search demand? What did coverage and traffic show?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
At 300 pages, I would not use keyword volume to decide what gets indexed. Crawl budget is irrelevant at this scale, and keyword tools are at their least reliable on long-tail “product plus tool” queries. The deciding factor is substance: index every integration page that says something specific and accurate about how the integration works, and fix, fold or hold back the pages that are just a logo and a sentence. Search demand should decide how much you invest in each page, not whether it exists in the index.
On what coverage and traffic show: no controlled public test comparing “index all” with “index only pages with demand” on a SaaS integration directory exists that I can point to. The pattern below is what the mechanics predict, and the rest of this answer explains how to measure it on your own site so the decision rests on your data rather than on anyone’s anecdote.
Why search demand is the wrong gate
Keyword tools model volume from panels and ad data. For queries like “[your product] [niche tool] integration” they often show zero or nothing at all, because individual volumes are tiny. Those queries are still valuable: the person searching is usually mid-evaluation or already a customer checking whether your product fits their stack. Low volume and high intent is exactly where volume estimates mislead most.
There is also a sequencing problem. The only reliable measure of demand for a specific integration page is Search Console impressions after the page is indexed. Noindexing pages because you cannot see demand makes sure you never will.
Integration pages also do jobs that have nothing to do with search volume:
- Sales teams send them to prospects who ask whether you support a particular tool.
- Procurement and IT reviewers use them to tick requirements off a checklist.
- Partners link to them from their own marketplaces and co-marketing.
- People ask AI assistants “does X integrate with Y”, and the answer has to come from somewhere.
That last point has a documented consequence for Google. Its AI features documentation states: “To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet”. A noindexed integration page cannot be the supporting source when AI Overviews or AI Mode answer that question. The answer may then come from a partner’s marketplace listing, a third-party review site or an outdated forum thread. Other assistants have their own crawlers (OpenAI, for example, documents OAI-SearchBot for ChatGPT search), so the exact rules differ by platform, but the principle holds: a page nobody can retrieve cannot correct a wrong answer.
Why crawl budget is not the constraint
Google’s crawl budget guide is written for sites with around a million or more unique pages changing weekly, or 10,000 or more changing daily. Three hundred integration pages, plus a directory and some category pages, is nowhere near that. Noindexing some of them will not help Google crawl the rest of the site faster.
The actual constraints are indexing selection and quality perception. Google does not index every URL it crawls. Near-identical template pages are among the most likely to be left out, and when a thin template makes up a large share of a site’s URLs, it can weigh on how Google assesses the site as a whole.
What makes an integration page worth indexing
The test is whether the page answers the questions a buyer or admin has before connecting the two tools. A page worth indexing usually covers most of these:
- What data syncs, in which direction, and how often (real time, scheduled or manual)
- Triggers and actions available, or the objects and fields mapped
- Who built and maintains it: native, partner-built, or via a connector platform such as Zapier or Make
- Which plans or tiers include it, and any extra cost on either side
- Setup steps and typical setup time, with a link to the help documentation
- Known limitations, such as unsupported objects, rate limits or one-way sync
- Specific use cases by role, with screenshots of the actual interface
A page with the partner’s logo, a one-line description and a “Connect” button fails that test whatever its search volume. It is also the version most likely to sit unindexed.
Hypothetical example:
A SaaS product lists 300 integrations: 40 native integrations with detailed documentation, 110 partner-built integrations with some technical detail, and 150 available only through a connector platform, each with a logo and one sentence. The 40 native pages are the obvious candidates for full investment. The 110 partner-built pages can reach “worth indexing” with a structured template drawing on data the partners already supply. The 150 connector-only pages are better served by category pages (“accounting integrations”, “calendar integrations”) that list them with a short, accurate section each, until any individual one earns its own page through customer demand or sales requests.
Where the real risk sits: pairs, duplicates and dead integrations
Three hundred single-integration pages are not the risky part. The risk starts when a template multiplies.
- Pair pages. Pages for “connect A to B using our product” grow combinatorially: 300 integrations produce 44,850 possible pairs. At that point Google’s scaled content abuse policy (“when many pages are generated for the primary purpose of manipulating search rankings and not helping users”) is the relevant frame. Publish a pair page only when the workflow is real and documented.
- Marketing page versus help article. “Slack integration” on the marketing site and “How to connect Slack” in the help centre often compete for the same queries. Split them by intent (evaluation versus setup), cross-link them, and do not let both carry the same content.
- Retired, beta and “coming soon” integrations. An indexed page for an integration that no longer works is a customer-support problem, and AI answers can repeat it as if it were current. Redirect retired integrations to the relevant category page, and keep beta or waitlist pages out of the index until they ship.
What coverage and traffic are likely to show, and how to measure it
The mechanics predict a long-tailed distribution. A small number of integrations with large platforms tend to attract most of the clicks. The long tail gets a handful each, but collectively they can matter, and they often convert well because the searcher has a specific need. Treat that as a hypothesis to check on your own data, not as a finding.
On coverage, a thin template usually shows up in the Page indexing report as a cluster of “Crawled - currently not indexed” and sometimes “Duplicate, Google chose different canonical than user”. The second label means Google considers two of your integration pages to be the same page, which is a direct signal that the template is not differentiating them.
- Isolate the template. Put integration pages in their own XML sitemap and filter the Page indexing report by it. Record indexed versus not indexed, and the reasons.
- Segment performance. Filter the Performance report to the integrations path. Split results by integration type (native, partner, connector-only) and by page completeness, not only by URL.
- Follow the conversions. In analytics, track trials and demo requests by landing page. In the CRM, tag deals where a specific integration was a stated requirement. That second source often shows value that organic traffic figures miss.
- Check discovery. If the directory relies on JavaScript filtering, make sure every integration page is reachable through plain HTML links from category pages, not only through a search box or filter state.
The decision matrix
Page substance Measured demand Action Evidence that would change it Substantive Clear impressions, clicks or sales requests Index and invest: use cases, video, customer examples Traffic arrives but never converts: reduce investment, keep indexed Substantive None measured yet Index and maintain accuracy; minimal further investment No impressions after six months and no sales use: fold into a category page Thin Clear demand Enrich first; highest priority on the list Enriched page still not indexed: check internal linking and duplication Thin None measured Fold into a category page, or noindex until enriched; keep in the in-product directory Sales or support keep sending prospects to it: enrich and index Retired or pre-launch Any Redirect retired pages to the category; keep pre-launch pages out of the index Integration relaunches A staged test that settles it on your own site
If the team disagrees, test rather than argue. Take the thin pages and split them into two groups matched by partner size and category. Enrich one group to the standard above and leave the other unchanged. Over 8 to 12 weeks, compare the indexed share, impressions, clicks, and trials or demos attributed to each group. Also filter Search Console’s Search generative AI performance report (impressions only, no clicks) to the same URLs to see whether enriched pages start appearing in AI features. Log partner launches, co-marketing and product releases during the window, because they move integration traffic independently of anything you change.
If enriched pages are indexed at a clearly higher rate and start earning impressions, the template was the constraint, and the rest of the thin pages should be enriched in order of sales and support demand. If both groups behave the same, look at internal linking, duplication with help centre pages, or site-level trust before writing more copy.
Do not promise a sitewide lift from noindexing thin pages. Google’s core updates documentation says that where a site has unhelpful content, “deleting the unhelpful content can help the good content on your site perform better”. That is most relevant when the integration template makes up a large share of a small site. On a site with thousands of other pages, removing 100 thin integration pages is unlikely to produce an effect you could measure.
The decision rule
Index on substance, invest on demand. Every integration page that accurately explains what the integration does, how to set it up and what it cannot do should be indexable, because it serves buyers, sales, partners and AI answers whatever a keyword tool says. Pages that cannot meet that bar yet belong in category pages until they can. Then let Search Console impressions, sales requests and conversions, not modelled volume, decide which pages deserve the next round of work.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.