Part 1 covered three ways to keep people on your site: interactive elements that give them something to do, internal links that guide the journey, and getting intrusive popups and ads out of the way. The grounding for all of it was NavBoost, Google’s system for re-ranking results based on how people actually behave, confirmed under oath at the DOJ antitrust trial and named in the 2024 API leak. The goal, then and now, is to be the last, longest click: the page where the search ends because the person got what they came for.
Every tactic in that note was really the same move: remove a specific reason for the visitor to leave. This one continues the list, with five more, focused on the content itself. Even a well-built, fast, uncluttered page loses the visitor if the writing doesn’t answer the question quickly and completely. Here are five ways to make the content the reason people stay.
(If you missed part 1 last week, be sure to check it out.)
Win the First Few Seconds
The fastest bounce is the one where the person can’t tell, in about two seconds, that your page answers their question. They landed, they scanned the top of the screen, they saw a paragraph of throat-clearing about your company’s founding vision, and they were gone before they reached the part that actually helps.
Lead with the answer. If someone searched “how long does a sitemap take to get crawled,” the first thing on the page should address that, not three sentences about the importance of technical SEO in today’s competitive landscape. State what the page delivers, or better, start delivering it. You can add context, nuance, and depth immediately after. But the person needs to know within a glance that they’re in the right place, or they go back to the results and try the next one.
This is the same moment where speed and layout stability matter, so handle them here too. A page that’s still assembling itself, where the text you started reading suddenly jumps down because an ad or an image loaded above it, is a manufactured reason to leave. It’s the same logic as the popups in part 1: you spent effort earning the click, then made the payoff annoying to reach. This is the user-facing half of Core Web Vitals, and there’s no need to relitigate the technical side here. The point is simply that a visitor who has to wait for the page to settle, or chase the content as it shifts around, is a visitor you’re inviting to give up. Don’t make them wait, don’t make them watch it jump, and once it’s there, make the relevance obvious instantly.
Make It Scannable
Winning the first few seconds gets them reading. Scannability keeps them from giving up halfway down.
People don’t read web pages start to finish. They scan for the piece that answers their specific question, and if they can’t find it fast, they leave, even when the answer is sitting right there in a dense block of text. Short paragraphs, clear subheads, and the occasional list are not stylistic preferences. They’re how you let someone locate their answer without having to read every word to find it.
This matters more the longer and more thorough your content is. A comprehensive page is a good thing, but comprehensiveness with no visual structure reads as a wall, and walls get abandoned. Subheads act as signposts. A person looking for one specific detail can jump to it, get their answer, and stay, instead of bouncing to a thinner page that happens to be easier to navigate. The thoroughness only pays off if people can find their way through it.
Answer the Follow-Up Questions
This is the most important tactic in part 2, because it attacks the return-to-search behavior at its root.
Almost no question is really a single question. Someone asking “do I need an XML sitemap” is going to want to know how to create one, whether their CMS already generates one, how to submit it, and whether it even matters for a small site. If your page answers the first question and stops, you’ve left them with four reasons to go back to Google. If those follow-ups get answered without a trip back to search, the search is over, and it ends on your site.
The key phrase there is “without a trip back to search,” not “on the same page.” Because the real skill is judging scope. Some follow-ups are small enough to answer in a paragraph right where they come up. “Does my CMS already generate one” is a sentence or two, so you handle it inline. But “how do I create an XML sitemap” is a full article’s worth of scope, something like “3 Ways to Build Your XML Sitemap” or a dedicated “How to Create an XML Sitemap” guide. Trying to cram that onto a page about whether you need one in the first place bloats the page and answers the question badly. The better move is a clear, well-placed link to the page that covers it properly.
That is where this connects back to part 1’s internal linking. The two tactics are the same instinct at different scales: answer the next question inline when it’s small, and link deliberately to the page that owns it when it’s big. Either way, the person gets their answer without returning to the results.
This also connects directly to query fan-out. When a searcher or an AI system works through a topic, it expands the original query into a cluster of related sub-queries, the follow-ups, the comparisons, the “what about” cases. A page (or a tightly linked set of your pages) that anticipates and answers that whole cluster is what satisfies the entire search, not just the opening line of it. That holds whether the fan-out is happening in a person’s head as they read, or inside an AI system deciding which source covers the topic completely enough to cite.
The practical move is to think past the literal query. Before you publish, ask what this person will wonder about next, then decide for each one: answer it here, or send them to the page that answers it well. What you never want is to leave the question unanswered so they go back to Google to ask it. Cover the cluster, inline or by link, and you’ve collapsed what could have been five searches into one satisfied visit that stayed with you.
Keep It Current, and Show It
A visibly stale page is a reason to leave. If someone lands on a guide with a 2019 date in the title, screenshots of an interface that no longer looks like that, or prices that are obviously out of date, they don’t trust it, and they go back to search for something newer. It doesn’t matter whether the underlying advice still holds. The staleness signals “this might be obsolete,” and that’s enough to send them looking.
So keep your important pages genuinely current, and show it. Update the content when things change, refresh screenshots and figures, and display a real last-updated date so the person knows they’re reading something maintained. That visible signal of currency removes a specific reason to bounce, because it answers the unspoken question of whether there’s something more recent worth finding.
To be clear about what this is not: it is not changing the date on a page you didn’t actually update in the hope of looking fresh to Google. That’s the freshness myth, and it does nothing but erode trust when a reader notices the “updated” date on content that plainly wasn’t. The move here is real maintenance, shown honestly. Do the update, then date it.
Recommend What’s Next, Deliberately or Not at All
Most sites handle “what to read next” badly, and the bad version actively works against everything else in this note.
The lazy version is an automated block: your latest posts, or a feed of loosely related articles the CMS picked by tag. That isn’t guidance, it’s clutter, and clutter is exactly what part 1’s discussion of the leaked clutterScore field warned about. A generic related-posts widget stapled to the bottom of every page adds visual noise, offers nothing genuinely relevant, and often sends people toward lower-intent pages or off the useful path entirely. It looks like retention and behaves like distraction.
Done deliberately, it’s the opposite. A hand-picked “what to read next,” chosen specifically for that one page, does the same job as a well-placed internal link: it answers the reader’s next question before they have to go find it. And it doesn’t have to be a single link. A strong version offers three or four options mapped to where the reader might be in their own journey on the topic. Someone finishing your guide on whether they need an XML sitemap might see the beginner primer, the how-to-build-one article, and the piece comparing methods, so the person who wants the basics, the person ready to act, and the person weighing options each find their own next step. That’s genuine journey guidance, not a “latest posts” feed.
So the rule is simple. If you’re going to add a “recommended next” section, curate it per page, deliberately, a short and strongly relevant list chosen for that specific page and the different points a reader could be starting from. If you’re not willing to do that, don’t add the widget at all. A lazy related-posts block does more harm than the empty space it would have filled.
The Through-Line
Across both notes, every tactic is the same move at a different point in the visit: remove a specific reason for the person to leave. Give them something to do, guide them with links, clear the friction, answer fast, make it scannable, cover the follow-ups, keep it current, point them deliberately to what’s next. Nine different tactics, one job. Be the last, longest click by making sure the person gets what they came for and never has a reason to go looking elsewhere.
Run down part 2 and that’s all each piece is doing. Answer the query in the first few seconds so they don’t bounce on arrival. Make it scannable so they can find their answer. Cover the follow-up questions, inline or by link, so they don’t return to search for the next piece. Keep it current so they trust it. Point them deliberately to what’s next so they keep moving with you instead of away.
None of this is gaming a signal. You can’t fake “the person got exactly what they needed and had no reason to look elsewhere,” and Google’s systems are built specifically to neutralize the click patterns of people who try. The only durable way to win the behavior is to earn it, which means the work that keeps visitors happy is the same work that compounds in your rankings, and now in whether AI systems cite you too. Do right by the person on the page, completely, and the signal takes care of itself.
