How to Optimize Content for Google Discover

Key takeaways

Use a landscape image at least 1200px wide with max-image-preview:large enabled, or Discover will refuse to show a large card at all.
Publish inside topics you've already built authority around — Discover rewards accounts that behave like known publishers in a niche, not one-off viral swings.
Freshness and update velocity beat perfect prose; a stale page with a recent factual update can re-enter the feed faster than a brand-new post with no track record.

What Google Discover actually is (and isn't)

Discover is the personalized feed inside the Google app and on Google.com's mobile homepage that shows content based on a user's search history, YouTube activity, and topics they follow — not a query they typed. Google's own documentation describes it as surfacing content "without a specific search query," which is the single fact that should reshape how you write for it. Google's Discover documentation confirms there's no keyword to target and no guaranteed inclusion mechanism — pages become eligible, they aren't ranked in the traditional sense.

This trips up a lot of SaaS founders who treat Discover optimization like keyword research. It isn't. You're optimizing signals that make Google's recommendation model confident your page is worth interrupting someone's scroll for. That's a trust problem more than a relevance problem.

The technical baseline: image, speed, markup

Three technical requirements do most of the work, and missing any one of them silently disqualifies you from the large-card format that gets the clicks.

Image size and licensing. Add the max-image-preview:large meta tag (or leave the robots directive unset, since large is now default in most themes) and use a source image at least 1200px wide. Anything smaller gets downgraded to a small thumbnail or dropped from consideration entirely.
Core Web Vitals. Discover is a mobile-only surface, so a page that loads slowly on a mid-range Android phone underperforms even with a great image. This is the same mobile performance work that matters for optimizing a SaaS homepage for AI search — Discover just makes the penalty for skipping it more visible because it kills the card format outright.
Structured data and canonical tags. Discover pulls from the canonical URL, so if you syndicate content across subdomains or use a CMS that generates duplicate canonicals, you're diluting whichever version Google decides is the "real" one.

None of this is exotic. What's non-obvious is how often it breaks silently — a CMS plugin update strips the max-image-preview tag, or a CDN serves a resized 600px image instead of the original, and traffic drops with zero warning in Search Console because Discover reporting lags by days.

Why E-E-A-T matters more here than in regular search

Google has been explicit that content in Discover needs to meet a higher bar for who's behind it, because there's no query to filter intent — the system is making a judgment call about what a specific person wants to see, unprompted. Google's Search Central guidance on content quality ties this directly to author expertise and site reputation, not just page-level optimization.

In practice this means:

Bylines with real credentials outperform anonymous "Team" posts in Discover eligibility, because the system has almost nothing else to judge trust by outside a query.
Sites that publish inconsistently — a burst of content, then silence for months — see Discover impressions collapse even if the old content is still technically indexed. The feed favors accounts behaving like active publishers.
First-person specificity (a founder describing what actually happened when they tried something) tends to get pulled into Discover more than aggregated "top 10" content, because it reads as a primary source rather than a repackage.

This is also why customer-facing proof works well in this feed. If you've built out customer review content for SEO, repurposing the strongest, most specific quotes into a Discover-eligible post — with a real name, a real photo, a real result — tends to outperform generic "why customers love us" copy, because Discover's model can't tell the difference between confident marketing copy and confident nonsense, but it can tell the difference between vague and specific.

Headlines: informative, not clickbait

Google explicitly penalizes exaggerated or sensational headlines in Discover — clickbait doesn't just underperform, it can get a whole domain throttled from the feed. The headline needs to accurately describe what's in the piece, front-load the specific claim, and skip the vague teaser format that works on social media but reads as manipulative to Google's classifiers.

Compare:

Weak: "This One Change Transformed Our Traffic"
Better: "We Cut Google Discover Impressions in Half by Removing One Meta Tag — Here's What Happened"

The second version is longer and less "punchy" by social-media standards, but it's verifiable, specific, and gives Google's system something concrete to match against the page content. Discover's classifiers appear to reward headline-body consistency heavily — if the headline promises a number or an outcome, the body needs to deliver it in the first few hundred words, not buried at the end.

Topical consistency beats one-off viral hits

Discover works off interest graphs built from a reader's behavior, which means it favors accounts that publish repeatedly within a recognizable topic cluster over accounts that occasionally produce a single high-performing piece. A SaaS blog that publishes one viral AI-industry hot take and then goes back to product changelogs won't build lasting Discover traffic from that hit — the algorithm has no consistent signal to attach the reader interest to.

This is the same mechanism that makes interlinking cornerstone content valuable beyond crawl efficiency: it signals to both users and Google's systems that a topic is a genuine, ongoing focus of the site rather than a one-time experiment. If you want sustained Discover traffic, publish inside 2-3 topic clusters consistently rather than chasing whatever trended last week.

Evergreen framing helps here too, oddly, even though Discover has a reputation for favoring news-adjacent content. A well-maintained evergreen guide that gets a genuine factual update — a new statistic, a changed process, a corrected step — can re-enter Discover as if it were new. This is different from writing purely for freshness; it's closer to what we cover in writing evergreen content for SaaS blogs, where the goal is durability with periodic, honest revision rather than constant republishing tricks.

Freshness signals that actually work (and the ones that don't)

Changing the published date without changing the content is a well-documented way to get temporarily suppressed from Discover — Google's systems compare content diffs, not just timestamps. What does work is a substantive update: a new section, a corrected fact, an added example, paired with an updated "last modified" date.

At Seolyn, when we've audited why older posts stopped showing up in Discover after a brief spike, the pattern is almost always the same: the post got picked up for a genuinely current angle, generated a burst of clicks, and then quietly aged out because nothing about it changed afterward while competing content kept updating. The fix isn't a content refresh in the vague sense — it's adding something a returning reader wouldn't have seen before. Our own process for this mirrors what we describe in updating old blog posts for AI search: treat the update log as evidence, not decoration.

What breaks when founders try to automate this

Indie hackers who try to mass-produce Discover-optimized content usually hit the same wall: templated images. If every post uses the same stock-photo style or the same auto-generated header graphic, Discover's image-diversity signals start treating the account as low-effort, and card impressions taper off even though nothing else changed. The fix is boring but effective — vary composition, use real screenshots or product photos where possible, and avoid text-heavy header images, since Discover downranks images that look like ads or have significant text overlay.

The second common failure is publishing AI-generated summaries of other people's news without adding a distinct angle. Discover's quality systems are explicitly built to detect and suppress this pattern, because it's indistinguishable from content farming at scale. If you're commenting on industry news, the piece needs a specific, attributable point of view — what you tried, what broke, what number changed — not a rewritten aggregation.

Frequently Asked Questions

Q: How long does it take for a new post to show up in Google Discover?

There's no fixed timeline — some pages appear within hours of publishing if they match an active interest spike, while others never appear at all despite ranking fine in regular search. Discover eligibility depends on image quality, E-E-A-T signals, and topical match to active reader interests, not crawl speed.

Q: Does Google Discover use different ranking factors than Google Search?

Yes. Search ranks pages against a specific query using relevance and authority signals; Discover has no query and instead matches page topics against a reader's inferred interests, weighting image quality, headline accuracy, and publisher consistency more heavily.

Q: Can I see Google Discover performance data for my site?

Yes, Google Search Console has a dedicated Discover report showing impressions and clicks, but only for sites that have received meaningful Discover traffic — it won't populate for sites with little or no Discover activity.

Q: Does republishing an old post with a new date help it appear in Discover again?

Not on its own, and it can backfire. Google's systems compare actual content changes, not just the timestamp, so a date change without a substantive update — a new fact, section, or corrected detail — is likely to be ignored or treated as a manipulation signal.

Q: Is Google Discover only available on mobile?

Discover primarily appears in the Google app and on the Google.com mobile homepage; it has limited presence on desktop, which is why mobile page speed and mobile-rendered images matter more here than for standard desktop search rankings.