Winning Google Discover and the Feed

Traffic to a page nobody searched for

Most of what I do assumes a query. Someone wants a thing, types words into a box, and the machinery of ranking decides who gets the click. Google Discover breaks that assumption. There is no query. The feed watches what a person reads, taps, and lingers on, then serves them content it predicts they will want, before they have thought to ask. That is the whole trick, and it is why Discover behaves nothing like the ten blue links you spend your week optimizing.

I have watched this surface send meaningful traffic to pages that rank for nothing. I have also watched it evaporate overnight for reasons no report will explain to you. If you treat Discover like search, you will lose. It is an interest engine, not an intent engine, and the work of earning a place in it is different work. Here is how I approach it after fifteen years of moving numbers in large programs.

Why the feed is a different game

Search rewards the best answer to a stated need. The feed rewards the most compelling thing to put in front of a person given everything the system already knows about them. Those are not the same objective, and confusing them is the first mistake teams make.

A few consequences fall out of that difference:

• You cannot target a keyword. There is no keyword. You are competing for a topic to be considered interesting enough, to the right audience, at the right moment.
• Freshness and timeliness matter more than they do in classic search. The feed leans toward what feels current and alive, which is a real tension for anyone whose content strategy is built to be evergreen.
• The click is a vote, and so is the non-click. Impressions without engagement teach the system your content did not deserve the slot. That feedback loop is faster and less forgiving than anything in organic search.
• You do not control when you appear. You can do everything right and still wait. Discover surfaces content on its own schedule, which makes it a supplement to a demand-capture strategy, never a replacement for one.

The reason this matters to a serious program is simple. Zero-click behavior is eating into the traffic you used to count on, and the feed is one of the few places left where interest-driven discovery still routes a real visit to your site. It is worth learning to earn.

What actually earns a place in Discover

Let me be direct about the thing everyone wants to do and should not. The lazy way to win the feed is clickbait: manufactured curiosity, a headline that overpromises, an image engineered to make a thumb stop. It works for exactly as long as it takes the system to notice people bounce. Then it stops working, and it takes your credibility with it.

The durable way is less exciting and far more effective. You earn the feed by being genuinely worth a person's attention on a topic they already care about, and by making that value legible to a machine deciding in a fraction of a second.

The visual is load bearing. Discover is a visual surface. A large, high quality, relevant image is not decoration, it is the single biggest lever you have on whether anyone engages. This is where the discipline of image SEO for a multimodal search world stops being a technical nicety and becomes a distribution strategy. Original, high resolution imagery that actually depicts the content, not a stock photo, not a logo, earns the slot. Thin or generic visuals get skipped.

The headline promises and delivers. You want a title that is specific and honest. Curiosity is fine. A gap between the promise and the payoff is fatal, because the feed measures whether people who tapped were satisfied. Write the headline you can actually cash.

Topical authority carries you. The feed is more likely to surface your content on subjects where you have demonstrated real depth. Building genuine coverage of a subject, the kind that comes from answering the questions behind the query rather than chasing one-off traffic, is what signals to the system that you belong in that interest category. Scattered, unfocused publishing does not accumulate this way.

The FEED framework

Here is the checklist I run content through before I expect anything from Discover. I call it FEED, and it is four gates, all of which have to pass.

• F, Fresh angle. Does this piece have a reason to exist right now? Not dated news, but a current, alive point of view on a subject people are engaged with. Timeless substance with a timely framing beats both stale evergreen and disposable hot takes.
• E, Engaging visual. Is there a large, original, high resolution image that genuinely represents the content and will make a thumb stop for the right reason? If the best image you have is a stock photo, you are not ready.
• E, Earned authority. Have you demonstrated real depth on this topic elsewhere on the site, so the system trusts you in this interest category? A single orphaned post rarely earns the feed. A body of work does.
• D, Delivered promise. Does the headline set an expectation the content fully satisfies, so that the people who tap are glad they did? The satisfied tap is the entire currency of this surface.

Run a piece through those four gates honestly. If it fails one, it will likely fail in the feed. Fix the gate before you publish, not after.

Measuring a surface that hides its logic

You will not get keyword data for Discover, because there are no keywords. What you get is impressions, clicks, and click through rate for the surface as a whole, reported separately from search. Watch that stream on its own terms. A spike is not a permanent win, and a drop is not always a penalty. The feed is volatile by design.

The number I care most about is click through rate against impressions, because it tells me whether the content I surfaced actually deserved its slot. A wall of impressions with a weak tap rate is the system telling you the headline or the image wrote a check the content did not cash. That is a content problem, not a distribution problem, and no amount of chasing volume fixes it.

One more measurement note. Discover traffic is interest-led, which means it tends to arrive cold, with no prior brand relationship. That makes it a genuine top-of-funnel asset, and it makes brand search the most undervalued metric you should be watching alongside it. If the feed is doing its job, some share of those cold visitors come back later by name. That downstream lift is the real return, and it will never show up in a click through rate.

The takeaway

Discover is not search, and the fastest way to fail at it is to treat it like search. It is an interest engine that rewards genuinely worthwhile content, made visible through a strong image and an honest headline, backed by real depth on the subject. Run the FEED gates, respect the visual, and never cash a headline you cannot deliver on. The same instincts that serve you as interfaces move toward voice and images, which I get into in optimizing beyond text, serve you here: earn the attention, do not trick it.

If you are trying to turn no-query, interest-driven surfaces into a real channel instead of a lucky accident, the channel is open by introduction. Bring me the content you think should be winning the feed and is not, and we will find out which gate it is failing.