contentEffort Isn't About Time. It's About What You Left Behind

That changed with the update that created a clear “before” and “after” in SEO. In September 2023, Google rolled out its last standalone HCU update. Since then, the helpful content system has become part of Google's core ranking systems, and we've talked a lot about content quality ever since.

But what actually counts as good content? Bad content? That's where things get difficult.

If you've been on LinkedIn or YouTube lately, you've probably run into the hype around “content effort,” or contentEffort as it's called in Google's leaked API documentation. Read enough of those posts and you'd think Google just handed us a secret ranking factor that measures how hard you worked on an article.

I don't think that's the interesting part. In a lot of ways, contentEffort just sounds like something SEOs have argued about for years. Content quality. We've been told, endlessly, to write helpful, original content for users instead of search engines. The problem is that quality is a human judgment. I can look at two articles and tell you which one is better fairly easily. Doing that consistently across the entire web is a completely different problem. Maybe that's what makes contentEffort worth thinking about: one small piece of how Google might be trying to turn something subjective into something a machine can actually work with.

So what do we actually know about contentEffort?

Not much, honestly.

The leaked field is stored as a list(VersionedFloatSignal). Interesting name, interesting data type. But it doesn't tell us what Google means by “effort,” how it's calculated, or how much weight it carries, if any.

This is where a lot of the recent discussion has overreached a bit. People found a field name and built checklists out of it. Add original images. Add expert quotes. Cite more data. Spend more hours researching. Cut back on AI. Some of that might be decent advice on its own merits, but none of it comes from the leak itself. Past the field name, everything is interpretation. So let's interpret it anyway, as a thought experiment rather than a claim about how the algorithm works.

How could Google actually measure something like this?

Google already employs human quality raters who judge pages on things like original research, testing, and first-hand experience. It's not much of a leap to imagine those judgments being used to train or calibrate a model — feed it enough examples of high- and low-effort content, and let it pick up the patterns.

From there you could run the model against individual pages with something like: assess how much effort appears to be behind this, regardless of how well it's written. That distinction is the whole point. Take five existing articles, run them through an LLM, and you can end up with something that reads better than any of the originals. You still haven't added anything.

What might such a model look for? Specific detail instead of vague statement. Original data. First-hand observation. Real sourcing. On the other side: generic phrasing, repetition, a lot of words saying very little.

None of it proves how many hours someone put in. Google doesn't need to know I spent eight hours on this piece. It just needs to look at what those eight hours produced.

contentEffort isn't necessarily timeEffort

Someone with fifteen years in iGaming can write a genuinely good article on certain topics in 45 minutes. Someone with zero background might spend eight hours researching the same subject and still come up short. By the clock, the second person worked harder. By the result, the first article is probably better.

If contentEffort is measuring anything quality-related, it likely cares about what the finished piece demonstrates rather than the hours behind it. That gets even messier once AI enters the picture. Three weeks of original interviews and research, turned into a finished draft in five minutes with an LLM's help, isn't low effort by any reasonable definition. Six hours of manual writing that just repeats what the top ten results already say isn't high effort either, no matter how long it took.

Which raises the real question: is Google measuring effort, or predicting the traits of content that tends to result from effort? Not the same thing.

Maybe this is just what we've been calling content quality

Quality was never one thing. There's no HTML tag for it, no word count where a page crosses over into “good.” Google can't just write contentQuality = good into the ranking algorithm. If they want a machine to evaluate it, they need to break the idea down into pieces something can actually measure or predict, and contentEffort might be one of those pieces among several others we don't know about.

What's interesting here isn't “Google cares about quality” — everyone already knew that. It's the possibility of a small window into how they're trying to operationalize it. A versioned float score, rather than a simple flag, fits that idea reasonably well. Versioned suggests whatever produces the number gets revisited or retrained over time, though we're guessing at that too.

A practical thought while writing this

Something occurred to me while working on this piece. Instead of spending energy guessing whether Google runs something like this against my content, why not just run it myself first?

If the thought experiment above is even roughly in the right direction, an LLM is already decent at telling the difference between “rewrote a press release” and “added context, data, or a first-hand angle.” So before publishing something, try handing a model your own draft and asking: does this add anything a reader couldn't already find on the top-ranking pages for this topic? Where does it feel generic, repetitive, or like a reworded version of something that already exists?

That's a genuinely different use of AI than generating the article in the first place. You're using it to stress-test your own work, roughly the way Google might, before anyone else sees it. It won't catch everything and it's obviously not Google's actual system. But it turns a vague worry — is my content effort even high enough — into something you can check before a page goes live, rather than something you find out about from a ranking drop three months later.

How could it be used?

Even in the best case, contentEffort is one signal among many. A high score wouldn't guarantee rank one. A low score wouldn't guarantee page seven. Links still matter. Relevance still matters. Authority, internal linking, technical SEO, search intent — all of it still plays a role, the same way domain rating on its own has never fully explained a ranking.

What would content effort look like in iGaming?

You see this constantly on iGaming news sites. A supplier puts out a press release, and within a few hours fifteen or twenty sites have published “their” version of it. It takes minutes now to run a press release through an LLM and end up with technically unique text that still tells the reader nothing new.

The gap shows up when a piece adds something the others didn't bother with. What happened the last time this supplier launched something similar. How the numbers stack up against the rest of the market. Why any of it actually matters to an operator or a player. Readers pick up on that difference almost instantly, even without thinking about it consciously.

The same thing happens with slot reviews. A lot of affiliates now run provider game sheets through AI and get clean, competent reviews out the other end — RTP, volatility, paylines, bonus features, all laid out nicely. It can look genuinely good. And there might still be nothing in it that only comes from actually having played the game.

AI AI AI AI AI AI AI

Someone's going to read this and think: so is this the end of AI content? No. AI doesn't tell you anything about how much effort went into a piece. The research can be yours. The data, the testing, the opinions — all yours, with AI just handling the part where it becomes sentences. Writing every word by hand, meanwhile, is no guarantee you've beaten a well-assisted AI piece.

Google is probably headed toward caring less and less about how you physically produced something, so long as the result is genuinely useful and shows you know what you're talking about. If that's true, the real question isn't whether Google can detect AI. It might not need to, as long as it can judge what actually ended up on the page.

So what is contentEffort?

Honestly, we don't know. A field name, a data type, one detail about versioning. That's roughly where the facts run out and the guessing starts.

Maybe something close to what's described here is happening. Maybe it isn't. Either way, it beats turning contentEffort into another checklist to chase. What we're really left with is a small hint at how Google thinks about content on the inside, and maybe one more attempt at answering a question nobody's fully settled: what actually makes content good?