When our blog workforce made the switch to an AI-first writing kind in June 2025, my reaction was once as soon as minimize up. There was once as soon as excitement on one aspect: I would possibly see the inefficiencies in our process that AI would possibly actually restore. On the other, anxiousness. And while numerous that anxiousness was once as soon as regarding the technology itself (how correctly the ones equipment would possibly actually write, how so much shall we believe them), my primary concern was once as soon as my workforce.

I’ve managed HubSpot’s blog writing workforce for three years, and previous to that was once as soon as a HubSpot Blog author myself. I understand what it option to have ownership over your process. Many of the writers I organize got into this artwork because of they love writing. They take pleasure in the craft of finding the right kind sentence and earning a reader’s attention. Web site guests numbers and conversions are secondary. Everywhere this AI transition, I felt the weight of what it might indicate for those whose entire approach to their craft was once as soon as changing speedy.

Such a large amount of most people conversation about AI content material subject matter gets the story incorrect. There’s an assumption that because of AI can write passably correctly, the arduous phase is over. Merely stick with the identical low-effort polish you’d stick with to an AI-drafted email correspondence or memo to your blog publish, then ship it.

Then again I will assist you to know from revel in: it doesn’t artwork that means. A blog publish is meant to compete against dozens (or quite a bit) of different posts on the similar matter. In this sport, generic, low-effort writing isn’t going to win, and that’s good knowledge on your content material subject matter workforce. It signifies that even though the writing process has changed, the craft remains very important.

So, how do you stability craftsmanship with the efficiency sure facets AI promises? And the best way do you assemble a process that takes advantage of what AI is actually good at without falling into its traps? We spent the general 365 days figuring that out. Proper right here, I’ll proportion what we discovered: where time was once as soon as saved, where it got added, and what it takes to make AI artwork without compromising the usual necessities you’ve spent years building.

Table of Contents

• Ahead of AI, writing was once private.
• What We Realized From Letting Writers Lead
• Evaluating Pre-AI vs. Put up-AI Writing
• AI content material at scale is simple. AI content material price studying is tricky.
• What This Manner for Folks Managers
• My Takeaways as a Content material Staff Supervisor

Forward of AI, writing was once as soon as non-public.

Forward of AI, our writing process was once as soon as, correctly, a writing process. While we had editorial necessities and guardrails to verify consistent top of the range, we didn’t put into effect a specific method for the best way posts should be written. We hired writers and editors for their ability to craft, and we depended on them to use that ability. We understood that writing is bigger than just a technical skill. Giving our writers the distance to amplify their own explicit particular person workflows yielded content material subject matter with character and unique insights.

When training new writers, we taught a simple, flexible framework to be in agreement people start to development their time:

• Outlining and planning took spherical 30 minutes to an hour.
• Research would possibly take one to two hours, depending on the complexity of the topic and the author’s familiarity with it. This included sourcing quotes, finding distinctive wisdom, and writing from non-public revel in.
• Drafting, the phase where anyone actually puts words to the internet web page, took spherical 5 to seven hours on affordable for a number one draft.
• Improving, QA, formatting, and finalizing in our CMS added every other roughly two hours on easiest of that.

For the usual blog publish, you’re merely looking at a whole day’s artwork, most often longer. Nearly all of that time lived inside the drafting phase — writing the words, finding the right kind development, and figuring out how to explain something clearly was once as soon as the core of the duty.

What We Learned From Letting Writers Lead

Once we formally adopted AI-assisted writing in June 2025, my manner was once as soon as deliberate: don’t over-prescribe. Allowing our writers to amplify their own processes had served us correctly for over a decade, and we’d have preferred to care for that sense of keep an eye on for our writing workforce as much as imaginable, similtaneously we added AI to the mix.

We gave writers get right of entry to to AI gear (in particular Claude and ChatGPT, together with NotebookLM) and house to resolve how you can artwork them into their process. Even with AI, I assumed writing is non-public, and I wasn’t going to tell a workforce of professional writers exactly how you can do their jobs. Our goal was once as soon as experimentation first, development later.

What followed was once as soon as no longer the clean efficiency win I expected. AI solved some problems and presented new ones. Then again by way of working through those problems with our writers, we built a process that everyone knows works.

Faster drafting doesn’t indicate sooner content material subject matter.

Initially, the indications were good. Writers were putting words on the internet web page sooner than previous to, and I thought that may translate immediately into sooner posts general.

It didn’t.

A couple of months in, after reviewing time tracking wisdom and having conversations with my workforce, a further tough symbol emerged. The quantity of posts stayed concerning the identical, on the other hand posts were getting longer, further detailed, and additional no longer simple to offer. Writers were spending time beyond regulation checking out tool, sourcing professional quotes, reviewing equipment in-depth, and going deeper on topics. The time AI saved in drafting was once as soon as being absorbed by way of the entire thing else a competitive publish requires.

What made it tougher was once as soon as that some writers felt the pressure of each and every expectations immediately: write sooner and write further. When I understood that this was once as soon as happening, I had to recalibrate. AI hadn’t diminished the workload. In many ways, it had upper the workload.

So my first lesson was once as soon as: When you’re measuring AI’s affect on your workforce purely by way of publish rely, you’re going to be dissatisfied. The win is in what those posts turn into, no longer how fast they get there.

Research was once as soon as our best marvel.

I moreover assumed AI would make read about sooner. In the end, instead of spending an hour getting up to the mark on a topic, a author would possibly ask Claude and switch on.

For general matter knowledge, that’s actually true. On well-documented subjects with numerous training wisdom to draw from, AI can explain concepts clearly, answer follow-up questions, and get a author up to the mark sooner than navigating a dozen browser tabs.

The shortcomings, however, were flagged by way of writers just about right away.

The main time it in truth surfaced was once as soon as on a listicle protecting competitor products. A author were given right here once more to me flagging that the pricing knowledge AI had surfaced was once as soon as incorrect. Not rather off, on the other hand completely old-fashioned. AI didn’t know the pricing had changed, and it didn’t flag its private uncertainty.

I maximum unquestionably should have spotted that one coming. In the end, a tool professional on wisdom from months prior to now obviously can’t know {{that a}} competitor changed their pricing. But it surely took a author worrying enough to flag it previous to I accounted for it.

From this, we discovered that for the remainder requiring accuracy about competitor pricing or choices, AI read about falls apart. That is good whilst you believe how the ones equipment artwork: An LLM’s training wisdom has a cutoff date, that implies it’s no longer going to reflect the most recent pricing changes, function updates, or product rebranding. And when AI equipment do take a look at a live web search, they’re best as good since the pages they can parse. Product pages are incessantly structured in ways that are tricky for an LLM to interpret, which makes even real-time lookups susceptible to inaccuracies. The outcome’s knowledge that sounds authoritative on the other hand may be months or years out of date.

We produce a lot of competitor listicles and product-focused content material subject matter, and for those, read about nevertheless requires necessary manual effort. So while AI has accelerated general-topic read about, competitor- and product-focused read about takes roughly an identical quantity of time as it at all times has.

The Hallucination Problem

The read about problem ran so much deeper than product pricing alone. Previous old-fashioned competitor wisdom, we’ve got been seeing fabricated statistics, misattributed quotes, and mistaken terminology, all delivered with the identical confident, authoritative tone. AI doesn’t know what it doesn’t know.

We place a best best elegance on accuracy, and the ones so-called AI hallucinations was fact-checking from a gradual final step into some of the a very powerful time-consuming parts of the process. In some cases, AI drafting took longer than manual drafting because of writers had to spend time verifying what the AI got incorrect.

What struck me was once as soon as how quickly my workforce caught it. That determination to publishing best honest, verified knowledge is strictly what separates a good author from a content material subject matter generator. It reinforced something I already believed: the author problems more than the equipment they use.

To take care of hallucinations, we presented a brand spanking new formal step in our process: faithful fact-checking. Pre-AI, writers were in control of the accuracy of their own read about, and a final editorial transfer would catch the remainder glaring. With AI inside the mix, fact-checking turned into its private distinct phase. For a traditional publish, I estimate it supplies spherical 30 to 60 minutes to the process. It’s longer for research-heavy pieces or the remainder involving stats, quotes, or product knowledge.

Sure, you should skip fact-checking and think AI gets the entire thing correct. Then again is that in truth the standard you want to hold your content material subject matter to? Publishing fallacious knowledge, even unintentionally, damages believe somehow that’s arduous to get better from.

Prime quality doesn’t police itself.

QA time grew too, for reasons previous what fact-checking was once as soon as already catching. Fact-checking is the process of verifying that the tips in a publish is proper. QA is ready the entire thing else: Does this publish sound like a person wrote it? Does it meet our editorial necessities? Is it structured somehow that search engines like google and yahoo and AI equipment can parse?

And no longer the use of a structured top of the range layer, AI-assisted drafts presented problems writers wouldn’t necessarily catch on their own, paying homage to:

• AI clichés and hedge-word-heavy phrasing (“it’s worth noting,” “leverage,” “it’s very important to grasp,” and so forth)
• Redundant sentences.
• Plagiarism from competing content material subject matter (no longer intentional, on the other hand from models quoting their provide material immediately inside the output)
• Broken or misguided links
• Missing image alt text
• Mistaken heading orders
• Sentences, paragraphs, and sections that aren’t optimized for LLM chunkability.

To take care of this, I built a custom designed QA checker immediately into Google Docs that flags the ones problems and additional. It’s turn into a required step in our workflow.

What AI Didn’t Touch

By way of all the changes above, some problems didn’t switch. The E-E-A-T necessities we had previous to AI — distinctive quotes, firsthand perspectives, forged wisdom sourcing — stayed intact. Editorial judgment stayed intact. The strategic considering this is going into deciding what a publish needs to say, and who it needs to say it to, stayed intact.

AI handles numerous the execution now: drafting, scoping, generating matter angles, and explaining concepts. Then again the layer of work that makes a piece worth learning (the human standpoint, the original belief, the rigorous sourcing) is still on the author. If the remainder, it problems further now because it’s what separates your content material subject matter from everyone else’s AI output.

Comparing Pre-AI vs. Submit-AI Writing

To grasp what changed, it’s serving to to seem it. Proper right here’s a rough breakdown of the way time was once as soon as distributed during a typical blog publish previous to AI, and the best way that distribution shifted after we adopted it:

Understand: Time estimates are averages in keeping with HubSpot Blog Group of workers inside tracking. Explicit particular person posts vary in keeping with matter complexity, length, and content material subject matter kind.

If truth be told, the reality varies more than the ones charts recommend. A author would possibly draft for an hour, perceive they would like further read about, go back to drafting, catch something that desires fact-checking, and return to drafting over again. A real workflow turns out further like this:

However, the time monetary financial savings are glaring.

AI content material subject matter at scale is discreet. AI content material subject matter worth learning is hard.

If AI can generate a blog publish in minutes, why no longer merely publish and switch on? Additional content material subject matter method further probabilities to rank, further ground area for leads, and additional output for a lot much less effort, correct?

Early inside the AI content material subject matter enlargement, all folks noticed a definite method emerging inside the SEO house: Submit as so much AI-generated content material subject matter as imaginable, as speedy as imaginable, and snatch market proportion previous to pageant caught on. Flood the zone. It operates on the accept as true with that amount alone may also be enough to win.

Except for, it isn’t, and the consequences caught up.

Slop at Scale

The content material subject matter that comes from this kind of fast-and-loose method is recognizable to any person who’s hung out in this industry.

A blog publish that strings together sentences without announcing so much. A listicle that may were written about any company in any industry. No distinctive wisdom. No authentic standpoint. No specific examples that couldn’t have been fabricated. And whilst you run it through a plagiarism checker, you’ll even to seek out content material subject matter that was once as soon as lifted from competing sources (no longer intentionally, on the other hand as a byproduct of the way LLMs artwork).

In brief, it’s generic. You’ve maximum unquestionably heard the period of time “AI slop” to give an explanation for this kind of content material subject matter as correctly. Professional readers and, further importantly, search algorithms can spot slop from a mile away, and it hurts far more than it’s serving to.

Marketers are catching on. In keeping with HubSpot’s 2026 State of Advertising File, 52% of marketers believe AI has made content material subject matter in reality simple to create that it’s turn into a lot much less environment friendly general.

Need further proof? In 2026, search engine marketing skilled Lily Ray tracked greater than 220 web pages that identified themselves as consumers of AI content material subject matter introduction platforms and situated a relentless building: instant expansion in herbal website guests, followed by way of a steep decline that, in a variety of cases, erased the sure facets utterly. All through her dataset, more than a part of the internet sites out of place no less than 30% of their best herbal website guests, and with reference to 1 / 4 out of place 75% or further.

Proper right here’s an example of what that looks like for one such website.

Supply

I understand why teams took to this implies to start with. The pressure to offer is authentic: 83% of entrepreneurs say they’re anticipated to create extra content material than ever on account of AI.

Then again this is exactly what major best with tempo turns out like in stick with. If top of the range becomes something you try to retrofit after the truth, it rarely works.

Get began with necessities, then add AI.

The teams doing AI writing correctly get began from the opposite end. They define what “good” turns out like — voice, depth, accuracy, originality — and then artwork backward to unravel where AI can be in agreement them get there sooner without compromising at the ones necessities. All writing is anchored in top of the range from the beginning.

On the HubSpot Blog, that suggests keeping our E-E-A-T must haves intact. We assemble top of the range guardrails into our workflow and are honest about which content material subject matter sorts have the benefit of AI involvement and which don’t. For example, a how-to knowledge and a piece of distinctive read about can have different must haves, and treating them the identical means will result in generic content material subject matter this is going ignored.

As former Director of the HubSpot Blog Karla Hesterberg puts it: “The only in truth simple process produces slop.” Teams that resist the slop, keep good writers inside of the driving force’s seat, and use AI as a lever reasonably than a replace are the ones producing artwork that converts.

For a further thorough dive into how we care for top of the range necessities in an AI-assisted editorial process, check out this publish from HubSpot Blog author Amy Rigby on how HubSpot creates high quality content material within the age of AI.

What This Approach for Other people Managers

Just about in one day, my place shifted in two directions immediately. On one aspect, I turned into further process-oriented than I’d ever been. I had to think further deeply about workflows, tooling, and the best way posts get assembled. On the other, folks keep an eye on got tougher. Not because of my writers were tricky, on the other hand because of I was asking them to switch how they artwork.

Process changes are reasonably easy to report, on the other hand maintaining motivation and morale through a transition that necessarily changed their jobs is a distinct problem utterly. You’ll assemble necessarily probably the most thoughtful AI writing process in the world, but if your writers aren’t bought in, none of it in reality works.

Writers come to this activity for quite a lot of reasons, and AI lands in a different way depending on which the explanation why that is.

Some writers are wired for effects. They get into content material subject matter promoting because of they’re good at writing they generally like seeing the business affect. For them, AI is a welcome exchange as it’s going to get them to the outcome further quickly.

Other writers were given right here to this artwork because of they love the craft. The writing process is something they care about deeply. For them, an AI-first kind can in reality really feel find it irresistible’s taking something away. I don’t see this as a potency issue or a failure to adapt. It’s an inside pressure that should be taken significantly.

As a manager, my activity is to recognize where every workforce member falls on that spectrum and artwork with them accordingly. I create a space for honest feedback, along with destructive feedback about AI. I know that if writers don’t in reality really feel comfy telling me what isn’t working, I’ll assemble processes on dangerous knowledge. The writers who most actually pushed once more on our early AI workflows gave me the clearest roadmap for what sought after fixing.

Steps for Managing Your Group of workers All through the AI Transition

Proper right here’s what I’d tell any manager navigating AI writing:

• Don’t over-prescribe early. Give writers get right of entry to to equipment and room to experiment previous to locking in rigid workflows.
• Mandate experimentation, no longer method. The teams that prescribed specific AI usage from day one created unnecessary friction.
• Recognize that you simply’re asking writers to switch how they artwork. That takes time and believe.
• Create house for honest feedback, along with positive feedback. If writers don’t in reality really feel protected telling you what isn’t working, you’ll assemble processes on dangerous knowledge.
• Understand where every author falls on the spectrum between outcome-driven and craft-driven, and organize them accordingly.

The target hasn’t changed: a workforce that produces artwork worth learning. The path to get there merely turns out different from what it did in 2024.

My Takeaways as a Content material subject matter Group of workers Manager

A 365 days into this transition, I’m proud of where we landed, even though we’re nevertheless figuring some problems out.

What made the difference wasn’t best the equipment, on the other hand the writers. I knew that shifting into, and this earlier 365 days confirmed it at each and every turn. The writers who pushed once more actually gave me the roadmap, and people who struggled helped me understand what was once as soon as actually being asked of them.

What sticks with me most from this revel in is something writers have started announcing no longer too way back: that they’re feeling like authentic writers over again. The stableness between AI efficiency and exact craft has started to in reality really feel sustainable. That happened no longer because of we found out a very good suggested or the right kind tool, on the other hand because of we on no account let transfer of what made our content material subject matter good inside the first place.

When you’re an article lead navigating the AI shift nowadays, know this: What is going to get you through is having writers who care enough to be honest with you, and being a manager who creates house for that honesty. Don’t underestimate what it method for a author to in reality really feel like a author. It displays up inside the artwork.

AI is a tool. Keep your writers at the center, grab the usual line, and the rest follows.

WordPress SEO

[ continue ]

WordPress Maintenance Plans | WordPress Hosting

read more