Blog / Post 01
Stock-photo content
How AI-assisted drafts changed what good content review looks like in 2026.
I read four B2B blog posts last week. By the time I closed the last tab, I could not have told you which company had written which one.
The pieces were not bad. Intros made sense, sections were sensible, takeaways were clean. There was nothing to circle in pen and send back. The four posts slid past me, one after another, in a steady wash of competent prose with no particular author behind any of it.
Most B2B content has started to feel like this in 2026, both to read and to ship. Output volume has gone up while recall has gone down. Fluency is real now, and the cost of it is forgettability.
What is happening here is not really about AI being bad at writing. AI writes well. The trouble starts when those drafts come back for review.
Most teams have noticed that content feels samey. Fewer have named what changed inside the workflow to cause it. Almost none have updated their review process to catch it.
The gap has a name. Stock-photo content. The visual world named this problem years ago, and content marketing is catching up. A stock-photo blog post is competent, professionally produced, and engineered to drop into any brand without friction, which is exactly what makes it forgettable the moment the reader closes the tab.
What is stock-photo content?
Stock-photo content is the AI-assisted draft that passes approval because the draft looks competent on the surface and stays indistinguishable in every way that matters to a reader. The piece reads like content without reading like anything in particular.
The structure is recognizable. The intro is clear. The headings make sense, the examples pass without comment, and the conclusion gestures toward some version of the future. The old review process was built to catch broken logic, messy prose, missing transitions, and obvious factual gaps. A stock-photo draft passes that review, and the pass itself is exactly what makes the draft dangerous.
A bad AI draft is easy to fix because the problems announce themselves. A stock-photo AI draft is more confusing, because the draft looks like it has already passed the review it still needs. Grammar and structure are intact. The thing missing is judgment: a proprietary example, a visible position, a sentence that could only have come from this company, this writer, this team, this accumulated body of experience.
In April 2026, Semrush analyzed 20,000 keywords and 42,000 blog posts. Content classified as human-written appeared in the top Google result 80.5% of the time. Purely AI-generated content appeared there 10% of the time. Detection methods have known limits, but the useful takeaway is simpler than the methodology: search rewards finished products that show human originality, and production processes that can be defended in a meeting do not move the needle on their own.
The question worth asking now is whether anyone on the team made enough decisions on top of the model's output for the piece to read like it came from somewhere specific. For most content teams, the answer is no. The writers have not stopped caring. The review process was designed for the old failure mode, bad drafts, and never rebuilt for the new one: drafts professional enough to ship and generic enough to disappear.
Quick diagnosis: if another company in your category could publish the piece tomorrow with only the logo changed, the draft is stock-photo content.
The two failure modes behind stock-photo content
The thing that catches stock-photo content is taste.
Taste gets treated like a little editorial perfume, something mysterious and subjective, and the word is more practical than that. In content work, taste is the ability to decide which sentence, example, claim, format, and compromise should ship under the company's name. AI raises the value of that ability and also reveals two opposite ways taste can fail.
Failure 1: accepting what the model offered
AI offers the safest version of every choice when asked to help with a piece. The model picks the structure most articles on the topic already use, the example that will offend no one and surprise no one, the softer wording of the sharp claim, the section weights that imply every idea deserves equal airtime, and the conclusion that gestures at the future the way every other piece does.
Each choice looks reasonable in isolation. The combined effect drags the piece toward the dead center of the category, where the draft becomes stock photography in prose form: competent, professional, and interchangeable.
Stock-photo drafts feel unfair to critique for this reason. The writer can point to any section and ask, reasonably, what is wrong with it. The honest answer is that the problem is structural, not local: the piece has no cumulative decisions to point to.
The draft is not wrong. It is unclaimed.
What to check for in the draft
Use this as the first pass before line editing:
| Signal | What it usually means | What to do |
|---|---|---|
| The examples could belong to any company in the category | The model reached for a generic pattern instead of lived knowledge | Replace at least one example with a real customer, sales, product, editorial, or operational moment |
| The strongest claim has been softened | The draft is protecting consensus instead of expressing judgment | Rewrite the claim the way the strategist would say it on a call |
| Every section is the same length and intensity | The piece has structure but no priority | Cut, expand, or reorder sections so the argument has weight |
| The conclusion summarizes instead of leaving a usable idea | The draft is closing politely instead of landing | End with a decision rule, diagnostic question, or memorable reframe |
The fastest first read is to ask, "Where is the first place this piece proves we have seen the work up close?" A draft that fails this test by the second section probably still belongs to the model.
Failure 2: using taste as obstruction
The second failure mode is less comfortable because the pattern looks like quality.
A strong writer with strong taste can spend three weeks on a piece that needed four days, refuse to write a comparison page because the format feels inelegant, rewrite a freelancer's perfectly usable draft because the draft does not sound like them, and turn every review into a referendum on voice when the reader only needed a better table. Each call feels defensible in the moment. The calls slow the team down in aggregate, without making the work meaningfully better.
Content people do not love to admit the next part. Sometimes the format is right even when the format feels boring. A comparison table can belong in the piece, the "best tools" article is rarely a moral failure, and the buyer is not shopping for literary elegance so much as trying to decide what to do before the next meeting.
Taste does its best work where taste protects specificity, usefulness, and point of view, and its worst where taste shields the writer's preference from the reader's actual task.
What to check for in the team
Stock-photo content shows up at the team level too, in the operating patterns around how drafts get reviewed and what gets put on the calendar.
| Signal | What it usually means | What to do |
|---|---|---|
| Time-to-publish keeps creeping up | Taste is being used to reopen settled decisions | Define which decisions are still open at each review stage |
| High-intent formats never make it into the calendar | The team is privileging editorial taste over buyer usefulness | Add comparison, alternative, roundup, and decision-support pieces intentionally |
| One strong writer keeps rewriting everyone else's work | Quality standards are not explicit enough to scale | Turn their instincts into a checklist, rubric, or annotated example |
| Debates keep circling preference instead of reader outcome | The team lacks a shared definition of "good" | Anchor review comments to the job of the asset |
The goal here is to stop spending high-taste energy in places where the spending does not change the reader's outcome.
Where taste belongs in the content workflow
The trap is treating taste as one setting on a dial. Turn the dial up and the work becomes precious. Turn the dial down and the work becomes stock photography. A more useful way to think about taste is as a placement decision: some parts of a content asset need strong editorial judgment, and other parts need the writer to get out of the reader's way.
Each kind of content has its own job. A founder POV piece should feel like a mind at work. A search article exists to answer the query. A comparison page works when the page helps someone actually choose. A case study earns its keep when the before-and-after is specific enough that the reader can picture the mess before the fix.
Use this table during planning, not after the draft is already written:
| Part of the content | A taste call that hurts | A taste call that helps |
|---|---|---|
| Topic | "Everyone is talking about this, so we should too." | "What can we say because we have seen this problem up close?" |
| Format | "This structure feels too obvious for us." | "What format helps the reader make progress fastest?" |
| Brief | "Let AI give us the angle." | "Feed AI our angle, constraints, examples, and audience context." |
| Draft review | "This reads well enough." | "Which parts could have come from any company?" |
| Examples | "Use a generic example so everyone relates." | "Use an example only someone in this role would know." |
| Final edit | "Make it smoother." | "What useful friction should stay?" |
| Publishing call | "I am not proud of this yet." | "Is it useful, accurate, and visibly ours?" |
The pattern across the rows is consistent. Taste matters most at the edges of the workflow: choosing the topic, defining the angle, deciding what makes the piece ownable, and protecting the final texture. Taste gets in the way in the middle, where format, search intent, and reader task already constrain the work.
A stock-photo piece has no fingerprints anywhere. A precious piece has fingerprints in the wrong places. A good AI-assisted piece carries human judgment where judgment creates value, and quiet competence where the reader just needs the information.
A review checklist for stock-photo content
Most content reviews still ask the old questions. Is the structure logical? Is the writing clear? Are the claims supported? Does it match the brief? Are there typos? Keep those questions. The old questions still matter, and on their own the old questions are no longer enough.
Add a second pass for sameness:
- Could another company publish this? If yes, add a specific example, named pattern, internal phrase, customer insight, or sharper claim.
- Where did the writer make a visible decision? If the answer is nowhere, the piece has not been edited enough.
- Which claim would the team say more strongly in a meeting? Put that version on the page, then make it defensible.
- Which section feels smooth but empty? Smoothness often hides weak thinking. Look for paragraphs that sound right until you ask what they actually changed for the reader.
- What did the draft avoid because the format felt inelegant? Add the table, list, recommendation, checklist, ranking, or comparison if the reader needs it.
- What should stay a little rough? Not sloppy, rough. A specific aside, a blunt sentence, a lived detail, a phrase that sounds like a person instead of a style guide.
The sameness pass should happen before the line edit. A polished draft makes people weirdly protective of it. The paragraph is fine, the paragraph is fine, the paragraph is probably fine. A stock-photo draft sneaks through at exactly that moment.
Use AI earlier, not just later
The default AI workflow is backwards. Teams ask AI to produce a draft and then ask a human to sand off the most obvious AI texture. The safest choices have already settled into the structure by then, because the model picked the frame, the examples, and what the argument would consider obvious before the writer arrived.
A better workflow puts AI before the draft, during the thinking, and after the draft as an auditor of the writer's choices. The owner of the piece stays human throughout.
Here is the five-part workflow I would use with a team this quarter.
1. Interview the writer before drafting
Ask AI to interview the writer before the outline exists.
Ask me seven questions that will help turn this topic into a specific, opinionated B2B article. Push for examples from work I have actually seen, claims I believe but might soften, and moments where my view differs from the standard advice.
The output you want is raw material, not polished prose. A phrase the writer would use without thinking about it. A story from a customer call. A frustration from reviewing drafts at 11 p.m. A distinction the model would never have invented. The raw material becomes the fingerprint.
2. Make the model find the tension
Drop in messy inputs: sales notes, SME interviews, Slack threads, customer objections, old briefs, call transcripts, competitor pages.
Where do these sources disagree? What tension is everyone circling but not naming? What would a generic article say, and what would our experience make us say instead?
Good B2B arguments often live in contradictions. Sales says buyers need comparison pages while brand says comparison pages feel cheap. Leadership wants thought leadership while search data says the audience wants templates. Writers want originality while customers want answers. Do not hide those tensions. Build the piece around one of them.
3. Argue against the premise
Make the strongest case against the article before drafting.
A piece that argues stock-photo content is the new content quality problem should have to defend itself against the counter-argument: that the real problem is weak briefs, bad incentives, lazy editing, or unrealistic publishing volume. Decide what survives. The step keeps the piece from becoming a slogan with sections attached, because the writer has to earn the claim.
4. Audit for sameness after the human draft exists
Use AI as a sameness auditor once the writer has drafted the piece.
Identify every paragraph that sounds like it could appear in a generic SaaS blog. Flag softened claims, generic examples, unsupported abstractions, and sections where the format avoids giving the reader a concrete tool.
Do not ask for a rewrite yet. Ask for flags. The writer should decide what to change. The distinction matters, because AI can spot patterns but cannot know which rough edge is worth preserving under the company's name.
5. Keep the final pass human
The final edit is mostly preservation. Preserve the useful rough sentence, the example that feels slightly too specific, the table the reader needs even when the writer finds it inelegant. Cut the elegant section that does not earn its place. Rewrite the conclusion until the conclusion gives the reader a decision they can carry into work tomorrow.
AI can help you move faster through the middle. The last call on what sounds true belongs with the human editor.
Workflow rule: use AI to widen the thinking, stress-test the premise, and audit for sameness. The decisions about what stays in the piece belong to the human editor.
How hiring changes in the stock-photo era
The easy question for a senior content hire is whether the candidate can write. The question still matters. Writing samples now tell you less than they used to. A polished sample might reflect the candidate's judgment, or a great editor, or a model with a prompt stack and three rounds of human cleanup. The finished artifact is useful information, and on its own the artifact is no longer enough.
A better signal is how the candidate reviews someone else's almost-good draft.
Try this in the final round:
- Give the candidate an AI-assisted article your team would be tempted to publish.
- Give them 15 minutes.
- Ask them to mark three places they would push back and one place they would defend the draft against your own preferences.
The strongest candidates will catch more than errors. The strongest will catch indistinguishable examples, claims that have been softened past usefulness, sections that avoid the format the reader needs, and paragraphs that sound fluent without changing the reader's understanding. The very strongest candidates will also know when to leave something alone.
The hiring signal in 2026 is whether this person can tell which parts of the work need judgment, which parts need structure, and which parts need to stop being touched.
Senior content roles are becoming more editorial and systems-oriented. The drafts will keep arriving cleaner. The job is to keep the drafts from staying generic.
The new review standard
Good taste in the age of AI is, in the end, one operational call: refusing to ship a piece that any company in the category could have published. The call gets made at the brief stage, in the examples, in draft review, in the final edit, and in the hiring decisions about who gets to make all those calls.
Most teams will postpone the redesign of their review process until the cost shows up in a quarterly business review. A diagnostic that brings forward the conversation: pull the last five AI-assisted pieces the team published, and on each one, mark the single sentence you would defend in a meeting against a critic from outside the company. The number you find is the conversation to take into the next planning cycle.
The redesign itself is bigger than one meeting, because the redesign changes who reviews drafts and what they read for. The work takes a quarter to do well, which is why the teams that begin now will still look like themselves when the rest of the category looks like stock photography.