When the digital camera arrived, plenty of accomplished photographers argued it would ruin the craft. Anybody could now shoot a thousand frames, keep one, and call themselves a photographer. They weren’t entirely wrong. A flood of bad photography did follow.
But nobody today blames the camera for it. A camera is equipment, and the person holding it decides whether the result is worth looking at. That distinction feels obvious now. It didn’t feel obvious in 1999.
The same argument is happening about AI in marketing, and the same mistake is being made.
The word slop is doing a lot of work
AI slop gets applied to the flood of generic content filling feeds, search results, and inboxes. The frustration behind the term is legitimate. A great deal of what’s being produced right now genuinely isn’t good.
The diagnosis is where it goes wrong. Most people conclude that AI produces bad work. The more accurate conclusion is that AI produces work at the level of the direction it was given, and most direction being given right now is thin.
HubSpot’s State of Marketing report for 2026 found that 86.4 percent of marketers are using AI. In the same study, 65 percent said consumers are getting better at spotting and ignoring AI content, and roughly half believed AI has made marketing content less effective overall. Read those together. Nearly universal adoption, and a majority who think the output is working against them.
That’s not a technology failure. That’s what happens when a profession adopts a tool faster than it develops a method for using it.
What actually produces bad AI output
Marketers who came up before desktop publishing learned a process that still applies. Pencil sketches, storyboard artists, copywriters who took a week to get a headline right, and a creative director who marked up every draft. At Springbolt, a model gets briefed the same way a junior designer was briefed thirty years ago.
That means a real brief. Tone, audience, what the piece has to accomplish, what it must not say, brand rules, and examples of work that hit the mark. Then somebody reads what comes back and marks it up. Then it goes back. Then it gets read again.
On a piece that matters, thirty rounds is normal. Sometimes more. That number surprises people, and it may be the most useful thing a marketing leader can hear. In 1998, going eight rounds with a junior writer was a normal Tuesday and nobody called that a broken process. Somehow when drafts started arriving in four seconds, the industry decided round one should be the finished product.
Most disappointing AI output traces back to a single round of direction. One prompt, one draft, one conclusion that the tool doesn’t work.
The seventy percent rule for AI content
Most AI-assisted projects land in roughly the same place. Strong structure, wrong details. The composition is right, the specifics are off. Plan for output that’s about seventy percent finished and budget the remaining thirty percent as human work.
That last thirty percent isn’t polish. It’s the part that requires knowing things the model has no access to. It doesn’t know a CEO killed a similar headline two years ago. It doesn’t know the sales team refuses to use the word partnership. It doesn’t know which competitor just filed suit, or why a campaign that worked in 2019 would fail at the same company today.
On design work, most generated output still gets finished in Photoshop. The model handles composition, color direction, and concept in minutes. A person fixes the type, the lighting, and whatever it decided to do with somebody’s hands. Twenty minutes of correction on top of a ten-minute generation still replaces most of a day. That’s a win, even if it doesn’t look like the clean automation story the industry was sold.
Why this matters for your marketing budget
Gartner’s 2026 CMO Spend Survey found that 70 percent of CMOs say their internal marketing processes aren’t mature enough to implement and scale AI effectively, and only 30 percent report mature AI readiness. In the same survey, 15.3 percent of marketing budget was already going to AI.
That gap is the whole story. The tools got bought. The operating model didn’t get rebuilt. When output disappoints, the instinct is to blame the tool rather than examine the process, because examining the process means admitting it was never designed for this.
A second finding is worth sitting with. Labor’s share of the marketing budget rose from 21.9 percent in 2025 to 24.5 percent in 2026. Everyone predicted AI would cut people costs. The opposite happened, and the reason is straightforward. AI collapsed the cost of execution. It did nothing to the cost of judgment, and judgment is the expensive part.
Two things to check this week
Ask the team a specific question. Not whether they’re using AI, because they’re. Ask how many rounds of direction they typically give before accepting an output. If the answer is one or two, that isn’t a quality problem. That’s a process problem, and process problems are fixable in a way technology limitations aren’t.
Then look at the brand guidelines. Most were written to settle arguments about logo placement. Almost none were written to be loaded into a system as working instructions. That document is now one of the more valuable technical assets a marketing organization owns, and if it’s vague, everything downstream of it will be vague too.
AI won’t solve a marketing problem on its own. It’s a tool, and like every tool before it, it rewards people who take the time to learn it and frustrates people who expect it to think for them.
The camera never made anybody a photographer. It just made the good ones faster.
Sources
HubSpot, State of Marketing 2026 (86.4% AI adoption; 65% on consumer detection; content effectiveness).
Gartner, 2026 CMO Spend Survey, fielded January to March 2026, n=401 (15.3% AI budget share; 70% process maturity gap; 30% AI readiness).
Gartner, June 2026 release (labor share of marketing budget, 21.9% to 24.5%).
About Springbolt Group
Springbolt Group is a full-service, AI-driven marketing, creative, and data agency. The team handles brand strategy, creative and storytelling, and data, including third-party data brokerage, for clients ranging from founder-led startups to enterprise brands. For CMOs and marketing directors trying to move faster without lowering the bar, that’s the problem Springbolt works on every day. More at springboltgroup.com.
About the Author
Raul Varela is Executive Creative Director at Springbolt Group. He entered the business in 1991 as a traditional art director, back when campaigns started with pencil sketches. Today he leads the creative work at Springbolt Group. Across three decades he’s helped launch hundreds of companies for clients, from small businesses to Fortune 100 brands, and founded several of his own.
He was a co-founder of PICKUP, the enterprise last-mile delivery platform, where he helped take the company through Series A and Series B rounds totaling roughly $40 million, scale to 90 markets covering half the country, and sign contracts with most of the major big-and-bulky retailers in the United States. The company was later acquired by Point Pickup.
He’s taught at the university level for over a decade across several North Texas institutions, and is currently an adjunct professor at East Texas A&M, where he teaches UI/UX design and entrepreneurship courses in the master’s program.