The “Good Enough” Mindset Is the Secret to Getting Real Work Done With AI
Eliott Ardisson
Founder & CEO - Basalt Studio
Why chasing AI perfection kills adoption — and how a "good enough" mindset unlocks real productivity gains for founder-led SMBs.
Key Takeaways
- Expecting AI to be perfect before using it is the most common reason adoption fails — teams test once, find flaws, and quit.
- The “good enough” mindset treats AI as a drafting engine: it handles the structural heavy lifting, you handle judgment and refinement.
- This approach works best on repetitive, structured tasks where iteration is already part of your normal workflow.
- Human review isn’t optional — it’s the mechanism that makes the whole system work and prevents quality drift over time.
- The goal isn’t to lower your standards. It’s to reach those standards faster by not starting from a blank page.
The Problem With Holding AI to a Standard You Don’t Hold Yourself
Most business deliverables don’t come out right on the first try. Proposals get revised. Marketing copy gets rewritten. Internal memos go through three versions before anyone sends them. That’s just how professional work functions.
So it’s worth asking: why do so many teams abandon AI tools the moment the first output isn’t perfect?
The answer is a kind of category error. People treat AI output like a finished product when it’s better understood as a first draft from a very fast, very well-read junior colleague. That junior colleague has read a lot, synthesizes quickly, and doesn’t have full context on your specific situation. Of course their first draft needs work. That’s not a failure. That’s the starting point.
The “good enough” mindset is a reframe, not a lowering of the bar. It means getting AI into your workflow as a drafting and structuring tool, then applying your judgment — domain knowledge, tone, client context, strategic nuance — to produce the finished output. The standard stays the same. The path to it changes.
Why Perfectionism Kills AI Adoption Before It Starts
Teams that demand flawless AI output from day one follow a predictable arc. They spend time setting up a tool. They run a few tests. The outputs are close but not right — maybe the tone is off, or a detail is missing, or the format doesn’t match their workflow. They conclude the tool “isn’t good enough” and move on.
What they missed is that the tool was probably 70–80% of the way there. The missing 20–30% was refinement work they would have done anyway in a normal drafting process. But because they weren’t expecting to do that refinement — because they wanted the AI to do it all — the output felt like a failure instead of a starting point.
McKinsey research on AI adoption in professional services has consistently pointed to expectation calibration as one of the strongest predictors of successful implementation. Teams that frame AI as an augmentation tool adopt it sustainably. Teams that frame it as a replacement either over-delegate and create quality problems, or get disappointed and disengage entirely.
The middle path — treat AI as a capable but imperfect collaborator — is where real productivity compounds.
What “Good Enough” Actually Means in Practice
This isn’t a philosophy. It’s a workflow structure. Here’s how it breaks down.
Tasks that suit the “good enough” approach:
- First-draft proposals and scope documents
- Email responses to common client inquiries
- Meeting notes and action item summaries
- Research briefs and competitive summaries
- Job descriptions, intake forms, FAQ drafts
- Content outlines and initial article structures
What these have in common: they follow a recognizable pattern, they tolerate iteration, and getting to 70–80% complete is the hard part. The refinement — adding specificity, adjusting tone, verifying facts — is fast once the structure exists.
Tasks that don’t suit this approach:
- Financial calculations and reporting that require precision
- Legal documents or compliance-sensitive language
- Final external communications with no review step
- Strategic decisions or crisis management responses
- Any context where an error damages a client relationship
For these tasks, AI can still play a supporting role — research, background prep, summarizing context — but the primary output stays in human hands.
The discipline is knowing which category you’re in before you start.
Structuring Prompts for Useful Drafts, Not Perfect Outputs
The way you prompt an AI system shapes what kind of output you get. If you write a vague prompt and expect a polished result, you’ll be disappointed. If you write a structured prompt and expect a useful draft, you’ll get exactly that.
A practical framework: Role + Context + Task + Constraints
Instead of: “Write an email about the project delay.”
Try: “You’re a project manager at a professional services firm. Our client’s website project will be delayed one week due to scope additions the client requested. Write a professional email explaining the delay, maintaining trust, and confirming revised timelines. Keep it under 200 words. Tone: direct and confident, not apologetic.”
The second prompt produces something you can actually use in five minutes. The first produces something generic that doesn’t fit your situation.
Adding what might be called “draft mode” language also helps calibrate expectations on both sides: “Create a first draft of…”, “Outline the structure for…”, “Generate three options for…”. These phrases signal that you’re looking for a starting point, not a finished deliverable. That framing tends to produce output that’s easier to refine because it’s less over-confident.
Building the Review Step Into the Workflow
The “good enough” mindset only works if the review step is non-negotiable. This is where many implementations break down — teams get comfortable with AI outputs and gradually stop scrutinizing them. Quality drifts without anyone noticing until a client flags something.
A practical review checkpoint takes roughly five to ten minutes and covers four areas:
Accuracy: Are any facts, figures, or specific details wrong or assumed? AI will sometimes invent plausible-sounding specifics. Verify anything that matters.
Voice and tone: Does this sound like your company? Is the formality level right for this recipient? Are there phrases that feel generic or slightly off-brand?
Strategic fit: Does this output actually serve what you’re trying to accomplish? AI doesn’t know your relationship history with this client, your competitive positioning, or your internal priorities. That’s your job to overlay.
Completeness: What’s missing? What question will the recipient have that this doesn’t answer? What context needs to be added?
Building this checklist into your process — literally as a checklist, not just as a mental note — prevents the drift. It also makes onboarding new team members to AI-assisted workflows much easier, because the review step is explicit rather than assumed.
Where This Plays Out in Real SMB Contexts
The “good enough” approach isn’t abstract. Here’s what it looks like in the kinds of businesses where it gets applied.
Recruitment and HR firms: A recruiter receives a job brief from a client. Instead of writing the job description from scratch, they prompt an AI system with the role requirements and get a structured draft in two minutes. They spend five minutes adjusting the language to match the client’s culture and their own agency’s tone, verify the requirements are complete, and post. The total time drops from forty minutes to seven. Quality is the same or better because the recruiter isn’t tired of looking at the document.
Real estate brokerages: An agent needs to send a follow-up sequence to leads that went cold. AI drafts three emails with different angles — urgency, value, social proof. The agent reviews, personalizes the subject lines and opening sentences, adjusts any pricing references, and approves. What would have taken an afternoon takes forty minutes.
Accounting practices: A manager needs to summarize a client’s quarterly performance for an internal briefing. They paste in the key figures, prompt an AI to structure a narrative summary, and get a coherent draft. They verify the numbers, adjust the interpretation in two places where context matters, and send. The grunt work of turning numbers into sentences is gone.
HVAC contractors: In our work helping trades businesses think through AI implementation, one of the most common workflows we encounter is after-hours inquiry handling. AI handles the initial intake — capturing the nature of the problem, the urgency level, and the contact details — and drafts a response with appropriate next steps. A human reviews the flagged urgent cases first thing in the morning. Nothing falls through the cracks, and the contractor doesn’t need to be available at midnight.
In each case, the human isn’t removed from the process. They’re moved to the part of the process where they add the most value.
Common Mistakes That Undermine the Approach
Setting the bar too low. “Good enough” doesn’t mean barely acceptable. If an AI draft needs to be completely rewritten rather than refined, the problem is either the task selection or the prompting — not the concept. Fix the input before you abandon the approach.
Skipping the review. This is the biggest risk. AI outputs can look confident even when they’re wrong. A missing review step doesn’t just create quality problems — it erodes trust in AI tools across the team when something eventually goes out that shouldn’t have.
Applying it to the wrong tasks. Some teams get enthusiastic and start using AI for things that require precision or genuine creativity from the start. The tool gets blamed when the real issue is task selection. Keep the high-judgment, high-stakes work in human hands.
Expecting instant adoption. Teams need a few weeks of consistent use before the workflow feels natural and the time savings compound. If you run a two-day pilot and don’t see dramatic results, that’s not a signal to stop — it’s a signal to keep going.
Measuring Whether It’s Actually Working
Forget trying to calculate a specific ROI figure in the first month. The more useful early indicators are:
- Time to first draft: Is it dropping? By how much?
- Revision cycle length: Are you spending less time in back-and-forth? Are fewer edits needed on final outputs?
- Volume capacity: Is your team able to take on more work without working longer hours?
- Team sentiment: Is repetitive, draining work decreasing? Are people spending more time on the work they’re actually good at?
Quality metrics matter too. Track client satisfaction, error rates in final deliverables, and revision requests from clients. If quality is holding steady or improving while volume is going up, the approach is working.
If quality is slipping, the review step is the first thing to examine. In most cases, the issue is that someone stopped reviewing carefully, not that AI is producing worse output.
The Mindset Shift That Makes the Difference
The teams that get lasting value from AI aren’t the ones with the most sophisticated tools. They’re the ones who stopped waiting for AI to be perfect and started treating it as a capable first-draft engine that frees them up to do the higher-value work.
That shift sounds simple. In practice, it requires overcoming a real psychological pull — the feeling that accepting imperfect AI output is somehow lowering your standards. It isn’t. Your standards apply to what goes out the door, not to the intermediate draft you’re refining. Every professional already works this way. AI just adds a faster, cheaper starting point to a process that already exists.
In our work helping founder-led SMBs think through where AI fits in their operations, the breakthrough moment usually isn’t a technology decision. It’s a team deciding that “useful now” is better than “perfect eventually.”
If you’re trying to figure out where the “good enough” mindset applies in your specific operation — which workflows to start with, what a realistic review process looks like, and how to build team buy-in — that’s a conversation worth having. You can book a free AI strategy call with Basalt Studio at cal.com/eliott-ardisson-kzq7zs/ai-strategy-call to work through the specifics for your business.