BLUF: The TCREI framework is a 5-step AI prompting method developed by Google: Task, Context, References, Evaluate, Iterate. It is designed to help marketers and business professionals generate higher-quality, more accurate outputs from AI tools by providing structured input before each prompt. The acronym stands for “Thoughtfully Create Really Excellent Inputs.”
By Nicola Ziady | Published: May 2026 | Updated: May 2026
Most AI outputs underperform because the input is vague. The TCREI framework fixes that – in under 5 minutes – by forcing structured thinking before you type. It is model-agnostic: the approach works equally across ChatGPT, Gemini, Claude, and other large language models.
Why this matters: According to Jasper’s 2026 State of AI in Marketing report, 60% of marketing teams with structured AI workflows report returns of 2-3x or higher – yet adoption of structured prompting approaches remains low across the industry.
The TCREI Framework: All 5 Steps Explained
Step 1 Task
Define exactly what you want
Describe the output you need. Include a persona (what expertise should the AI draw from?) and a format preference (how should the output look?). Persona shifts tone and output quality – it is directional, not decorative. “Act as a marketing director with no patience for vague briefs” produces meaningfully different results than “write me a marketing plan.”
Step 2 Context
Give the AI enough to work with
Context is the difference between output that is technically correct and output that is actually useful. Provide information about your audience, constraints, situation, and the stakes involved. The more specific you are, the more the output reflects your real-world needs rather than a generic interpretation of your task.
Step 3 References
Show examples
If you have past campaigns, competitor copy, tone of voice guidelines, or writing you want to emulate – include it. References narrow the gap between what you imagined and what the AI generates. Think of it as briefing a new freelancer: you would show them examples, not just describe the work.
Step 4 Evaluate
Apply your own expertise before using the output
Before using any output, ask yourself: did the input I gave actually produce what I needed? Evaluation is where your expertise matters most. You know the brand, the audience, and the goal. The AI does not. Most mediocre AI-assisted work reaches publication because someone accepted an output that was “close enough.” Being critical at this stage is the highest-leverage thing you can do.
Step 5 Iterate
The best outputs come from a conversation
Prompting is not a one-shot process. If your first output is 70% of the way there, build on it. Tell the AI what worked, what did not, and what you need adjusted. A single follow-up instruction – “make it punchier, cut two words, remove the word ‘now'” – typically moves output quality from adequate to publication-ready.
TCREI Prompting: Weak vs. Strong Examples
Task – email subject line
Weak
“Write a subject line for a reactivation email.”
Task – email subject line
Strong
“You’re a copywriter who believes the best subject line is the one that doesn’t feel like one. Write five options for a reactivation email targeting B2B marketers who went cold three months ago. Under 50 characters. No exclamation marks. No emojis. No fake personalization.”
Context – gift recommendations
Weak
“Give me holiday present ideas under $50.”
Context – gift recommendations
Strong
“Give me five holiday present ideas under €50. The recipient is 35, loves horse-riding, traditional Irish music, and just got accepted to compete in the Dublin Horse Show.”
Why Your AI Output Quality Scales With Your Prompting Skill
A common misdiagnosis: a flat first output leads to the conclusion that “AI isn’t good enough for this task.” That diagnosis is expensive.
According to a March 2026 Harvard Business School study (Bojinov & McFowland, Working Knowledge), AI amplifies existing expertise but cannot bridge a domain knowledge gap. In other words, the better you already are at your craft, the more your prompting skill compounds.
The implication: AI is not the bottleneck. The structure of your input is. The TCREI framework does not add time to your process – it redirects the thinking you were already going to do from vague intent into structured input.
Frequently asked questions
No. The TCREI framework is a checklist, not a script. You do not need to write your prompts in Task – Context – References – Evaluate – Iterate order every time. What matters is that before you send a prompt, you have thought through all five elements. Sometimes context belongs first. Sometimes you lead with a reference. The framework is a mental model for completeness, not a template for word order.
TCREI stands for Task, Context, References, Evaluate, Iterate. The acronym was developed by Google and is sometimes expanded as “Thoughtfully Create Really Excellent Inputs.” It is a structured 5-step framework for writing better AI prompts.
The TCREI framework was developed by Google and is taught through Google Career Certificates as part of their Prompting Essentials curriculum. It is designed to be model-agnostic – applicable across any large language model including ChatGPT, Gemini, Claude, and others.
Yes. The TCREI framework is model-agnostic. The underlying principle – that structured input produces better output – applies regardless of which AI tool you use. Specific outputs will vary by model, but the prompting approach is universal.
Not for every prompt. For simple tasks – “summarize this in three bullet points” – you do not need all five steps. For any output you plan to publish, present, or act on, running through the full checklist takes about 30 seconds and consistently improves the result.
That is what the Iterate step is for. A poor second output usually means your Context was too vague or your Task was not specific enough. Tighten those two elements first. If output quality is still low, add a Reference – even a rough example of what good looks like typically unlocks a significantly better result.
Approximately 2 minutes once the framework is familiar. The TCREI structure is not about writing more – it is about thinking more clearly before you type. Most of the context required is already in your head. The framework stops you from leaving it there.