Swap Generic Claims for Specific Details
If you only learn one thing from this course, make it this: generic claims are the biggest giveaway in AI writing, and specific details are the biggest fix. A model can produce endless sentences about "many businesses" and "significant improvements". It can't know that your Leeds store had three shouting customers in one week.
Find the generic claims
Read your draft and circle every word or phrase on this list. Each one is a placeholder where a real detail should be:
- Vague amounts: many, various, numerous, several, a number of, significant, substantial
- Unnamed sources: studies show, experts say, research suggests, according to industry reports
- Unnamed people: users, stakeholders, customers, team members, professionals
- Unnamed time: recently, in recent years, nowadays, over time
- Empty results: improved efficiency, better outcomes, positive impact
A typical AI paragraph has five or more of these. A typical paragraph from someone who knows their subject has one or two.
The specificity ladder
For each circled phrase, climb as many rungs as you honestly can:
| Rung | Example |
|---|---|
| 1. Vague | Customers responded positively. |
| 2. Category | Repeat customers responded positively. |
| 3. Number | About 40 repeat customers emailed to say thanks. |
| 4. Named | About 40 repeat customers emailed, including the owner of the café next door. |
| 5. Story | The café owner next door, who'd complained for two years, emailed to say it was the first time our delivery arrived before 8 a.m. |
You don't need rung five everywhere. One story per piece and a few numbers and names is usually enough to change how the whole thing reads.
Where to find specifics fast
You usually have more detail than you think. It's just not in your head when you start writing. Check:
- Your sent email and chat history. Search the topic. Real dates, names and quotes are sitting there.
- Your calendar. When did the project start? How many meetings did it take?
- Your own data. Sales numbers, website analytics, survey replies, support tickets.
- Customer and colleague words. A one-line quote from a real person beats a paragraph of description.
- Your brain-dump or interview answers. If you did those, half the work is done.
Never invent them
This matters for search too. Google's guidance on helpful content asks whether a page shows first-hand experience and real expertise. Invented details are the opposite of that. Lesson 7.3 covers AI content and SEO in full.
Before and after
The second version makes a smaller claim, about one company instead of "many organizations". It's also far more believable, and far more interesting to read.
Circle every generic phrase in a draft. Pick the three that matter most and climb at least two rungs on each. Check the draft in the AI detector before and after, and watch the "Specific details" bar. Then do the read-aloud test on the result.
Was this lesson helpful?
Thanks. Your answer helps us improve the course.