Support slop is apology boilerplate that avoids saying what went wrong or what happens next. It reads as automated because it is evasive, and a customer who is already annoyed reads evasion as a bot even when a person typed it.
The phrases, and what to write instead
| Avoid | Write instead |
|---|---|
| Dear valued customer | Use their name. |
| We apologize for any inconvenience this may have caused | "Your order is four days late. That is our mistake." |
| I sincerely apologize for any inconvenience | Apologise for the specific thing, once. |
| Rest assured | State the next step and when it happens. |
| As per our policy | Explain the reason in plain words. |
| Unfortunately, we are unable to accommodate your request | "We can't refund this, because X. Here is what we can do." |
| Thank you for reaching out | Open with their issue. |
| Great idea! I'll pass that along to the team | Say whether anything will happen. If it will not, say that. |
| We take your feedback very seriously | Delete. Show it instead. |
| Your call is important to us | Delete. |
| I completely understand your frustration | Show it: repeat the specific problem back. |
| We are working diligently to resolve this | Say who is working on it and when you will update. |
| enthusiastically, flawlessly, efficiently | Cut the adverbs and state the fact. |
| Please don't hesitate to reach out with any further questions | "Reply here if it happens again." |
Score this draft before you send it
Paste your text into Slop Score. It checks every rule on this page plus the full 200+ list, marks each hit in place, and cleans the mechanical ones. Runs in your browser; nothing is uploaded.
Check my draft →The rule underneath all of them
Every replacement above does the same thing: it swaps a claim for evidence. That is the whole method, and it is why no tool can finish the job for you. A model can delete results-driven. Only you know the number that belongs there.
Density is what readers notice, not any single word. One phrase from this table is a coincidence. Four in a short piece is a pattern, and people now recognise the pattern quickly. The full 200+ tell list covers the vocabulary, sentence patterns and machine artifacts that apply everywhere, whatever you are writing.
Questions
Why do customers react badly to AI-written support replies?
Not because a machine wrote it, but because the reply avoids the two things they want to know: what went wrong and what happens next. Boilerplate apology reads as a brush-off. A blunt reply naming the fault and the fix lands better than a polished one that says nothing.
What makes a support reply sound like a bot?
Apology without a subject ("any inconvenience"), policy citations without reasons, and the word "valued". All three are ways of not saying the specific thing, and specificity is what proves a person read the ticket.
Can we use AI drafts for support without sounding robotic?
Yes, if the agent replaces the placeholders with the actual fault, the actual fix and the actual date. The failure mode is shipping the draft with its generic apology intact, which is exactly what a rushed agent will do under queue pressure.