I’ve been using the Perplexity app and I’m unsure how to fairly review it. Some features work great for quick answers, but I’ve run into issues with accuracy, reliability, and understanding what data it uses. I’d really appreciate guidance from others who’ve used it—what should I focus on in my review, and how do I decide if it’s worth recommending to others?
I’d split your review into a few concrete sections so it feels fair and specific.
- What it does well
- Fast for short factual stuff. Definitions, summaries, high level overviews.
- Nice for “I need a quick orientation” type questions.
- The UI is simple, so people can jump in without a tutorial.
Example line: “For quick summaries and simple factual questions, it works fast and is easy to use.”
- Accuracy and reliability
This is where you get specific.
- Note if it ever gave you wrong info. Add 1 or 2 clear examples.
Like “It told me X about tax rules in my state, which was wrong when I checked the official site.” - Mention if it sounds confident even when wrong.
- Mention how often this happens for you. For example “Out of ~30 complex questions, I had to double check at least 10.”
- Sources and data
If you feel unsure what data it uses, write that directly.
Some points you can include:
- “It shows links, but it is not always clear how much it relies on each source.”
- “I still needed to click through to original articles to verify important info.”
- “I am not fully sure how my own data and chats get stored or used, and the app does not explain this in simple language.”
If you care about privacy, say things like:
- “I want a clearer, plain language privacy explanation inside the app, not only a long policy page.”
- “I would like easy settings to control if my chats are used to train models.”
- When it works vs when it fails
Give readers guidance.
For example:
- “Good for: learning concepts, brainstorming, summaries, rewriting text.”
- “Not good for: medical, legal, financial decisions, time sensitive news, niche technical topics.”
That keeps your review balanced and practical.
- Your rating logic
Explain how you got to your stars or score. Example:
- “3/5. Speed and ease of use are strong, but accuracy and trust issues mean I double check anything important.”
- Or “4/5 for casual use, 2/5 if you need reliable expert level info.”
- Tone suggestion
Try something like:
- “I like the idea and I use it, but I treat its answers as a starting point, not final truth. I still cross check with trusted sources.”
If you want to be extra fair, you can end with:
- One thing you wish they improved first, for example “clearer privacy controls” or “more obvious warnings about checking facts.”
That gives future users and the devs something concrete to work with, and your review feels honest, not ranty.
If you’re aiming for “honest but not ranty,” I’d actually write the review around your use-cases instead of feature buckets like @reveurdenuit suggested.
Something like:
- Start with how you actually use it
Example:
“I’ve used Perplexity daily for 3 weeks for:
- quick factual checks
- explaining technical concepts
- planning / brainstorming
- a few heavy ‘trust-this-answer’ things like health/finance.”
That frames the rest: people know what kind of user you are.
- Tell 2–3 concrete stories, not a checklist
Instead of “accuracy is a problem,” do mini stories:
-
“It nailed this” story
“I asked it to explain a math concept / coding API / historical event in plain English, and it turned a confusing mess into something I got in 2 minutes.” -
“It kinda failed me” story
“I asked about [specific topic: tax rule, medical thing, niche tech]. It gave a very confident answer that turned out to be partially wrong when I checked [official source]. I would’ve made a bad decision if I trusted it blindly.” -
“Mixed result” story
“For [topic], it gave a decent overview, but I ended up clicking through 3 of the sources it showed and realized it glossed over important caveats.”
You don’t need a ton of examples, just 2–3 with details. That feels a lot fairer than a generic “sometimes it’s wrong.”
- Be blunt about the “trust” problem
You can literally say:
- “If I’m just learning or brainstorming, it’s amazing.”
- “If I need something I’d normally ask an expert or lawyer or doctor, I treat its answer as a draft hypothesis and then go verify manually.”
You’re basically telling people: “This is a thinking assistant, not a source of truth.”
- Talk about data & privacy like a normal person, not a lawyer
Forget the policy language. Describe your actual feeling:
- “I still don’t fully get what they train on or keep. The app mentions sources, but it’s not transparent in human language about how my chats are stored or reused.”
- “I’d like a simple ‘Privacy 101’ screen in the app that explains:
- what is saved
- what is used to train
- what is shared, and with who
- how to opt out.”
If you’re confused, say “I’m reasonably tech literate and still not sure how my data is used.” That hits harder than generic ‘privacy is a concern.’
- Describe how you’ve changed your behavior because of it
This is super honest and really useful:
- “I stopped asking it anything sensitive (health, financial account details, personal info) because I’m not fully sure how that data is handled.”
- “I now always open at least one of the cited links for anything important.”
- “I don’t use it for real-time news since it tends to miss context or get details slightly wrong.”
Behavioral changes tell readers more than any star rating.
- Then drop your rating as a split, not a single number
Instead of “3/5 overall,” do something like:
- “5/5 for learning and summarizing complex topics quickly.”
- “2/5 for anything where being wrong has real consequences.”
- “3/5 overall, because I constantly have to fact-check and I’m not fully comfortable with the data situation.”
That lets people self-sort. Some folks only want a summarizer and will see your review as basically positive.
- One short “wishlist” paragraph
End with 2–3 things they could realistically fix:
For example:
- “Clear, simple privacy controls in-app (not buried in legal text).”
- “Stronger warnings around medical / legal / financial questions.”
- “More obvious indication when info might be outdated or low confidence.”
That keeps your tone balanced: you’re criticizing, but you’re also pointing to specific fixes.
If you write it that way, it comes across as: “I use this, I like parts of it, but I’m not going to pretend it’s magic or perfectly trustworthy.” Which is probably the fairest review you can give.