If your product calls Google’s Perspective API for toxicity scoring, you are on a deadline. Google is discontinuing the service, and every team still sending comments to commentanalyzer.googleapis.com needs a replacement.
The good news: you probably do not need to throw away your integration and start over. Respectify was built with Perspective users in mind — including a near drop-in compatible API that keeps the scoring flow you already have, with room to add coaching and richer moderation when you are ready.
This post walks through why teams are migrating, what changes in your code, and what to test before you flip the switch.
Why teams are leaving Perspective
Perspective did one thing well: it returned attribute scores — toxicity, insult, threat, and related signals — that you could threshold, queue, or route into your own moderation logic.
What it did not do was moderate on its own. You still had to decide what to do with a high score: hide the comment, send it to a human queue, or show a warning. The commenter rarely got an explanation or a chance to fix what they wrote.
With Perspective shutting down, the question is not just “what scores toxicity now?” but “what should moderation look like going forward?”
Respectify answers both:
- Perspective-compatible scoring for teams that want to keep their existing flow
- Full comment moderation that catches bad-faith posts before they publish, explains why, and helps commenters revise
If you only need the first part today, you can migrate in an afternoon. If you want the second part later, you can turn it on without ripping out what you already built.
Respectify vs. Perspective at a glance
| Perspective API | Respectify | |
|---|---|---|
| Core output | Toxicity attribute scores | Scores or pre-publish moderation with feedback |
| Blocks before post | No | Yes (optional) |
| Explains flags to commenters | No | Yes |
| Suggests rewrites | Limited (suggestscore) | Yes |
| Dog whistle / relevance checks | No | Yes |
| WordPress plugin | No | Yes |
| Pricing | Free (rate-limited) | Usage-based, from $5 |
For a fuller comparison, see our Perspective API alternative page.
The compatible endpoints
Respectify exposes two Perspective-style endpoints:
POST /v0.2/perspective-compat/analyse— the main scoring call, analogous to Perspective’s analyse flowPOST /v0.2/perspective-compat/suggestscore— for score corrections and feedback, if you already use Perspective’s suggest flow
Most common Perspective attributes map directly to Respectify. Some Google attribute names are preserved as aliases. A smaller number are approximate matches rather than exact replacements — the attribute mapping documentation lists those cases clearly.
If your application already thresholds on TOXICITY, INSULT, or similar attributes, this is the path of least resistance.
Migration in four steps
1. Create a Respectify account and API key
Sign up and grab an API key from your dashboard. The $5 minimum top-up is enough to run real traffic through your staging environment and compare scores against what Perspective returned.
SDKs are available for TypeScript, Python, and PHP, or you can call the REST API directly.
2. Swap the endpoint and authentication
For most migrations, three things change:
- Replace Perspective’s URL with Respectify’s compatibility endpoint (
/v0.2/perspective-compat/analyse). - Replace Perspective’s query-string API key with Respectify’s auth headers (Bearer token).
- Keep your request body shape — attribute lists, comment text, and language hints carry over for the common case.
Everything else in your pipeline — thresholds, queues, logging, admin tools — can stay as-is while you validate output.
The migration overview in our docs is the authoritative reference for request and response details.
3. Test the parts that matter for your product
Before sending production traffic, exercise the edge cases your users actually produce:
Languages. If you send languages in the request, Respectify preserves them. If you do not, Respectify does not claim language detection — the response falls back to an English-oriented default moderation context. Plan accordingly if you moderate non-English communities.
Spans. If your UI highlights flagged text using span offsets, test with emoji, quoted text, ampersands, and content that looks like HTML. The compatibility endpoints return UTF-16 offsets aligned with Respectify’s returned spans.
Score corrections. If you use Perspective’s suggestscore flow, Respectify supports it on POST /v0.2/perspective-compat/suggestscore. Note that feedback sent to this endpoint is retained intentionally to improve the service — do not send content you are not comfortable storing.
4. Go live, then layer on more checks
Once scores look right in staging, point production traffic at Respectify. Watch your dashboard for usage and balance.
When you are ready to go beyond scoring, Respectify can catch comments before they post, flag off-topic replies, detect dog whistles, and coach commenters toward better-faith rewrites — all configurable per site, topic, and audience. You do not have to enable everything on day one.
When to use scoring-only vs. full moderation
Stick with perspective-compat scoring if:
- You already have moderation workflows built around attribute thresholds
- You need to migrate quickly before Perspective shuts down
- You want to validate Respectify’s scores against your existing rules first
Add Respectify’s full moderation if:
- You want bad comments stopped before they publish, not cleaned up after
- Your team spends too much time in review queues
- You want commenters to learn why something was flagged instead of guessing
Both paths use the same account and API key. Many teams migrate the scoring endpoint first, then enable pre-publish checks on a subset of traffic.
WordPress users
If you run comments on WordPress rather than a custom app, you do not need the compatibility API at all. Install the Respectify WordPress plugin, add your API key in Settings, and comments are analysed before they publish. See our WordPress comment moderation guide for the full walkthrough.
Common questions
Is this really a drop-in replacement?
It is a near drop-in replacement. Endpoint swap, auth header change, test spans and languages — that is the typical migration. You are not re-architecting your moderation system.
Will my thresholds still work?
For the common Perspective attributes, yes. Check the attribute mapping docs for any attributes you rely on that do not have a direct equivalent.
What does it cost?
Usage-based, billed per comment. Top-ups start at $5 with optional auto-topup. No monthly subscription. View pricing.
Where do I get help?
- Migration overview
- Perspective-compatible scoring docs
- Contact us if you are migrating a high-volume integration and want to talk through the cutover
Start the migration
Perspective served a generation of products that needed toxicity signals without building ML in-house. Respectify picks up that job — and gives you a path to moderation that actually teaches commenters, not just scores them.
Ready to switch? Get your API key, read the migration guide, and see the full Perspective API alternative overview.