You have thousands of testimonials. You can find six.
Praise piles up faster than anyone can read it. Marketing wants quotes about pricing. Sales wants Head-of-Ops voices. Product wants the objection nobody’s solving. Without structure, every request becomes another long Cmd-F session.
Sentiment Analysis reads each new or edited testimonial on its own and tags twelve dimensions — sentiment, themes, use cases, features, personas, objections, outcomes, and more. The library becomes searchable proof, not a graveyard.
Every quote, x-rayed across twelve dimensions.
One AI call, twelve answers — plus a 0–100 showcase score, two summaries and the asset it suits best. An edited testimonial is analyzed again; approving or tagging one costs nothing.
A label (positive, mixed, neutral or negative) and a score from 0 to 100.
Up to eight topics, named in plain English and scored 0–100, with one marked as primary.
What the customer uses the product for, scored 0–100, with one marked as primary.
Product features the customer names, scored 0–100, with one marked as primary.
Who is speaking, read from the words and the author’s title and company, scored 0–100.
Role, industry and company size, when the testimonial gives them away, each with a confidence.
The problem the customer had before they found the product.
What the customer hesitated over or pushed back on.
The results the customer describes, with the numbers when they give them.
What makes it believable: specific numbers, named results, a before and after.
Short phrases that capture what the customer said.
Up to eight pull quotes, each scored 0–100, with a line on why it works.
Quote in, tags out — on its own.
A new or edited testimonial is found within minutes and queued on its own, and the queue is worked every minute. Nothing to press, no export to run.
status: PENDING
source: form_submission
automatic: on
job: PENDING
model calls: 1
credits: 1
themes: up to 8
on failure: refund
filter: sentiment, theme
sort: AI score
Themes in plain English. Counted for you.
Each analysis names a testimonial’s themes in plain English and marks one as primary. The Insights page counts your analyzed testimonials by primary theme, by feature and by sentiment, and puts the strongest proof beside the counts.
Ask in plain English. Get filters you can see and edit.
AI search turns a request into the inbox’s own filters — sentiment, theme, feature, use case, rating, date, source and more — using only values your project already has, and it uses no credits. Save the result as a personal or team view.
It runs on its own. You decide when it doesn’t.
Analysis spends AI credits, so the cost is shown before anything runs and automatic analysis has a switch for each project.
Tagged once. Used across the product.
What it reads. What it costs.
One AI call per testimonial, its answer checked against a fixed schema before it is saved, and a credit you can see for each one.
| Reads | The testimonial’s text, a video’s transcript when it has one, its title and star rating, the author’s name, title and company, where it came from, and its custom field answers. |
| Returns | A sentiment label and 0–100 score; a primary theme, use case, feature and persona; up to eight entries in each list; a 0–100 showcase score; a short and a marketing summary; the asset it suits best; its language. |
| Checks | The answer must fit a fixed schema, or nothing is saved. Scores are the model’s judgment of the words, not a measured result. |
| Analyzes | Approved and pending testimonials. Archived and spam ones are skipped. |
| Runs | On its own for new and edited testimonials while the project’s switch is on, and on request with Start Analysis or Reanalyze. |
| Costs | 1 credit per analysis. A refresh where nothing it reads changed is free, and a failed analysis gives its credit back. |
| Credits | Included each month, per project: Premium 50, Business 150. Packs of 100, 500 and 1,000 can be added. |
| AI search | Turns a request into the inbox’s own filters, using only values your project has. Uses no credits. |
| Plans | Premium and Business |