Skip to content
Log in Start now →
Product
PlatformPricingDimensionsIntegrationsSecurityCompare
Solutions · by use case
Customer, guest and fan experienceProduct feedback and prioritizationRevenue, churn and retentionAgenciesContact center performanceAI agent and chatbot analyticsMulti-location performanceReviews and competitive benchmarking
Solutions · by industry
Hotels and hospitalityRestaurants and food serviceFitness, golf and recreationRetail and e-commerceSports and live entertainmentInsuranceBanking and financial servicesSoftware and SaaSAutomotiveConsumer products and CPGContact centers and BPOsPublic sector and nonprofitReal estateGaming and betting
Learn
Artifacts GalleryCustomer storiesResearchBlogThought leadershipLessons Log in Start now
Dimension Labs

Use case

AI agent and chatbot analytics

Read every conversation your bot or AI agent has. Find where automation fails, why customers ask for a person, and what each failure costs.

Start now › See pricing ›
Ask Where does the bot fail, and what does it cost?

We work with enterprises such as NBA, Travelers, Mars and Concentrix.

NBATravelersMarsDisneyConcentrixIntuitSeattle MarinersDropboxAmerican Red CrossResideoNetflixThomson ReutersAsicsStakeEasygoFleishmanHillard
[01] The problem

Containment looks great. The customers who gave up don't show up in it.

A conversation that ends without a handoff counts as contained, including the ones where the customer quit and called instead. Bot dashboards count turns. They don't read them.

Right now
  • 01Containment counts abandonment as success
  • 02Failure modes are found by reading transcripts by hand
  • 03Nobody knows which intents are safe to automate next
  • 04Escalations have no named cause
  • 05Model changes ship without measuring their effect

We started in conversational AI analytics, in 2016. This is the problem we know best.

[02] EnrichOne record, structured

Language in. Structured data out.

Every record is read in full and written back with dozens of typed Dimensions, the same way every run. Here is what that looks like for this use case.

enrichment · one record
Raw text

I asked for a person three times before it let me through. It kept sending me the same help article, which doesn't cover my situation. Twenty minutes to change one address.

Bot conversation · handed off · resolution joined
Dimensions written to the record
  • Human-Escalation Attempt3 attempts
  • Bot Containmentfailed
  • AI Failure Modeloop · repeated answer
  • Self-Service Attemptarticle not relevant
  • Effort Levelhigh · 20 minutes
  • Automation Opportunityaddress change

Each colour links a phrase in the raw text to the Dimensions extracted from it. Hover either side.

[03] The product
3.1 · Read everything

Every record, enriched.

Start with one export of bot conversations, AI agent transcripts and handoff tickets, or connect every channel. Every record comes back with typed Dimensions like bot containment, AI failure mode, human-escalation attempt, self-service attempt and automation opportunity, written the same way every run, then joined to intent, resolution, handoff and account data.

  • Read every record, not a sample
  • Join each record to intent, resolution, handoff and account data
  • Add your own Dimensions in plain English
  • Backfill years of history in the same run
dimension labs · blend
3.2 · Prove the cause

Ask why. Get proof.

Ask in plain English. The agent writes the queries, controls for what would confound the answer, and ranks each issue by its measured effect on true containment, escalation and churn, with an effect size and a confidence interval. Findings that fail the robustness checks are never reported as causal.

  • Where does the bot fail, and what does it cost
  • Separate real containment from customers who gave up
  • Rank the next intents to automate by value
  • Measure whether a model or prompt change worked
dimension labs · ask
Example finding

Where does the bot fail, and what does it cost?

Failure modes ranked by the escalations and repeat contacts they cause, with the transcripts behind each one.

Every bot and agent conversation · handoffs joined
3.3 · Deliver

Ask for the report. Get the report.

Ask for the readout your automation team needs and the agent builds it: summary, ranked findings, charts and recommendations, each figure checked. Schedule it weekly or monthly on fresh data, or publish it as a live monitor.

  • Produce the monthly readout without building it
  • Publish a live monitor leadership can open and question
  • Track whether the fix you shipped moved the number
  • Export to PDF, slides or your BI tool
dimension labs · report
[04] How the platform worksRuns daily

From raw conversations to a proven answer.

Language from every channel and your business data flow into one pipeline that enriches, governs and tests every record, then delivers wherever your team works.

Language · 37 sources
ZendeskIntercomSalesforceLiveChatDriftSlackQualtricsMedalliaSurveyMonkeyQuestionProGenesysAmazon ConnectAvaya Cloud8x8TwilioSMSAdaKore.aiYellow.aiCognigyRasaMicrosoft Bot FrameworkAmazon LexAmazon AlexaOracle Digital AssistantBotcopyCapacityFacebook MessengerKikX (Twitter)OpenAIEmailCall recordingsApp store reviewsGoogle reviewsOpen-ended surveysCommunity forums
Business + world data
SnowflakeDatabricksBigQueryAmazon S3Salesforce CRMCSV + ExcelREST APIRevenueOrdersRenewalsUsage eventsLoyaltyStore + location dataWeatherSeasonalityMarket events
Dimension Labs pipeline · Runs daily
01
Ingest + normalizeEvery channel and language into one schema
02
EnrichDozens of Dimensions written onto every record
03
Map + governValidated for drift, versioned, the same fields every run
04
Analyze + proveJoined to outcomes and tested for cause
Where it goes
Dimension Labs agentAsk, report, schedule
Your warehouseSnowflake, Databricks, BigQuery
BI toolsTableau, Power BI, Hex
Claude and ChatGPTvia MCP · coming soon
[05] ArtifactsBuilt by the agent

What you get. Reports, monitors and decks, built by the agent.

Reports · Live monitors · Apps · Decks · Charts See every artifact ›
[06] Proof

Intuit QuickBooks Establishes Structured Chatbot Intelligence Across 1M+ Monthly Sessions

Intuit QuickBooks supports more than seven million small businesses worldwide, providing financial management tools across accounting, payroll, invoicing, and payments.…

Read the customer story ›
[07] Explore8 use cases · 14 industries
By use case
Customer, guest and fan experienceWhich issue moves your scoreProduct feedback and prioritizationWhat to build, ranked by impactRevenue, churn and retentionWhat customers said before they leftAgenciesClient analysis in hours, not weeksContact center performanceWhat separates your best agentsAI agent and chatbot analyticsWhere automation fails, and what it costsMulti-location performanceWhy one site beats anotherReviews and competitive benchmarkingYou and every competitor, side by side
By industry
Hotels and hospitalityGuest experience, property by propertyRestaurants and food serviceSpeed, accuracy and the operator effectFitness, golf and recreationMember churn, club by clubRetail and e-commerceStore, site and delivery, on one scaleSports and live entertainmentFan experience and renewalInsuranceClaims, renewals and premium changeBanking and financial servicesFees, trust and attritionSoftware and SaaSChurn signals and product prioritiesAutomotivePurchase, delivery and serviceConsumer products and CPGProduct voice and defectsContact centers and BPOsEvery client program, one methodPublic sector and nonprofitConstituents, members and donorsReal estateLead conversations to closingsGaming and bettingPlayer conversations at scale
Dimension Labs/Start with one file From $299 a month

Find out what's really moving your number.

Start with one file. Ask why a number moved. Plans from $299 a month, no demo and no contract on monthly plans.

Start now › See pricing ›