Export with company and plan
Pull conversations for a fixed window with the company, plan, ARR where available and the date. A conversation without its company is a quote you cannot price.
[ 01 | THE METHOD ]
Five steps, by hand, from conversations and tickets
Do it once by hand before you automate it. You will learn what your customers call things, and you will know what a good automated read looks like.
Sample the closed conversations
Read 100 closed conversations, not open ones; the full exchange is what tells you whether a question was a request, a bug or a comparison. Freeze a first taxonomy.
Tag requests and mentions separately
Requests describe a job the product does not do. Competitor mentions are names, prices or features attributed to another tool. Keep the lists apart.
Count companies, weight by ARR
Distinct companies per theme, with revenue attached, ranked by revenue. Chat volume rewards the loudest; revenue rewards the accounts you need to keep.
Review weekly, quotes attached
Take ranked themes to product and mentions to marketing with the quotes behind each. Decide, assign an owner, and check what moved next week.
[ 02 | PITFALLS ]
Three ways the manual read goes wrong
Bot replies pad the volume
Automated replies and resolution bots inflate counts. Tag the customer’s words, not the thread length.
Short messages hide intent
A three-word question can be a request in disguise. Read the exchange, not the first message.
Mentions cluster after moves
When several companies name the same competitor in one week, something happened at that competitor. Find the move; the mentions are the effect.
[ 03 | WITH HYPERORBIT ]
The same method, run continuously by three agents
OVER THE INTERCOM CONNECTORConnect Intercom once over OAuth and Chorus reads every conversation as it closes, clusters requests into themes and sub-themes with the quotes and the ARR of the companies behind each, and keeps the read current. Recon logs each competitor mention and ties it to the move that preceded it. Atlas folds the pattern into account health and watches renewal windows, so a run of comparisons from one company reaches its owner with a play rather than a transcript.
Chorus, the voice-of-customer agent
Clusters reviews, calls, tickets and surveys into ranked themes and sub-themes, each with the quotes behind it and the ARR of the accounts that raised it.
Recon, the competitive intelligence agent
Watches competitor sites, changelogs, pricing pages and reviews every day, scores each move, and drafts a battlecard for the ones that matter, held for review.
Atlas, the account intelligence agent (beta)
Fuses what Chorus and Recon find with usage, support and CRM data into one health score per account, then watches every churn and renewal window you set.
[ 04 | GO DEEPER ]
Related on hyperorbit.ai
INTEGRATIONS76 connectorsEvery source connects once over OAuth and is read continuously.
SOLUTIONVoice of customerEvery channel clustered into ranked, priced themes.
WORKED EXAMPLECause and effectA competitor free tier, 31 more pricing complaints 8 days later, $214K ARR touched.
PRICINGStart freeStarter: one feedback source, 50 signals and 10 insights a month.
[ 05 | FAQ ]
Questions people ask
Does HyperOrbit integrate with Intercom?
Yes. Intercom is one of the 76 connectors, connected once over OAuth and read continuously, and one of the delivery integrations Enterprise plans include.
Does Chorus read chat differently from tickets?
The clustering is the same: by intent, with evidence. Chat is shorter and faster, so the read updates more often; the themes and sub-themes are what you review.
Can I run the method by hand?
Yes. The five steps above need an export, a spreadsheet and a weekly hour, and they are the method the agents run continuously.
What does it cost to start?
Starter is free: Chorus on one feedback source, 50 signals and 10 insights a month. Intercom can be that source.
Every signal, on one orbit
Connect your first source in an afternoon. The first pass lands before your next standup.
Book a demo