Ryz Labs/Replace/Replace Intercom
Replace

Replace Intercom with Custom Support Software: Runbook

For teams weighing an Intercom renewal: how seats, Fin outcomes and add-ons add up, how conversations come out, and when an owned build pays back.

Owning your support software instead of renewing Intercom is worth pricing when full seats keep spreading across success, sales and product, when conversations need your own account data inline, and when you want every transcript in a warehouse you run. For a dozen or so seats, or a company that gets most of its value from Fin as shipped, Intercom is the better buy. The test is simple: full seats × seat price × 36 months, set against engineers × months × $7,000 and an annual run cost you assume.

Where an Intercom bill comes from

Seats are priced per month but billed annually; the pricing page lists no month-to-month seat price. Advanced costs $85 per seat per month, billed annually, and includes 20 free Lite seats (list price checked November 21, 2026; check your contract). Expert, the highest tier, is bought self-serve rather than by quote, and Intercom reserves contracts for enterprise terms. Usage-based AI and workspace-level add-ons sit on top of the seat line.

ItemPricePerDetail
Essential$19, annual billingseat, monthlyShown as reduced from $29, with "new customers save 35%"
Advanced$85, annual billingseat, monthly20 Lite seats at no charge
Expert$132, annual billingseat, monthly50 Lite seats at no charge; self-serve
Fin AI Agent$0.99outcomeAt most one charge per conversation; Fin on its own goes through sales with a monthly minimum
Fin Voice$1.99voice outcomeMetered
Copilot$29 annual or $35 monthlyteammate, monthlyOptional
Pro$100workspace, monthlyOptional
Proactive Support Plusfrom $99workspace, monthlyOptional

Intercom list prices as published on November 21, 2026. Your contract may differ from list; check it.

Two lines on the invoice grow for unrelated reasons. Seat spend tracks people: a success manager or solutions engineer who outgrows a free Lite seat becomes a paid full seat. Fin spend tracks traffic: at $0.99 an outcome, it rises with the number of customers the AI agent helps, whatever your headcount. Then there are tier moves, $19 to $85 from Essential to Advanced and $85 to $132 from Advanced to Expert. Essential's $19 is labeled as a new-customer price, so read what your renewal says instead of assuming it holds.

Should you keep Intercom?

For a product-led company with a standard support motion, usually yes. Use these rows to frame the internal discussion.

Your situationWhat we'd do
Roughly 15 full seats or under, with conversations that follow familiar patternsStay. In our 12-seat example, upkeep alone costs more than the seats.
Fin handles a big share of conversations and you're happy with it as Intercom runs itStay. A customer-facing AI agent is a separate build with its own model spend.
No one could own a support application once it shipsStay. Owned software needs an engineering owner by name.
The main complaint is API throughputUsually stay. Private apps already share 10,000 calls per minute per workspace.
Full seats are spreading into success, sales and productRun the numbers. Owned cost doesn't rise per person.
Agents need billing, plan or usage data on screen and must write changes backRun the numbers. The conversation becomes part of your own data model.
Quality review, analytics or model evaluation needs every transcript in your warehouseRun the numbers, or begin with the split below.
Split (middle path)Customers keep talking to Intercom and Fin; escalations, approvals and the transcript archive move to code you own, synced from the Intercom API.

Seat math over 36 months

Keep two running totals. Intercom's grows by full seats × $85 each month. Yours grows by $7,000 per build engineer per month during the build, keeps Intercom's seats until one month past launch (the parallel month), and adds run cost from that parallel month onward. Owning breaks even in the first month from which your total never again exceeds Intercom's. The run cost below is a placeholder for you to replace: yearly upkeep in engineer-months at $7,000, plus hosting.

Full seatsAnnual seat spendSeat spend, 3 yearsBuild costYearly run assumptionOwn-it total, 3 yearsBreak-even
12$12,240$36,7202 engineers for 3 months at $7,000 = $42,000$18,000 (2 engineer-months, $14,000, plus $4,000 hosting)$95,580Never; $1,500 a month of upkeep exceeds the $1,020 seat bill
70$71,400$214,2003 engineers for 5 months at $7,000 = $105,000$30,000 (3 engineer-months, $21,000, plus $9,000 hosting)$218,200Month 38, just outside three years
220$224,400$673,2004 engineers for 7 months at $7,000 = $196,000$66,000 (6 engineer-months, $42,000, plus $24,000 hosting)$505,100Month 24

Advanced at $85 per seat per month on annual billing, as listed November 21, 2026 (list price, check your contract). Free Lite seats aren't counted. The own-it column includes 4, 6 and 8 months of Intercom seats (the build plus a parallel month).

The 220-seat case step by step: $196,000 of build, 8 months of Intercom at $18,700 ($149,600), and 29 months of upkeep at $5,500 ($159,500) come to $505,100 over three years, compared with $673,200 for renewing. Month 23 still favors Intercom ($433,600 against $430,100); from month 24 owning stays lower ($439,100 against $448,800). The 70-seat case shows why honest math matters: it costs $4,000 more than renewing over three years and only crosses in month 38. The 12-seat case never crosses at all.

Fin outcomes and Copilot aren't in these rows; if you pay for them, add them to Intercom's side. If your owned tool summarizes, tags or answers conversations with a model, it pays per LLM API call, which belongs on your side. Run your own quote through the 3-year build vs buy TCO calculator.

How an Intercom replacement runs

1. Build a seat map

Write down every full and Lite seat, the team behind it, which inboxes and conversation types it handles, and which workflows and integrations react to new conversations. Then split customer-facing pieces (chat, Fin, outbound messages) from back-office ones (assignment, escalation, tagging, approvals); a split usually leaves the first group on Intercom. Exit check: each person is marked as moving or staying.

2. Pull transcripts, not only metrics

Intercom offers several exports, and they don't all contain the conversation itself. Per Intercom's conversation export article, the Reports dataset export is a CSV of IDs, timestamps, attributes and metrics covering up to two years, and conversation and message content is not included; browser downloads stop at 10,000 rows, and bigger files arrive by email. Inbox bulk export to PDF or text works on 200–300 conversations per run and leaves out internal notes, fine for a legal request but not a migration. Saved-view CSVs refuse views above 100,000 rows. For transcripts, Intercom's route is to list conversation IDs, then fetch each conversation, and its Export API allows at most a 90-day timeframe per request. Cut your history into 90-day windows, gather IDs per window, and fetch each by ID. Exit check: sampled transcripts match what agents see in Intercom, and any missing notes are documented.

3. Model threads in PostgreSQL

A conversation is a thread, so store it as one: a conversations table, a messages table holding customer replies, teammate replies, bot replies and notes in order, plus contacts, companies and a tags join. Unique keys on Intercom IDs make repeat fetches harmless. The payoff is joining contacts to your own users table on your internal user ID. Exit check: one query returns a customer's support history next to their plan and billing data.

4. Start with back-office workflows

Rebuild what happens after a conversation lands: assignment rules, escalation to engineering, refund and credit approvals, SLA reporting. These are the most business-specific pieces and the easiest to run in two places at once. Customer chat stays on Intercom until the back office has proven itself. Exit check: tests built from exported conversations cover each workflow, and each sits behind a feature flag.

5. Sync, auth and webhooks

Authenticate the sync with an Access Token from a private app in your own workspace, sent as a Bearer header to api.intercom.io, as Intercom's authentication docs describe; collecting other people's tokens in place of OAuth breaks Intercom's terms. The rate limiting reference gives private apps a shared 10,000 calls per minute per workspace, enforced in 10-second windows, with 429 rate_limit_exceeded beyond that, so pace requests and back off on 429. List endpoints page with starting_after and allow at most 150 per page, and because search cursors are stateless, edits during a crawl can produce duplicates or gaps (pagination docs); dedupe on ID. Webhooks belong to an app, not a workspace, and the receiving endpoint must take HTTPS POST and answer HEAD validation. On your side, put SSO in front of the agent tool and map roles from the seat map. Exit check: webhook sync plus a nightly reconciliation that finds no unexplained gaps.

6. Pilot one team

Choose a group whose work is mostly internal, often escalations or billing, and move them to the owned tool while conversations still enter through Intercom and sync across. Compare assignment results, response times and reopen rates with that team's own history. Exit check: pilot metrics match or beat the team's baseline, and every gap has a reason.

7. Final pull, then cancel

Move teams in order of how internal their work is, and decide on its own merits whether customer chat and Fin ever move. Take the final transcript pull before you cancel. Per Intercom's cancellation article, cancelling doesn't delete data, but none of it is visible until you resubscribe, and workspaces without a subscription are deleted after 180 days of inactivity, with a warning email 90 days earlier (inactive workspace deletion). Exit check: final counts reconciled window by window and stored in your database.

Before you cancel: checklist

Two weeks with an AI pod on your team

Week one

Week two

Coding assistants take the repetitive work for our AI pod teams: mapping threads to tables, producing fixtures from exported transcripts, writing the windowed export scripts. Engineers review every change, and the test suite has to pass, before it ships.

Team for an Intercom exit

This is mostly application and integration work. Plan on a tech lead who decides the schema and the move order, senior full-stack engineers for the agent tool and the sync (our TypeScript developers often take both), a data engineer if transcript history is large, and QA a few days a week during the pilot. Many teams begin with a pod of 3 to 5 senior engineers (a tech lead plus engineers), and a full pod is about seven. Moving only escalations and the archive is lighter: a lead with one or two engineers covers it.

Ryz engineers on your team start from $7,000 per month (our published range is $7,000–$15,000 per engineer per month, mostly mid-level to senior), quoted per team. Ryz engineers work on your team, reporting to your leads. Once the renewal date is known, scope your team with seat count and conversation volume in hand.

Related

FAQ

At what seat count does owning beat paying for Intercom?

There's no universal number; it depends on your build size and upkeep. Multiply full seats by the seat price and 36 months, then compare with engineers times months times $7,000, plus Intercom until one month after launch, plus yearly run cost. With Advanced at list in our example, 12 seats never cross, 70 seats cross in month 38 and 220 seats in month 24. Fin and Copilot spend go on Intercom's side, LLM API usage on yours.

Which Intercom export gives us full conversations?

Not the Reports dataset export, which leaves out conversation content. List conversation IDs and fetch each conversation through the API, in 90-day windows, since the Export API caps each request at a 90-day timeframe. Page with starting_after, dedupe on ID, and finish before cancelling, because a cancelled workspace's data can't be viewed.

What sets the timeline?

How many teams and workflows move, how much history you bring, whether customer chat moves at all, and how fast the pilot team's metrics settle. We size it from the week-one seat map rather than guessing beforehand.

How would the build be priced?

Engineers times months times the monthly rate, from $7,000 per engineer, plus a yearly estimate for hosting, upkeep and LLM API usage. Intercom seats stay on the bill until each team moves, and those months count toward your side.

Can we keep Fin and move everything else?

Yes, and for many teams that is the right shape. Customers keep talking to Intercom and Fin, while escalations, approvals and the transcript archive move to owned code synced from the API. People who only did internal work may no longer need full seats.

Questions we didn't answer? Email info@ryzlabs.com.

Thanks — your message has been sent. We’ll get back to you soon.

Something went wrong while sending your message. Please try again or email info@ryzlabs.com.

Explore Ryz Labs

Software development servicesAI development companyIndustriesStaff augmentationAI pod teamsNearshore software developmentHire engineers by roleEngineer ratesGuidesBuyer guidesCase studiesHow we vet engineers

Ryz Labs

Senior engineers on your team. AI pod teams that ship.

Tell us what you're building. You get a scoped plan, a price and the names of the people who would do the work.

  • Only the top 1% of tens of thousands interviewed make it
  • On US business hours, including New York hours
  • Trusted by Fortune 500 engineering teams

Tell us who you need

A Ryz partner replies with a scoped team plan.

Start a conversation →

Prefer to talk? Book a 15-min call →