Four Things Vendors Call a CRM Integration

One word, four products#

Six vendor pages read on 8 September 2026 all promise the same thing in the same words: send from your CRM, log it on the timeline, personalize from contact properties.

Read further and the products diverge completely. One installs an app. One hands you a webhook builder. One paints a send box into your browser. One only pulls leads in.

A buyer comparing pricing pages cannot see that difference, because the difference is not a feature. It is where the running code lives.

• Tuco AI publishes a native HubSpot workflow action, described as drag and drop Send iMessage steps inside HubSpot Workflows (tuco.ai/integrations/hubspot, read 8 September 2026).
• Blooio publishes a HubSpot connection built on HubSpot's own native webhook action, with the setup steps written out for the buyer to follow (blooio.com/integrations/hubspot, read 8 September 2026).
• Sendara publishes a Chrome extension for Pipedrive, described as working inside the Pipedrive interface so reps do not switch tabs (sendara.io, 13 July 2026, read 8 September 2026).
• One self-managed vendor publishes a Meta Lead Forms connection and states plainly that it is a lead source, not a message mirror, and that nothing you text is written back.

Shape one: the code runs inside the CRM#

Installed apps and native workflow actions live in the CRM's own automation engine. A deal stage changes at 2am, the action fires, the message goes, and nobody had to be awake.

Vendors describe this as native, and the honest test is whether the send appears as a step in the CRM's workflow builder rather than as a link out to a second dashboard.

Blue Reacher ships native connections for GoHighLevel, HubSpot, Close, Salesforce, Pipedrive and Follow Up Boss, so this shape is the one our own integration pages describe.

• Failure mode: the vendor's app has to keep pace with the CRM's API changes, and a marketplace listing can be pulled or paused by the CRM rather than by the vendor.
• What to verify: open the CRM's workflow builder and confirm the send appears as a selectable action with your merge fields available inside it.
• What to verify: confirm a reply writes back to the record as an event the CRM can trigger on, not just as a note a human can read.
• Tuco AI's HubSpot page lists attribution properties written back to HubSpot so reply rate can be reported in HubSpot dashboards (tuco.ai/integrations/hubspot, read 8 September 2026).

Shape two: you build it out of the CRM's webhooks#

Webhook builds are real integrations and they work, with one cost buried in the setup instructions: the automation primitive belongs to the CRM, so the CRM's plan tier decides whether you can use it.

Blooio's HubSpot page is unusually direct about this. Setup runs about 15 minutes if you already have a HubSpot Professional workflow license, and the page states it works on HubSpot Professional and Enterprise plans.

Price that honestly. A $39 monthly messaging bill sitting on top of a CRM upgrade is not a $39 monthly decision, and the CRM upgrade is the larger of the two numbers by an order of magnitude.

• Blooio's HubSpot connection is published at $39/mo with no per-message fees and no A2P registration required (blooio.com/integrations/hubspot, read 8 September 2026).
• Failure mode: nobody owns the connection. A webhook you assembled is not on the vendor's support surface and not on the CRM's, so a silent 401 is yours to find.
• What to verify: which CRM plan tier the webhook or custom-code action requires, in writing, before the messaging contract is signed.
• What to verify: what happens to the build when your CRM admin leaves, since the knowledge of how it was wired usually leaves with them.

Shape three: the code runs in a browser tab#

Browser extensions inject a send box into a page the CRM renders. Sendara's Pipedrive write-up describes exactly that, a Chrome extension so reps send and place calls without switching tabs.

For a rep working a queue, that is genuinely useful and often faster than a native app. The limit is structural rather than a quality problem, and the limit is time.

An extension does nothing while nobody is looking at the screen. Speed-to-lead at 11pm, a nurture sequence over a weekend, an automated reactivation pass: none of those exist in this shape.

• Sendara publishes a flat $249/mo for the first line, $149/mo for each additional line and a one-time $99 setup, with a dedicated line provisioned within 24 hours (sendara.io, read 8 September 2026).
• Failure mode: the record depends on the extension writing back. Close the tab mid-conversation and the CRM's version of that thread can end where the rep stopped watching.
• Failure mode: browser extension permissions are a security review item at any company with an IT function, and that review lands after the contract in most buying processes.
• What to verify: whether an automated send is possible with every browser closed, and if so, which of the other three shapes is doing that work underneath.

Shape four: leads travel in and nothing travels back#

One-directional connectors are the shape most often mistaken for the others, because they sit on the same integrations index page with the same logo treatment and the same word above them.

That vendor's Meta Lead Forms page states the boundary in its own words: this is a lead source, not a message mirror, and nothing you text is written back to Facebook or Instagram.

Nothing is wrong with that design. A lead form is not a system of record. The problem is a buyer counting it as CRM coverage on a comparison spreadsheet where it is not comparable.

• Its own capability list for that connection reads No on text messages, No on photos and video, No on voice memos, No on calls and recordings.
• Failure mode: the conversation lives in the messaging tool only, so anyone reporting from the source system sees a lead created and no follow-up activity at all.
• What to verify: whether the connection writes any event back, and if it does, which specific object it writes to.
• What to verify: where a compliance officer would go to reconstruct a full conversation history if asked for one.

The capabilities that quietly do not survive the trip#

Shape decides whether anything runs. The next layer down decides what actually arrives, and this is where the category is unusually quiet, because almost nobody publishes a capability table at all.

One self-managed vendor published one on 8 September 2026 covering nine CRM connections against ten questions each. Reading it is the fastest education available in this category right now.

Their table is about their own connections, so the numbers are theirs. The useful part is the shape of the answer set, because the same gaps recur across the vendors that publish nothing.

• Group chats read No on eight of the nine connections in that table. A thread where a prospect adds their operations lead is the highest-intent thread in a pipeline and it is the one least likely to reach the record.
• Reactions read No on seven of nine, and In part on the ninth. A tapback is intent with no message row attached, so a CRM that drops reactions reports an engaged contact as silent.
• Reply from inside the CRM reads No for Pipedrive and In part for Follow Up Boss, Attio and Meta Lead Forms, which is a different claim from the same CRM being called integrated.
• Two connections in that table are marked Coming soon while still carrying a full row of capability answers, which is worth knowing before a row of ticks gets copied into a spreadsheet.

Five questions that separate the four shapes#

Vendor pages answer none of these directly, and every one of them is answerable in a ten-minute call or a trial account. Ask them in this order.

Question one does most of the work, because the answer to it constrains every answer that follows. The rest exist to catch the cases where the first answer was optimistic.

Write the answers next to the price rather than under it. A cheaper line on a shape that cannot run overnight is not cheaper, it is a different product.

• With every browser on my team closed, can a message send from a CRM event? If the answer needs a caveat, the shape is an extension.
• Which plan tier of my CRM does this require, and is that the tier I am on today?
• Does a reply land as a triggerable event on the record, or as text a human has to open and read?
• Do reactions, group threads and media arrive on the record, and if not, which of them are on a roadmap with a date?
• When an opt-out is captured in the messaging tool, how long before every other channel in the CRM stops sending, and what is the longest link in that chain?

Change the order you evaluate in#

Most buying processes compare price, then features, then integrations, which puts the structural question last and the reversible questions first. Invert it.

Shape first, because shape is the only one of the three that cannot be changed by a support ticket. A missing feature can ship. A browser extension does not become a background worker.

Then capability, then price. A line that costs more and posts every reply, reaction and group thread onto a record your automations can read is doing a different job from one that does not.

• Open the vendor's integrations index and sort their own pages into the four shapes before reading a single price.
• Ask for the capability table. If one does not exist, ask the ten questions from that published version against your own CRM, since a vendor who has answered them once can answer them again.
• Test the overnight case in a trial: fire a CRM event with nobody signed in and check what reaches the record by morning.
• Keep the CRM plan tier requirement in the same document as the messaging quote, because the two numbers are one number.
• Re-check after any CRM migration, since a shape that worked in one system is frequently a different shape in the next one.

Frequently asked questions