Make helps TripleTen process 300k CRM events per month with Meta’s Conversions API for CRM

TripleTen, an online tech education provider, runs performance marketing campaigns across Meta and other social media platforms to attract people looking to change careers into tech. Meta is their biggest acquisition channel. But until they built a feedback loop between their CRM and Meta's algorithm, the leads it delivered were optimized for the wrong audience.

To address it, TripleTen and connected HubSpot to Meta's Conversions API for CRM via Make. Now, TripleTen is sending qualified lead signals back to Meta after every AI-powered qualification call. Meta's algorithm uses that signal to find more students who match their best leads, not just people likely to fill out a form. 

At the moment, the team processes around 300,000 CRM events per month, sees higher funnel conversion rates, and has achieved a lower customer acquisition cost compared to campaigns optimized on standard lead events.

"When we switched to a new event, we started to see that these leads have a higher conversion rate through the funnel, and their end customer acquisition cost is just better than it was before."

Maksim Epifanov, User Acquisition Lead at TripleTen

The challenges: Quality leads locked behind data restrictions

Form fills gave Meta a misleading signal

The first stage of TripleTen's funnel is a form submission. That event is easy to track and easy to send to Meta directly from the website. The problem is that it tells the algorithm almost nothing useful. 

A form fill indicates curiosity, not intent. It says nothing about a student's motivation, financial readiness, or likelihood to enroll. Optimizing campaigns on that signal meant Meta kept finding people who would click and submit, not people who would convert.

"The first stage of the funnel is when the user fills the form, and you can send this event directly from the website. But this event doesn't mean you have a target user. We decided to optimize our campaigns not on leads, but on quality leads."

Maksim Epifanov, User Acquisition Lead at TripleTen

Qualification data had no path from HubSpot to Meta

TripleTen conducts a structured qualification call with every new lead, run by an AI sales agent that asks about career motivation, income, availability, and readiness to start. Every answer is captured in HubSpot. From that data, the team can predict with reasonable confidence whether a lead is a target student.

The challenge was that none of this data was reaching Meta. The algorithm was making targeting and bidding decisions based on form submissions, while the most valuable signal about lead quality sat unused in HubSpot.

Scaling without a development team

Building a custom server-side integration to pass HubSpot events to Meta would have required developer resources and a significantly more expensive HubSpot plan. For a team that needed to move fast across multiple ad platforms simultaneously, that path was too slow and too costly.

The solution: Conversions API for CRM & Make connect HubSpot data to Meta

An AI qualification call as the conversion event

TripleTen built a qualification process where every new lead speaks with an AI sales agent (built on the Retaily platform) for a call of 15 to 20 minutes. The agent asks structured questions, and the answers are stored in HubSpot against each lead record. 

The team researched what an ideal call looks like: a certain duration, positive sentiment, a specific level of engagement. Only leads whose calls meet those criteria generate the target event that gets sent to Meta.

Make connects HubSpot to Meta in real time

Anton Potorochin, the automation engineer behind the integration, built Make scenarios that watch HubSpot for qualifying lead events and send them to Meta via Conversions API for CRM as they happen. 

Each event carries a calculated monetary value based on lead quality signals, including the student's stated ability to pay. Leads who can commit upfront are weighted more heavily than those who need financing, giving Meta a calibrated signal rather than a binary one.

The same scenarios send events across Meta, Google Ads, TikTok, and LinkedIn simultaneously, processing around 300,000 events per month and generating approximately 2 million credits in Make.

From idea to solution in one week

Without Make, the team estimates they would have needed dedicated developer time to replicate even part of the logic. With Make, the integration went from initial request to a testable, deployable solution in a single week.

"From the request to the best testable solution we can deploy, it's really within one week."

Anton Potorochin, Founder of vatech.io

The results: Higher conversion rates, lower acquisition costs

• Higher funnel conversion rates and lower customer acquisition cost compared to periods when campaigns were optimized on standard lead events or form submissions

• Around 300,000 CRM events are processed per month across Meta, Google Ads, TikTok, and LinkedIn, generating approximately 2 million Make credits

• Integration deployed in one week, with no developer resources required beyond the automation engineer

The signal that keeps getting better

TripleTen's target audience, people who want to change careers but have not yet searched for how to do it, are not reachable through intent-based channels. Meta is well-suited to finding them when the signal it receives reflects real business outcomes.

Every qualified call that flows from HubSpot through Make to Meta adds to the algorithm's definition of a good student. Every campaign gets a little more precise. That compounding effect is what the loop is built for.

"Meta has very good algorithms. If they receive positive and good signals from our business, it's very easy to find the right people."

Maksim Epifanov, User Acquisition Lead at TripleTen