The Seven Deadly Claims 2/7: "We Integrate With Hundreds of Systems."
The logo wall is a staple of every CIM. ERPs, CRMs, HR systems, banks. And a sample number: "300+ integrations." The number is almost always true. At some point, somebody connected the product to each of those systems. The question nobody asks is how? I’ll get to why this matters in the age of AI….bear with me.
When I dig in, the logo wall usually mixes four very different things:
File-based. Batch file uploads, often over SFTP, on a schedule. Customer IT or your services team sets it up. When it breaks, the customer finds out because the numbers are wrong.
API only. A documented API. The customer's IT team or a systems integrator builds and maintains the connection.
Services-built. Your professional services team builds the integration for each customer, billed or absorbed.
Productized. A prebuilt connector a business user or partner configures inside the product in hours, with field mapping and data transformation built in. The product monitors it, alerts the customer when something fails, and retries or fixes itself.
Only the fourth one is a software capability. The first three are a services business living inside a software business.
That matters for three reasons. Integration is usually the longest part of implementation, so it sets time to value. It's a large share of service hours, so it sets service margin. Integrations that fail silently are one of the most common sources of early escalations and churn (ask any AM, CSM or CCO!). A customer who waits 90-180 days for a live data feed, then loses trust when it breaks without warning, doesn't renew with enthusiasm and is likely at risk of churning.
The questions I ask go straight at the mix. Of the integrations you list, how many are prebuilt connectors, how many are API-based, how many are file-based, and how many are custom builds? For your top ten systems by customer count, who sets up the integration and how long does it take? Can a customer or partner set one up without your services team? Can a business user map fields and transform data, or does it need a professional services expert or developer?
Then the operational questions, which are the ones that separate good from great. How do you monitor integrations? Do you know before the customer does when one breaks? Does the system alert the customer, retry, or self-heal? If the answer to "do you know before the customer" is no, you've learned something important about the support load and the renewal risk.
Here’s a real life example: in the last couple of years I worked for a PE backed software company … really a great business overall with a compelling product portfolio, rule of 50 financials (heavily weighted towards EBITDA vs. top line growth). I was shocked to learn that 50% of the employees in the company worked in professional services and implementation cycles could stretch from 6-12 months. As I dug into the issue, several things became apparent:
1) Implementation was never a product design consideration so shadow applications only available to the PS team emerged to configure and setup the customer’s instance (sound familiar my CPO readers?)
2) Yesterday’s customer was conditioned to accept the 6-12 month implementation cycle AND pay for it. Today’s customer not so much….New AI first entrants are offering days or weeks to implement with low to zero implementation fees.
3) Integrations with the customer’s system(s) of record were ALWAYS the long pole in the tent and required a specialized sub group of professional services folks to figure out. This team grew and grew, but could never keep up. When someone left, all the context left with them. When something broke post production (and it always does), this team would get dragged back in to figure out what went wrong and how to fix it… queue the firefights that emerged with professional services, customer support, CSMs, product managers and engineers lighting up Slack/Teams channels to triage and resolve the issue. The other channel that lit up was between the CPO, CTO, CRO, CCO and CEO demanding answers and resolution. Now, like most firefighters, we always extinguished the blaze (eventually). The trailing impact was that the churn rate stayed stubbornly flat or increased YoY. The GRR struggle became a perpetual nightmare and the NRR rate dropped dramatically. Would YOU buy more software from a vendor who put you through that inconvenience and stress on a regular basis?
In yesterday’s PE playbook, this type of issue almost NEVER surfaced in due diligence during a process. 30% EBITDA + 15% top line revenue growth? Check. GRR = 93%? Check. NRR = 105%? Check. AI agents shipped? Check. Who cares? Answer: everyone should care, a lot.
AI in your product raises the stakes here. AI is only as good as the data flowing into it. Stale, fragile, file-based or subpar API based feeds give it stale, fragile inputs. After the happy customers get over the sugar high of your AI agent or chat bot, the adoption and engagement metrics stall out (which prompts a mad scramble to get the numbers back up). The AEs close the first wave of customers based on the nice demo or AI vision, then they start to see deal cycles stretch out and late stage pipe begins to convert to “Closed Lost” or “Pushed to the next quarter.” Let the mad scramble to correct all of this begin, because we’re in a process and OMG, we cannot afford to miss this quarter’s bookings target. Rinse and repeat until a successful transaction occurs or, more likely, we pull the company off the market because the PE buyers who can write a big enough check are hyper cautious. Sound familiar?
KEY TAKEAWAY: That’s the 2026 PE exit market right now — and it's not changing anytime soon. Shameless plug: that’s why we need a new approach for PE backed SaaS businesses. The good news is that Provex Advisory is designed for this new world. The sooner we face up to the changing landscape, the sooner we’ll see proper. value creation and exits at the right valuations.
For deal teams, ask for the integration inventory with method, owner and customer count for each, twelve months of integration incident logs, time from contract signature to first live data feed by quarter, and a live connector setup by someone who isn't an professional services expert or engineer. Inspect for AI native integration tooling (and be wary of DIY integrations using agentic coding… book a discovery call if you want to hear the horror stories). And for god’s sake, require a live demo… talk is cheap, it takes money to buy whiskey…..
For portfolio companies, the path forward is to productize the top five or ten integrations by customer count, add monitoring and alerting before anything else, and stop counting logos. A buyer will be far more impressed by "our top ten connectors cover 80% of customers, set up in a day, monitored end to end" than by a wall of 300 logos. Offer to show your prospective PE investors/buyers a live demo. Highlight it! Use newer AI native integration tooling to instrument these integrations — in the last 6 months I’ve done several proof of concepts with newer unified API/iPaaS players in DAYS that would blow your socks off. Book a discovery call if you want to learn more.
The red flag: most integrations with major systems of record are file-based or services-built, and nobody monitors them. That caps implementation capacity and makes every new customer more expensive to onboard than the model assumes.
Next week: acquisitions that are "Fully Integrated."
