When PE firms assess SaaS targets, the conversation centers on revenue quality. Net revenue retention. Churn cohorts. Customer concentration. CAC payback. These are the right questions. They are also the questions every buyer is asking, which means the seller has prepared for them.
The domain that swings valuation in SaaS acquisitions and rarely gets examined pre-LOI is data model governance.
Specifically: is the schema documented and version-controlled? Are database migrations tracked and reversible? Does anyone other than the lead engineer understand the data architecture well enough to hand it off? Is the data model a shared institutional asset, or is it tribal knowledge maintained by one person who may not survive the transition?
These questions do not appear in a quality of earnings report. They barely surface in a technical due diligence engagement unless someone specifically commissions that scope. And yet the answers to them are, in many SaaS acquisitions, the difference between an integration that runs on schedule and one that runs 18 months over budget.
Here is why it matters at the LOI stage. When a SaaS company is acquired, the integration timeline is almost always modeled on assumptions about how the data will move. How the acquired product will connect to the acquirer’s systems. How customer data will be normalized. How reporting will consolidate.
Those assumptions are made based on what the seller tells the buyer in the LOI room. If the data model is undocumented, the seller often does not know enough to tell you the truth — not because they are being deceptive, but because the architecture lives in one engineer’s head and has never been articulated in any other form.
The questions to ask before the LOI
- Is the database schema documented and accessible outside the engineering team?
- Are migrations version-controlled with a complete history?
- Can the data architecture be explained in writing by someone other than the lead engineer?
- Has the data model been reviewed by anyone outside the company in the last 24 months?
- What happens to data access if the lead engineer leaves in the 90 days post-close?
The consequence of undocumented data architecture is not technical. It is financial. Integration timelines extend. Engineering resources get absorbed by archaeology instead of development. Product roadmaps slip. The thesis that justified the acquisition multiple begins to erode.
In roll-up strategies, the problem compounds. Each add-on acquisition with a different, undocumented data model adds to the integration debt of the platform. What was manageable in isolation becomes structural risk at scale. The gap you tolerated in the third acquisition becomes the constraint on the fifth.
The posture question is not whether the target has a data model. Every SaaS company does. The question is whether the data model exists anywhere other than in one person’s memory. That distinction is worth asking before the LOI. It will cost you significantly more to discover it after.