Skip to content

SaaS

SaaS pitch deck mistakes investors catch in diligence

Updated

SaaS decks fail diligence in a few repeating ways. The product can be real and the spreadsheet can still describe a different business from the one you have. Partners are checking the definition of the revenue, the shape of the cohort, and whether the moat survives a company that already has the customer.

ARR that is not recurring

Pilots, services, onboarding fees, and one-time setup do not belong in the same number as subscriptions. Split them on the slide. If the interesting growth is in the services line, say so. Calling it ARR does not make the renewal exist. The unit-economics guide is where those lines change gross margin, not just the label.

Logos without a cohort

Ten logos can be ten experiments. Show who is still paying and how long you have actually watched them. A cumulative count that hides a recent group of customers who have not renewed is one of the red flags that reads as a pass even when the logos are genuine. If expansion is in the story, name the mechanism: seats, usage, or a second product. A target with no path is a wish.

A price that contradicts the motion

A product-led story with an enterprise contract and a security review, or an enterprise story with nobody who has sold one, describes two companies. Pick the motion you have evidence for. The sales cycle in the model has to include procurement if that is who buys. Leaving the review out makes the payback look shorter than the deal.

CAC that pretends the team is free

Founder-sold pipeline, a discount to win the logo, and implementation hours are costs. If you are still the person closing, the Series A checklist will ask what happens when someone else does it. Put that version on the slide, or label the current cost as founder-sold and stop projecting it.

A moat that is the model or the interface

Models get bundled. Interfaces get copied. The defense that survives the question is a workflow the buyer would have to rip out, data the incumbent does not have, or a system of record. The public ScribeAI sample is a seed pitch for an AI outbound product that leans on a fine-tuned model. It scores 78/100, with defensibility at 88 and incumbent retaliation at 90 (higher is worse). The dossier's pressure is an incumbent shipping the same workflow to accounts it already has. Use it as a pattern for the competition answer, not as a claim about your category.

Churn explained only as bad-fit customers

Some early customers will be the wrong fit. If that is the whole explanation, and the next slide sells to the same profile, the partner hears that you have not changed who you sell to. Say what you stopped selling, or stop using the explanation.

Check the deck against the company you have

Walk ARR, the cohort, the motion, and the moat against the file you will send. Then paste it into the free demo audit. If the score lands on defensibility or distribution, those are the slides to rewrite before you book the meeting. Marketplace businesses fail a different set of checks. Those are in the marketplace guide.

Questions founders ask

Is ARR the same as revenue?
No. Annual recurring revenue is the recurring portion, annualized, from customers who are actually subscribed. Pilots, services, setup fees, and one-time onboarding are different lines. If a customer can cancel before you deliver the work, it was not ARR.
Should I show customer logos?
Logos are fine as proof that someone signed. They are not a retention chart. Next to the logos, say the price, how you got them, and whether they renewed or expanded. If you only have a few customers, show those facts instead of implying a motion.
What belongs in CAC for a SaaS deck?
The fully loaded cost to win that customer: people, tools, discounts used to close, and implementation you cannot skip. Founder-sold deals are real customers and a misleading cost if you paste that cost into a hired sales model.
What if the moat really is the product experience?
Say what about the experience is hard to copy: a workflow the buyer would have to rip out, data you have that an incumbent does not, or a system of record. A cleaner interface or a finer model, by itself, is a feature. Features get bundled.