Outbound for AI SaaS companies

You are not displacing a competitor, you are creating a line item

Most B2B software is sold into a budget that already exists. The buyer pays for something, you argue that yours is better, and the decision is a comparison. An AI product usually has no line to sit in, which changes what the first email has to accomplish.

When you displace a competitor, the buyer already agrees the problem is worth money. Your job is preference. When you create a line item, the buyer has to agree the problem is worth money at all, and then find the money inside a budget that was set for something else. Those are two different sales, and the second one is slower even when the product is obviously better.

Practically, that means the cheapest opening is one that borrows an existing budget rather than asking for a new one. Every AI product replaces something the company currently pays for in cash or in hours. Photography, agency retainers, contractor days, overtime, a slow internal process. Naming the specific thing your product comes out of gives the buyer somewhere to take the money from, and turns an interesting demo into an approvable one.

The reverse also holds. If you cannot name what the spend comes out of, you are asking a stranger to invent budget, and the reply rate will show it.

Your claims are read against everyone else’s claims

Every buyer in Europe and North America has spent the last two years receiving mail promising that AI will transform their operation. Most of it was written by people who had not built anything. The result is a category-wide scepticism tax that your message pays whether or not you deserve it.

Three things trigger it immediately. Adjectives without numbers. The word "revolutionise". Any claim of a percentage improvement with no stated baseline. A reader who has been burned twice does not evaluate the claim, they pattern-match it and delete.

What survives is arithmetic the reader can check against their own operation. A cost per unit of output. A throughput figure with the time window attached. A before and after where the before is a number the reader recognises from their own business. If a buyer can do the multiplication in their head and land somewhere plausible, you have passed the filter that most of your category fails.

It also helps to be specific about what the product does not do. A stated limit is the cheapest credibility available, and in a category this noisy it is close to a differentiator.

The price floor decides the motion

Nearly every AI company launches with two ways to buy: a self-serve plan at some tens of euros a month, and an enterprise arrangement with a contract minimum. Those two numbers are usually set by the finance model. They should be set by the sales motion, because the floor decides what outbound can pay for.

The middle two rows are where most AI SaaS sits, and they are where outbound genuinely pays. Below that, the cost of contacting a company exceeds the value of winning it, and the honest answer is that outbound is the wrong channel no matter how well it is executed.

A contract minimum is not a pricing detail. It is the instruction that tells your outbound team how much time they are allowed to spend on one company, and it should be decided before the first list is built.

Running both motions at once is the common failure

The self-serve plan and the enterprise contract look complementary on a pricing page. In outbound they fight.

If a cold email mentions a low monthly figure, the enterprise buyer anchors there and every subsequent conversation is a negotiation down from a number you never intended to offer. If it leads with a contract minimum, the smaller company that would have happily paid the monthly price never signs up, because they read the mail as not being for them.

The resolution is not clever copy. It is segmentation. Decide, before the campaign, which band of company each list belongs to, and let each list see exactly one commercial frame. Self-serve gets found through content, product and search. Outbound is pointed exclusively at the companies large enough to sign the contract you actually want.

The tell that this has gone wrong is a pipeline full of interested small companies and a founder complaining about lead quality. The leads are fine. The message invited them.

Who to write to when the category is new

In an established category there is an owner of the problem, and finding them is a title lookup. In a new category the problem is spread across two or three roles and nobody has been given it, which is why "we sent it to the right title and heard nothing" is such a common complaint.

Three routes work better than a title filter.

Write to the person whose workload changes. Not the person who signs, the person whose week gets shorter. They have the strongest incentive to answer and they will bring you to the signer, which is a much better introduction than arriving cold at the signer yourself.

Write to whoever owns the number you move. If the product affects conversion, that is one person. If it affects cost per unit, it is a different one. The number, not the department, points at the human.

Write to the founder in companies under about fifty people. Below that size the founder is still the decision maker for anything that touches the core operation, and going around them wastes a month.

What proof actually looks like

A logo wall is not proof. A testimonial about how great the team is to work with is not proof either. Proof is a described situation, a number, and enough detail that a sceptic could estimate whether it would hold for them.

The strongest form in this category is a unit economic, because it travels. Cost per output, time per output, or volume per time window. A buyer with a different catalogue, a different market and a different budget can still multiply your unit number by their volume and see what happens. A percentage improvement cannot be multiplied by anything, which is why it persuades less despite sounding larger.

Two practical rules. Use the customer’s own before-number rather than an industry average, because the industry average is always disputed and the customer’s number never is. And name the constraints under which the result held, since a result with visible edges reads as measured rather than marketed.

The metrics that matter before revenue does

An early AI company running outbound for the first time usually watches meetings booked, which is the last thing in the chain and the slowest to move. Three earlier numbers tell you what is wrong while there is still time to fix it.

• Reply rate by segment, not overall. An overall rate averages a segment that works with three that do not, and hides the only useful finding in the campaign.
• Positive-to-total reply ratio. A high reply rate made of polite refusals means the targeting is right and the offer is wrong. A low reply rate with a good ratio means the offer is right and the list is wrong. These two failures look identical on a dashboard that only counts replies.
• Meeting-to-opportunity conversion. If meetings happen and nothing progresses, the message is selling something the product does not do, and no amount of extra volume repairs that.

Give it a full cycle before judging. Outbound into a new category takes longer to read than outbound into an established one, because part of what you are measuring is how long a buyer takes to accept that the problem is real.

When outbound is the wrong tool

Three situations where the honest answer is not yet.

The product cannot be described in one sentence a stranger understands. Outbound amplifies clarity and punishes its absence, and no sequence design compensates for a value proposition that the founder is still testing.

The contract value is under a couple of thousand euros a year with no expansion path. The arithmetic does not work, and doing outbound anyway just converts runway into a list of unqualified conversations.

There is no delivery capacity behind the pipeline. Booking meetings you cannot service is a way of burning a market you will want later. Fix delivery first, then open the tap.

If none of those apply, outbound is usually the fastest way for an AI company to find out whether the category it invented has buyers in it. We run the engine end to end, tuned to the contract size you are actually selling, and forward the interested replies to you.