There's an industry around selling AI to small and medium-sized businesses, and most of it is rotten.
This isn't a criticism of every AI vendor. A lot of them sell genuinely useful products. The problem is that the buyer almost never has the context to tell which is which, and the sales playbook is specifically designed to keep it that way. The result is that SME owners regularly sign contracts they don't understand, for outcomes that won't materialise, with vendors who knew it at the time.
What follows is the buyer's-side handbook. The seven moves the AI vendor playbook uses on every owner-founder it can get in front of, the question that exposes each one, and a scorecard you can score the next pitch against in real time. If you've sat through more than two AI pitches in the last year, you'll recognise most of these. That's not a coincidence. It's a template.
Why is the AI sales playbook so aggressive in 2026?
Three reasons worth knowing before you take another meeting.
Start with the market. The AI software space in 2026 is brutally competitive. There are thousands of products chasing the same SME budget, most of which are wrappers around the same underlying models. The differentiation is in the marketing, not the technology, which means the sales effort has to do a lot of work the product can't.
Now look at the buyer. SMEs are an easier sell than enterprises. Enterprise buyers have procurement teams, vendor management functions, and contract lawyers. The owner of a 25-person business has none of that. The same pitch that would be torn apart in a corporate setting walks straight into an SME boardroom and signs the same week.
Finally, the time horizon. Nobody in this market has long-term reputational pressure. Most AI vendors selling to SMEs in 2026 won't exist in 2028. They know that. The incentive isn't to build customer relationships that compound over five years. It's to close as much business as possible before the music stops.
The result is a set of sales behaviours that look reasonable on the surface and are aggressive underneath. The seven moves below are the ones you'll see again and again.
What are the seven moves you'll see in every pitch?
Move 1: The fake demo
How it works. The vendor walks you through a polished demo that shows their product doing something impressive in real time. It looks live. It's not. What you're watching is either a pre-built scenario that's been rehearsed to the millisecond, or a recording, or a sandbox environment that has none of the constraints of a real deployment.
Why it works. SME owners don't know what to ask. A live demo feels like proof, even when it isn't.
The question that exposes it. "Can I see you do that with my data, on a problem I bring to the meeting, right now?"
If the answer is anything other than "yes, give me five minutes," you've just watched a marketing video.
Move 2: The vanity metric
How it works. The vendor cites a productivity number that sounds enormous and specific. "Customers see a 47% reduction in admin time." "Teams using our platform are 3.2x more productive on reporting." The number is meaningless because the denominator is never defined and the sample is never disclosed.
Why it works. Numbers feel like evidence. Specific numbers feel like rigorous evidence.
The question that exposes it. "How many customers is that based on, over what time period, and how did you measure the baseline?"
If they can't answer all three parts, the number is a sales prop, not a finding. The honest version of that claim is almost always "our best customers report saving time on this task."
Move 3: The bundled feature list
How it works. The pitch walks through a list of fifteen capabilities the platform offers. By the time the list is finished, you're slightly overwhelmed and have lost track of which ones matter for your business. The vendor moves on to pricing before you can ask which features you'll really use.
Why it works. A long feature list signals seriousness. It also obscures the fact that 80% of the features are irrelevant to your business.
The question that exposes it. "Of the features you just walked me through, which three account for most of the value your customers get?"
A good vendor will answer this honestly. A bad one will pretend all fifteen are equally important, which is impossible.
Move 4: The fear close
How it works. Somewhere in the pitch, usually after the demo and before the pricing, the vendor introduces urgency. Sometimes it's competitive ("your competitors are already using this"). Sometimes it's temporal ("we're locking in 2026 pricing this quarter"). Sometimes it's existential ("businesses that don't adopt AI in the next 12 months will struggle to compete").
Why it works. Pressure overrides judgement. An owner who'd have asked three more questions in a calmer setting signs in the room.
The question that exposes it. "If I take this decision in March instead of January, what specifically will I have lost?"
If they can't answer concretely, the urgency is a sales tool. Real urgency has a specific number attached. Manufactured urgency has feelings attached.
Move 5: The integration handwave
How it works. The vendor confirms that yes, of course, their platform integrates with your existing systems. They show you a logo grid of every major business tool ever invented. They do not show you the integration working.
Why it works. Integration is the part of the buying decision that owners worry about most, so a fast reassurance is welcome.
The question that exposes it. "Can you connect this to my [specific system] in a setup call with me on the line, before I sign anything?"
The gap between "we integrate with Xero" and "we have a working integration with the version of Xero your accountant is on, in your country, with your chart of accounts" is enormous. A logo on a slide is not an integration. A vendor who won't demonstrate it before purchase is hiding something.
Move 6: The reference theatre
How it works. When you ask for customer references, the vendor produces a glossy case study or two, possibly with a video testimonial. The customer in the case study is usually three times your size, in a different sector, and was their first major customer. The story is real, the relevance to you is not.
Why it works. A case study feels like a peer endorsement. The fact that the peer is in a different industry, at a different scale, with different problems is easy to miss when the testimonial is well-shot.
The question that exposes it. "Can you give me the contact details of two of your customers in my sector, my size, who've been live for at least six months?"
If they can, fantastic. Call both. If they can't, they don't have the customer base they're claiming, or their customer base isn't satisfied enough to be put forward, or both.
Move 7: The implementation fog
How it works. The pricing slide shows the licence cost. What's missing is everything that comes before the licence starts producing value. Setup, training, data preparation, custom workflow design, integration work, the time your team will spend learning the tool. The vendor is selling the licence. You're buying a project.
Why it works. The licence price is what you compare across vendors. The implementation cost is what determines whether you get value, and it's never on the slide.
The question that exposes it. "What's the total cost of getting this fully operational across my team, including setup, training, and the first three months of internal effort?"
If the vendor's answer doesn't include a number for internal effort, they've never thought about it from the buyer's side. The honest answer is usually 1.5 to 3 times the licence cost in the first year.
How do these moves combine in a real pitch?
A typical SME-targeted AI pitch in 2026 runs roughly like this. The vendor opens with a five-minute story about AI transformation (vanity metric in the first 90 seconds). They show a demo (the fake demo). They walk through 15 features (the bundled feature list). They drop in a case study from a much larger customer (reference theatre). They handwave the integration question (integration handwave). They land on a pricing slide that excludes implementation (implementation fog). They close with urgency (the fear close).
If you saw a pitch that did all seven of these things in 45 minutes, you'd recognise the pattern as a sales process rather than a partnership conversation. The problem is that you usually only spot it afterwards. Hence the scorecard.
This isn't a claim that every AI vendor uses every one of these moves. Some are honest. Some have products that deliver real value. The point of the scorecard isn't to refuse to buy AI software. It's to make sure that when you do buy it, you're buying for the right reasons and from a vendor whose pitch has earned your trust. A vendor who scores well on the seven questions below is one worth working with.
The AI vendor pitch scorecard
Score each move from 0 to 2 during or immediately after the pitch.
| Score | What it means |
|---|---|
| 0 | The vendor did this thing and didn't address it when challenged. |
| 1 | The vendor did this thing but had a credible answer when challenged. |
| 2 | The vendor didn't do this thing at all, or addressed it pre-emptively. |
The seven moves to score:
| # | The move | The question | Score |
|---|---|---|---|
| 1 | The fake demo | Did the vendor show their product working live, on real data, on a problem you brought to the meeting? | ____ |
| 2 | The vanity metric | Were the productivity claims backed by named customers, defined time periods, and measurable baselines? | ____ |
| 3 | The bundled feature list | Did the vendor name the two or three features most of their customers genuinely use? | ____ |
| 4 | The fear close | Was any urgency in the pitch tied to a specific, concrete cost of waiting? | ____ |
| 5 | The integration handwave | Did the vendor offer to demonstrate the specific integration you need before any contract is signed? | ____ |
| 6 | The reference theatre | Did the vendor offer references in your sector, at your scale, who've been live for at least six months? | ____ |
| 7 | The implementation fog | Did the vendor give you a credible total-cost figure including setup, training, and internal effort? | ____ |
Add up the score. Out of 14.
| 12–14 | A vendor worth working with. The pitch was honest, the answers were grounded, and the relationship is starting from the right place. |
| 8–11 | A vendor with a real product whose sales process is more aggressive than the product warrants. Negotiate hard, demand specifics in writing, and don't sign in the room. |
| 4–7 | A vendor who's prioritising the sale over the customer relationship. Probably has a product that works for some customers. Probably won't work as well for you as the pitch suggests. Walk away unless you have a very specific reason to engage. |
| 0–3 | A vendor pitching a product they don't expect you to keep beyond the initial contract. Walk away. Do not return calls. |
So what should I do at my next AI vendor meeting?
Three things.
Before the meeting, print the seven questions above on a single page and have them in front of you. The vendor's slides will be designed to keep you off-balance. The questions will keep you on it.
During the meeting, ask the questions explicitly. Don't be polite about it. A vendor whose product is genuinely good will welcome the rigour. A vendor whose pitch falls apart under direct questioning has just told you everything you needed to know.
After the meeting, before any decision, score the pitch using the scorecard. If you can't do it from memory, the pitch wasn't memorable enough to be worth signing for. Sleep on anything that scores below 12. Do not, under any circumstances, sign in the room.
The AI vendors selling to your business in 2026 are running the same playbook on every SME owner they meet. Once you know what it looks like, it stops working on you. That's the whole point.
Take the free AI Readiness Diagnostic
The Tools & Tech dimension of the diagnostic gives you a structured view of what you're already paying for, where the genuine gaps are, and where you should and shouldn't be talking to vendors next. It takes around 15 minutes and the report comes back within one working day.
Take the diagnostic →