five signals.
one week.
you automated the part
they judge you on._

voice and video used to be identity checks, and most payment approvals were built on the assumption that they still are. the cost is one successful payee change, which in a company under fifty people is usually a five or six figure transfer that clears before anyone thinks to call anybody back.

the fix being sold is another security awareness module telling staff to listen for something slightly off. that parks the whole defense inside a junior employee's ear at 4pm on a friday while a senior voice tells them it is urgent. the vendor gets a renewal. you get a certificate.

the real fix is procedural and boring. any new payee, or any change to bank details on an existing one, needs a second named approver and a callback to a number already on file, never the number in the request. write it into the payment step so rank and urgency cannot waive it, because rank and urgency are exactly what the attack manufactures.

a competent first draft now costs nothing, so every outbound email in your market is competent, and they all read the same. same structure, same cadence, same tidy three-line value proposition. your buyer has seen four of them since breakfast.

the cost is not that your email is bad. it is that it is interchangeable, and interchangeable gets filed by pattern instead of read. a real offer dies next to four fake ones. the fix being sold is more volume through a better sequencer, plus a personalization token that drops the company name into a sentence nobody wrote. the course that teaches you to send ten thousand of these calls that scale. it is the same email in a wig.

the fix is narrow and it is work. line one is a sentence only your company could write, about something you actually saw in their business, in words a machine would not have guessed. the rest can be machine-assembled, because the rest was never doing the work. run the swap test on your last three sends this week. whatever survives the swap gets rewritten first.

a whole category sells finished revenue to owners too busy to audit the mechanism. the calls get booked, the leads get worked, the pipeline fills, and somewhere near the end there is a dashboard. ai gave that category a new coat of paint and a new name every quarter.

the subscription is the small loss. the thing you outsourced was your first impression, and a prospect who works out that a machine qualified them while pretending otherwise does not come back for the human version. the fix most owners reach for is a different vendor with a better website, which is how the same script keeps returning under new logos. the seller always shows the earnings number before the mechanism. that order is the tell.

before money moves, ask three questions. what model is under this. what data was it trained or tuned on. give me a customer i can call without you setting up the call. a vendor who built something answers all three in a sentence each. a vendor reselling something shows you the dashboard again. if the growth claim is specific and the explanation is vague, you are not the customer. you are the deck.

support volume is mostly routine, the routine part automates cleanly, and on a spreadsheet the headcount math looks finished. what the spreadsheet does not carry is that customers judge you almost entirely on the remainder: the billing error, the thing that broke twice, the refund just outside policy.

klarna's satisfaction fell on exactly those cases. the cost arrives two quarters later as churn that gets blamed on price, followed by a rehire at a new title and a higher rate. the fix being sold for this is an escalation button, which routes the angriest customer into a queue and tells them their call is important to us.

automate the routine and staff the remainder on purpose. route by complexity, not by volume, and keep a named person on anything involving money, an error you caused, or a customer asking for the second time. the org chart will call the rehire a new role. the customer will call it the third time they explained this.

for two years the sellable object in ai services was a front end. wrap a model, put a logo on it, charge monthly, and the buyer could see what they paid for. that window is closing, and the spending data is where it shows first.

the chat window is the line a cfo can picture living without, so it is the first thing cut in a budget review. the fix being sold is more features on the same screen, which raises the price of the thing they already planned to cancel.

sell the wiring. scope the connection into their crm, their billing, their inbox and whatever they treat as the record, and price it as the product rather than as onboarding you discount to close. integration is expensive to install and more expensive to remove, and that is the only stickiness that survives a cost review. the interface is what they buy with. the connections are what they keep.