Decisioning Is a Lending Word. Here's What Your Martech Vendor Left Out.

Decisioning is the system that automatically picks the next move for each customer, like which offer or message they get. Martech borrowed the word from lending, where a decision has to be explained by law, and left that requirement behind.

Every big martech platform sold today makes calls about your customers without asking you first. Which offer this person sees. Which email goes out, and when. Whether someone gets pulled aside as a high-value lead. The software decides, thousands of times a day, faster than any team could sit and choose.

Vendors have a word for that: decisioning. You’ve heard it in demos and analyst decks, and most people nod along rather than ask what it means. Worth knowing before you sign anything: the word didn’t start in marketing. It’s borrowed. And the part that got left behind in the borrowing is the part that should decide your purchase.

The word comes from lending

In banking, decisioning has meant one thing for decades. An automated yes or no on credit. Approve this loan, decline that card, set this limit. A decision engine takes an application, weighs it against rules and models, and returns an answer in seconds.

Lending’s version differs from marketing’s in one place: the leash. When a lender turns someone down, U.S. law requires them to give the specific reasons: too little credit history, too much existing debt, income under the line (1. CFPB, n.d.). A vague answer like “you didn’t qualify” or “the model said no” is a violation.

The black box gets no pass either. When a lender leans on a model it doesn’t fully understand, regulators have been blunt: a creditor’s lack of understanding of its own methods “is not a cognizable defense” (2. CFPB, 2022). In lending, if you can’t explain the decision, you’re not allowed to make it.

What martech kept, and what it dropped

Martech took the word and the promise. A decisioning platform now means software that picks the next move for each customer. The offer, the message, the moment, the channel. In a CDP or a journey orchestration tool, it shows up as next-best-action: the system reads a customer’s profile and behavior and chooses what happens next, on its own. Same core idea as the loan engine, an automated call made at scale on the data you feed it.

Then it dropped the leash. Nothing requires a martech decisioning system to explain why it served this person that offer. No law, no notice, no specific-reasons letter. In lending, a wrong decision gets explained, documented, and can be contested. In martech, a wrong decision quietly costs you a conversion, and nobody ever knows it happened. The vendors kept the confident half of the word and retired the accountable half.

You can see the word’s home by looking at who the analysts file under it. The vendors ranked in Forrester’s decisioning platform evaluation come from credit, risk, and enterprise decisioning, not marketing: FICO, Provenir, Pega, IBM (3. Forrester, 2025). Marketing borrowed a term rooted in credit risk, then put it to work with none of credit risk’s rules.

Why the missing piece is what the demo runs on

Here’s where it bites, in the room. In the demo, the platform decides “show this customer that offer,” and it looks smart. You have no way to check whether it was the right call. Grading it would take knowing the customer, the full context, and what happened afterward, and you have none of that. So you grade on how sharp it looked instead of whether it worked.

That gap is the whole event. A decision you can’t audit gets judged on polish and confidence, and the demo is engineered to supply both. The demo does something quieter than lie. It fills the space where proof should sit with the next best thing, which is belief.

Buyers feel this even when they can’t name it. In a survey of 350 B2B buyers, 27% pointed to vendors leading with features instead of business impact as a top frustration (4. Adience, 2025). The demo shows the machine working. It rarely shows the machine being right.

The questions the word’s own history hands you

Skip the longer RFP . The move is to make the vendor do what a lender has to do by default: explain a decision after the fact.

Ask to see one real decision the platform made for a real customer six months ago, with everything that fed into it. Ask how they retire a model that’s gone stale, and who signs off when it does. Ask what the system could tell you if you demanded the specific reasons behind a single call it made yesterday.

If the product can answer those, you’re buying decisioning in the full sense of the word. If it only answers inside the rehearsed demo, you’re buying a word with the spine taken out. The word carried a duty to explain itself when it was born. Your job as the buyer is to check whether it still does.