why we built jemma

Every optimization leader we’ve worked with has been asked the same question, “Yeah, but what’s the ROI of optimization?”

This usually gets asked by some loud VP or sometimes by a CFO or occasionally by a board deck that needs a number by Friday afternoon and it’s Friday morning.

And every single time, the answer starts the same way. Somebody opens a spreadsheet, someone else is asked to pull a report from Adobe Target. Somebody asks the analytics team to produce results, which takes three days but usually they can’t get to it until next month. Somebody digs through JIRA to remember what the test was actually measuring. Someone else tries to sort the real tests from the QA builds and the adopted winners and the personalization rules that got labeled as tests, gives up, and counts everything. Somebody drops it into a slide because it’s due Friday.

The answer that ends up in the VP’s inbox is, essentially, a reconstruction. Half the context lived in someone’s head. A quarter of it lived in a deck from last year. The rest lived in four different platforms that don’t talk to each other.

Hila and i have spent a combined 30 years running optimization programs for major brands. Retail, healthcare, travel, media, financial services. Different stacks, different team sizes, different maturity levels.

Yet the question was never different.

And the machinery behind the answer was never different either. Manual. Time-intensive. Rebuilt from scratch every quarter. The real intelligence lived in a few people’s heads, and when those people left, so did the program’s memory. And sometimes the programs died.

The problem we discovered is that every platform in the optimization space is built around the individual test. Build it, launch it, read the result. These are good tools. They do what they do.

But the program is a different thing. It has a portfolio. It has aggregate learnings. It has intelligence that only reveals itself when you look across a year of tests and ask “what did we actually learn, and what did it actually earn us?”

Nobody manages that. At least not well.

The test results live in one platform. The revenue context lives in another. The hypothesis and business framing lives in JIRA. The learnings from the last three quarters live in slide decks. The institutional wisdom about which tests mattered and why lives in someone’s head, and that someone might leave next quarter.

Running an optimization program well is a cross-platform, cross-team, cross-time intelligence problem. Every tool currently being sold for it is a single-platform, single-test, single-moment point solution.

i guess that’s why we decided to build jemma. It’s built for practitioners who already know what they’re doing. It doesn’t try to replace expertise. It doesn’t automate strategy. It doesn’t pretend to know the business better than the person running the program. It takes on the parts nobody has time for, quietly, in the background. Catalog grooming. Cross-platform data orchestration. Statistical rigor. The program-level intelligence that makes the VP’s question answerable without a three-day fire drill.

We care about the positioning, because most of what’s currently being marketed as “AI for optimization” gets it exactly wrong. It talks down to practitioners. It promises to replace the judgment that’s the most valuable part of the job. It confuses answering questions with doing the work.

jemma does the opposite. It respects that the practitioner is the expert. It works behind the scenes. It surfaces intelligence instead of instructions. It makes an experienced team more effective, not less necessary.

one more thing

We’re in Vegas this week for Adobe Summit, showing jemma in a private suite to a small handful of people who already live the problem. The feedback has been the kind of thing you remember for a while. Every demo has produced some version of the same reaction, which is roughly this is the thing i’ve been trying to build for my team out of spreadsheets for years.

Which tracks. Because that’s exactly the thing we got tired of watching people try to build.