Your Forecast Should Be More Than a Lookup Table

Open an opportunity in NetSuite and one of the fields sitting on it is probability. Oracle's documentation calls it "the most critical piece of pipeline information on an opportunity record," and that is not marketing language, it is mechanical. The weighted amount is the probability multiplied by the projected total. The pipeline total is the sum of those weighted amounts across every open opportunity and estimate. That total is the number your company plans against, so it is worth knowing where the percentage comes from.

It does not come from the deal. It comes from the stage. A customer status in NetSuite is its own record, carrying both the stage a deal is in and the probability that goes with it, and Oracle's documentation is explicit that the interface calculates the probability from the status rather than from anything about the opportunity itself. Move a deal to a new stage and the number moves with it.

The weights themselves are usually reasonable, too. A deal where the proposal is in the buyer's hands genuinely is closer to closing than one where a first meeting is on the calendar, and pricing those at seventy percent and fifteen percent is a fair reading of how deals tend to go. The model works. It turns a pile of unrelated conversations into a single number using the one thing the system always knows for certain, which is what stage each deal is sitting in.

The catch is what the model cannot see. Two opportunities in the same stage contribute exactly the same weighted amount, whether the champion is picking the date for the next meeting or stopped replying in June. Both proposals went out. Only one of them is still alive. The forecast has no way to tell them apart, because the only thing it knows about either deal is which bucket it is in.

Every CRM worth using has priced pipeline this way since the beginning, and that is not a failure of imagination, it is a good starting model that has held up for decades. There are two ways to get more out of it. The first is worth an afternoon of your time: go and look at your stage weights. Most teams either inherited theirs during implementation or simply kept whatever the system shipped with, and have never once checked either against their own closed-won history. If the stage where the proposal is out has been converting at forty percent for two years and the status record still says seventy, the forecast is not wrong because of your deals. It is wrong because of your table. Vendios makes those weights straightforward to change, and they should be revisited more than once.

The second is the part no amount of tuning will fix. A weight describes a stage. It cannot tell you whether a particular deal still deserves the stage it is in. That is a judgment about evidence, it has to be made deal by deal, and it has to be made continuously, across hundreds of deals, which is precisely why it never gets made.

Which is why Vendios is built AI first rather than AI eventually. Every call and every email is transcribed and summarized onto the NetSuite record as it happens, so the account history gets written by the act of selling rather than after it. That is the unglamorous half, and it is the half everything else depends on.

Then the AI reviews the deals every morning, the way a manager would if they had read everything and forgotten nothing. It is not there to rewrite your stage weights, those are yours. It is there to hold each deal accountable to the stage it is claiming.

It has the timestamps, and the timestamps are the easy part. They will tell you a deal has gone quiet. They cannot tell you why, and silence is not one thing. A buyer who mentioned on the last call that they are out until the third is quiet for a reason you already know about, and nobody needs to be chased about it. A buyer who has not answered the email you sent nine days ago is quiet for a completely different reason, and that one is worth a phone call this morning. As a gap between two dates those are identical. Read what was actually said and they stop being identical, and the observations start being worth something: the buyer went quiet right after asking about pricing, the security review promised three weeks ago never arrived, this deal is sitting at proposal with an expected close date it is not going to make. None of that is available to an AI working from the dates alone. It will still answer you, confidently, in better grammar than the rest of us and probably with an em dash or two.

All of that lands on the NetSuite records themselves, not a parallel database with a sync job in between. The probability field, the weighted roll-up and the forecast reports stay exactly where they are and keep doing what they have always done well.

The lookup was always a reasonable way to price a pipeline. It was also always the easy half. Which bucket, times how much, works fine as long as something is checking that each deal still deserves the bucket it is in, and for most of the history of CRM nobody has had the hours to do that checking. Now something does it every morning, before anyone opens the forecast.