Garbage In, Garbage Out: Fix Data at the Source

There’s a story Philip Richard tells about one of his lease operators from his days as a young production engineer. The guy would get out to the field at six in the morning, come home around four, and then spend three hours on his computer entering reports. Three hours. When he should have been at the dinner table with his family, he was copying numbers into spreadsheets that somebody in the office would barely glance at before pulling them into another spreadsheet.

That’s not a technology problem. That’s a design problem. And it’s one of the most common and most expensive problems in the oilfield today.

Philip is the CEO of Porosity, a workflow automation company built specifically for field operations in oil and gas. Before that, he was a petroleum engineer. Worked across most of the major shale basins. Wrote work over procedures. Maintained production. Sat in the same seat as the people he’s building software for now. And the thing that comes through in every conversation with him is this: the data problem in oil and gas doesn’t start in the server room. It starts at the wellhead.

So here’s the question worth sitting with: If your data is only as good as the process that captures it, how much are you really spending to work around bad inputs?

The Real Cost of “Just Get It Entered”

Philip’s CTO spent twelve years at AWS before founding Porosity. The man was dealing with more bytes of data than seconds since the Big Bang. And after all that experience with massive data sets at the highest level, the principle he brought into the oilfield was deceptively simple: garbage in, garbage out.

Everyone in this industry has heard that phrase. But very few companies have actually done something about it where it matters most, which is at the point of entry. In the field. On the lease. With the operator who’s tired and trying to get home.

What Philip described on the podcast is something most of us have seen firsthand. Operators collecting data on paper or in cluttered spreadsheets. Leak inspections getting communicated through text messages and emails. Third-party companies trying to flag a problem to an operator through a chain of phone calls. And at the end of the year, someone in the office is trying to stitch all of that together into a regulatory filing.

The data gets entered, sure. But nobody trusts it. And when nobody trusts the data, nobody trusts the decisions that come from it.

Make the Right Entry the Easiest Entry

Philip made a point on the show that stuck with me. He said the easiest way to prevent garbage data from coming in is to make the entry and the workflows around it as simple as possible. Not simple for the office. Simple for the person holding the phone in the field.

Porosity built their mobile application to feel like the everyday apps people already use. Philip compared it to something like Uber Eats or DoorDash, but tailored for the oilfield environment. That’s not a gimmick. That’s a design philosophy. When the tool feels familiar, the operator doesn’t fight it. They step through the workflow, capture the data, and move on. The data comes in clean because the process made clean the path of least resistance.

That’s a lesson that goes well beyond software. Whether you’re building a reporting system, a safety protocol, or a maintenance workflow, the principle holds: if you want better data, don’t lecture the field. Design a better process.

What Clean Data Actually Looks Like in the Real World

Philip didn’t just talk theory. He shared real numbers from real customers. And the results tell a story that every operator and every engineer in this industry should pay attention to.

One of Porosity’s customers was running their LDAR program the old way. Inspections recorded in spreadsheets. Leaks communicated through emails and text messages. Repairs assigned through a chain of calls and follow-ups. Their average time to repair a leak was about a week and a half. Philip did the math on what that costs. On an average leak of ten to twenty MCF a day, that’s roughly eight hundred dollars a day just bleeding into the air while someone shuffles paperwork.

When they brought everything into the Porosity system, consolidating data into one place and connecting leak detection directly to maintenance alerts, that average repair time dropped from a week and a half to a single day. Errors went down ninety percent. Time spent on data entry dropped fifty percent. And when you extrapolate that across a year, that fifty percent reduction added up to roughly sixteen thousand man-hours of pure data entry that simply went away. Reporting that used to take a month was done with a single click in a day.

Those aren’t projections. Those are real outcomes from cleaning up the source.

AI Doesn’t Fix Broken Inputs. It Amplifies Them.

One of the most honest things Philip said in our conversation was this: you can’t just slap AI on something and expect it to work. He and his CTO share that conviction, and they’ve built their entire product around it.

Philip talked about how, a decade ago, companies were hiring data analysts to study shale data. They were looking for correlations between how much proppant they pumped, how long they drilled their laterals, and what production looked like. And most of the time, the answer was nothing. No correlation. Because the data was coming from different vendors with no cohesion. Variables were being changed four at a time. Well pacing wasn’t even being accounted for. The analysis was dead before it started.

His point was clear: AI has to start with the outcome you’re trying to achieve, and then it needs clean, aggregated, human-verified data to work from. Not random data sets. Not sensor feeds with no context. Structured, validated data that a real person confirmed before the model ever touches it.

That’s where Porosity is building its long game. Every day, their platform collects thousands of human-verified data points from the field. Philip’s vision is that this data set becomes the foundation for predictive AI down the road, layered eventually across IoT sensors, pumping units, ESPs, gas compressors. But the foundation has to be there first. You can’t skip the hard part.

Philip Richard’s Advice for Operators and Engineers

Philip wasn’t just talking about Porosity on the show. He was talking about a shift in how this industry needs to think about the relationship between the field and the office. Here’s what stood out.

Stop trying to clean data in the office that was broken in the field. The fix has to happen at the source. If the entry point is clunky, complicated, or disconnected from the actual work, the data will always be unreliable. Make the tool so simple that the right entry is the only natural path.

Listen to the field. They’ll tell you what’s broken. Philip said his sales process often starts with the production superintendent or foreman who’s tired of whipping out a laptop in the truck to enter data after a full day in the field. Those people know exactly where the friction is. If you’re not asking them, you’re guessing.

Don’t chase AI before you have the data to support it. Philip was refreshingly honest about this. AI is the biggest buzzword in the investor world right now. But in oil and gas, the real value of AI is doing more with less, not replacing people. And that only works when the data underneath it is trustworthy.

Regulations aren’t going away. Build for that reality. Regardless of what happens at the federal level, Philip pointed out that states are requiring more and more inspections. Colorado, New Mexico, Wyoming, North Dakota, Pennsylvania. The compliance burden is growing, and operators who can generate reports with a single click instead of stitching spreadsheets together will be the ones who survive low commodity environments.

Think about the person, not just the process. Philip kept coming back to quality of life for the field operator. If the guy in the field is spending three hours at home entering reports instead of being with his family, the system is failing him. The best technology in this space doesn’t just capture data. It gives people their time back.

Final Thought

There’s a tendency in this industry to throw money at dashboards and analytics platforms and AI tools and expect better decisions to follow. But Philip Richard’s message is simpler and harder to argue with: none of that matters if the data going in is broken.

Fix the source. Respect the field. Design for the person doing the work, not the person reading the report. The rest follows.

That’s not a software pitch. That’s just good operations.

If this question hit home, you’ll want to hear how it plays out in real operations.

Join Philip Richard, CEO of Porosity, on Wisdom at the Wellhead as he breaks down how clean field data, simple workflows, and human-verified inputs are changing the way operators run leaner, faster, and smarter.