--- url: 'https://dbseer.com/data-dispatch/data-dispatch-november-2026/' title: 'Data Dispatch: November 2026 ' author: name: Mahyar Ghasemali url: 'https://dbseer.com/author/mahyar/' date: '2026-11-07T15:17:46-05:00' modified: '2026-11-07T15:17:48-05:00' type: post categories: - Data Dispatch image: 'https://dbseer.com/wp-content/uploads/2026/11/Slide1.jpg' published: true --- # Data Dispatch: November 2026  *Let this be your one-stop shop for all things data and AI. Our team at dbSeer is tracking trends, watching the market, and exploring what’s actually changing in how organizations build with data, so you don’t have to do it alone. If you’re not already subscribed on LinkedIn, sign up to have the Data Dispatch delivered to your inbox each quarter.*  Table of Contents - [What We’re Watching: AI Without a Data Foundation Is Instagram Without the Internet ](#What_Were_Watching_AI_Without_a_Data_Foundation_Is_Instagram_Without_the_Internet) - [What We’re Building: The Foundation Starts on the Front Line ](#What_Were_Building_The_Foundation_Starts_on_the_Front_Line) - [What We’re Working On: Your Ad Platform Is Already an AI. What Are You Teaching It? ](#What_Were_Working_On_Your_Ad_Platform_Is_Already_an_AI_What_Are_You_Teaching_It) - [What’s Next: On Our Radar ](#Whats_Next_On_Our_Radar) - [dbSeer Achieves Select Partner Tier in the Claude Partner Network ](#dbSeer_Achieves_Select_Partner_Tier_in_the_Claude_Partner_Network) - [dbSeer Joins The Franchise Consortium as an Accredited Supplier ](#dbSeer_Joins_The_Franchise_Consortium_as_an_Accredited_Supplier) - [dbSeer Joins the 2026 Fitness Technology Summit ](#dbSeer_Joins_the_2026_Fitness_Technology_Summit) ## **What We’re Watching: AI Without a Data Foundation Is Instagram Without the Internet**  Every leadership team we talk to is asking some version of the same question: how do we get AI working for us? It’s the right question. But it skips a step. Before anyone asks what AI can do for the business, someone has to ask what that AI is going to stand on.  We recently met with a data and AI leader at a global technology company, someone whose organization is about as AI-forward as they come. You might expect him to spend his days on models and agents. Instead, he told us he still puts 70 to 80 percent of his time and resources into data foundations and building out his data practice. In his view, AI is the benefit that sits on top of a solid platform. It is not a shortcut around one. Then he put it in a way we haven’t stopped thinking about: *AI without a data foundation is like Instagram without the internet*. The app can be beautiful, fast, and full of features. Without the network underneath it, it does nothing at all.  The research points the same way. In an [industrial example](https://www.youtube.com/watch?v=06-AZXmwHjo), several teams tried different model architectures on a steel defect detector and couldn’t move it past 76.2 percent accuracy, but improving the data raised it to 93.1 percent. The team behind the [Falcon language models](https://arxiv.org/pdf/2311.16867) saw something similar in their own experiments: filtering and deduplicating training data made the models perform better, while their architecture changes mostly made the models more efficient to train and run. Improving the data can deliver gains that model changes alone won’t.   [Qualcomm researchers](https://arxiv.org/html/2510.02243v1) made an AI assistant more accurate without ever touching the underlying model. They only changed how its source documents were prepared. That’s the part of AI most organizations overlook: an assistant is only as good as the data it searches, and the more it knows about your customers, your locations, and your definitions, the more that data matters. So, before you buy an AI pilot, ask what it’s standing on, and make the time to clean it up.   ![](https://dbseer.com/wp-content/uploads/2026/11/image-1.png) ***Where does your organization sit on the curve?*** *Our free Data + AI Maturity Assessment takes about two minutes. Answer six questions and get your stage, your biggest gaps, and a 90-day plan to move up.* [*Take the assessment here.*](https://assess.dbseer.com/)  ## **What We’re Building: The Foundation Starts on the Front Line**  ***Picture it:*** **It’s 4:50 on a Friday, and a field technician has just fixed a cooling unit at a customer’s site. The work order asks what failed. He types “fixed,” closes the ticket, and heads home. The customer is happy and the job is done.  A year later, the company launches the predictive maintenance model it has been planning for months. The model is supposed to learn which parts fail, how often, and under what conditions. It finds thousands of tickets that say “fixed,” “resolved,” and “done,” and almost nothing it can learn from. The AI isn’t the problem. The history it was meant to learn from was never written down.  That’s what a *“data foundation”* looks like in practice. IT doesn’t build it in a server room. It gets built one record at a time, by the people and business processes closest to your customers and equipment, usually at the end of a long day, and usually by someone who was never told why the field was there.  Now: multiply that by every technician, every site and every shift. That’s how a good AI strategy gets quietly undermined before it ever launches.  These small gaps are more common than most leaders assume. In an exercise reported in [Harvard Business Review in 2017](https://hbr.org/2017/09/only-3-of-companies-data-meets-basic-quality-standards), 75 executives each reviewed 100 records their own departments had recently created. On average, 47 percent contained at least one critical error.  Those errors get more expensive the longer they travel. Under the 1-10-100 rule, a quality principle originally introduced by George Labovitz and Yu Sang Chang (and the concept previously discussed in our [blog](https://dbseer.com/blog/why-most-ai-investments-underperform-and-what-those-who-succeed-do-differently/) here) it costs about $1 to get a record right when it’s entered, $10 to find and fix it later, and $100 when nobody catches it and the error flows into decisions. The extra minute it takes to fill in the work order is the cheapest return on data a business will ever get.   That’s why, in the age of AI, every employee is a data steward, whether they know it or not. Stewardship isn’t a new role or a new layer of process. It’s a shift in mindset: understanding what a field means, who depends on it, and why getting it right matters downstream. And where front-line teams change often, it can’t depend on a single training session. It has to be built into onboarding and into the work itself. Culture sets out the habits, and [DataOps](https://dbseer.com/service/dataops-foundation/) keeps them in place, with shared definitions, automated checks, and feedback loops that catch gaps before they reach a dashboard or a model. The organizations doing this well start small. They pick the handful of fields that matter most, define each one plainly, name an owner, and show the data back to the people who create it so they can see what it makes possible. That last step is the one most companies skip, and it’s the one that changes behavior. People take care of data when they can see it working for them.  ***Picture it again:*** ****Same technician, same Friday at 4:50. This time, he knows the work order is what the maintenance model learns from, because he has seen the report that shows which parts fail most across the fleet. He picks the part that failed, notes the cause, and closes the ticket in under a minute. A year later, the model has thousands of tickets like his to learn from, and it can flag which units are likely to fail before the customer ever calls.   *Now run the numbers.* Take a service business with 50 technicians, each closing 30 tickets a week. That’s about 78,000 work orders a year. If one in five closes with nothing more than “fixed,” that’s 15,600 incomplete records. Under the 1-10-100 rule, getting them right on the work order costs about $15,600. Finding and fixing them later costs about $156,000. Letting them flow into maintenance plans, parts orders, and forecasts costs about $1.56 million. Same business, same technicians, same minute at the end of the job. The only difference is whether someone knew *why* the field mattered.   Multiply that by every location, every shift, and every employee, and you have a data culture: the foundation every AI strategy should stand on.  ## **What We’re Working On: Your Ad Platform Is Already an AI. What Are You Teaching It?**  Most marketing leaders don’t think of it this way, but if you run paid search, you already hand major budget decisions to an AI every day. Your ad platform’s bidding algorithm decides who sees your ads, when, and how much you pay. The question isn’t whether you’re using AI in marketing. It’s what you’re teaching it to optimize for.  For one of our clients, a home services brand in a large franchise system, the answer used to be form fills and phone calls: less than ¼ of those turned into customers. dbSeer connected the brand’s CRM to its ad platform so real outcomes flowed back to the algorithm: qualified leads, booked appointments, completed jobs, and revenue.  ![](https://dbseer.com/wp-content/uploads/2026/11/image.png) The connection was the easy part. The real work was agreeing on definitions. *What counts as a lead? When is an appointment truly booked? How should revenue be valued?* We worked through each question with the client’s marketing and brand teams. On one of the system’s larger brands, we went further: rather than use a single average for how long a customer stays active, we brought in a statistician and calculated a separate figure for each of its 256 locations. *The full case study is coming soon on our website. *  But this work builds on an argument we’ve made before, in our white papers on [marketing analytics for multi-brand companies](https://dbseer.com/white-papers/why-digital-marketing-analytics-and-operations-are-critical-for-multi-brand-companies/) and [turning marketing data into operational insight and ROI](https://dbseer.com/white-papers/creating-marketing-analytics-operational-insights-to-understand-performance-and-roi/): when marketing data lives in one place and the customer record is built with care, every channel gets smarter. AI raises the stakes, because an algorithm amplifies whatever you feed it. Clear definitions become better decisions for your company.   ## **What’s Next: On Our Radar**  Every question you ask an AI assistant costs something to answer, and not every question needs your most powerful (and most expensive) model. Asking for a store’s hours and asking for a forecast of next quarter’s revenue by region are very different jobs, yet many AI systems send both to the same top-tier model and pay top-tier prices for each.  A growing pattern fixes this with a router: a lightweight system whose only job is to read each request and send it to the right place. Simple questions go to quick, low-cost models. Complex ones go to the most capable. It works like a triage nurse who doesn’t treat anyone, but makes sure the specialists spend their time where it counts. In benchmark tests by the [RouteLLM research team](https://www.lmsys.org/blog/2024-07-01-routellm/), routing cut costs by more than 85 percent on one test set compared with sending everything to the top model, while keeping 95 percent of its performance. Real savings depend on the workload and on how much quality a business is willing to trade, which is why routing must be tested against your own questions.  The bigger lesson is about how AI should be built. Models are improving every few months, so the systems that last are built in modular pieces that can be swapped as better options arrive, without starting over. And as AI usage grows, so does the bill. This is why we think about AI architecture and cost together: dbSeer’s FinOps work helps organizations right-size their cloud, data, and AI spend, set sensible usage limits, and put guardrails in place so costs don’t quietly grow.  ## **dbSeer Achieves Select Partner Tier in the Claude Partner Network**  dbSeer has achieved the Select partner tier in the Claude Partner Network Services Track. The Claude Partner Network is how Anthropic works with the firms that put Claude into production for customers. The Services Track recognizes partners for building a certified delivery team, delivering Claude use cases for clients, and proving the work through public customer references. With Select status, dbSeer now partners across all three layers of the modern AI stack: Anthropic for models, AWS for cloud infrastructure, and Databricks for data platforms. [Read the announcement.](https://dbseer.com/news/dbseer-achieves-select-partner-tier-in-the-claude-partner-network/)  ## **dbSeer Joins The Franchise Consortium as an Accredited Supplier**  dbSeer has been formally accredited as a member of The Franchise Consortium (TFC), a selective community of franchise suppliers committed to advancing responsible franchising. Franchisors rely on TFC to find supplier partners who have been vetted for real expertise and a demonstrated commitment to the franchise industry. Led by executives with decades of franchising experience, TFC’s network brings together legal, advisory, operational, and growth specialists, and dbSeer joins as one of a select few technology and data partners in that community. [Read the Announcement.](https://dbseer.com/news/dbseer-becomes-member-of-the-franchise-consortium/)  ## **dbSeer Joins the 2026 Fitness Technology Summit**  This past month the team at dbSeer went to the invitation-only Fitness Technology Summit held in the Washington D.C. area. This Summit allowed for open discussion and exchange regarding the future of fitness clubs and technology. For more of our thoughts on the Summit,[ read our recap here.](https://dbseer.com/news/what-we-heard-at-the-2026-fitness-technology-summit/)   > **Ready to turn your data challenges into competitive advantages?**  > > > > *Reach out to us today. *