Moments of impact from Don Kellogg's career
Recon AnalyticsOur clients were drowning in industry news from dozens of sources and struggling to keep up with regulatory changes and competitor moves that directly affected their strategic decisions. The volume of information was overwhelming, and they needed a way to stay informed without spending hours sifting through content. This created an urgent need for an intelligent solution that could not only filter relevant news but also provide our unique analytical perspective. I designed and built an end-to-end agentic digest system that transforms how clients consume industry intelligence. Starting with a comprehensive mapping of the most critical news sources for each industry, I developed a workflow using APIs and orchestration tools to automatically crawl, filter, and summarize articles. The core innovation was building an automated analysis engine that connects breaking news with Recon Analytics' proprietary first-party survey data, providing clients with our expert take on what the day's top stories mean for their business. I created two consumption channels to meet different client preferences: a personalized morning email tailored to each client's specific interests, and a private podcast feed that delivers the same insights in audio format for on-the-go listening. The system now operates at significant scale, generating dozens of customized digests each morning that process thousands of news articles across a wide array of industries. By synthesizing and analyzing trends as they happen while injecting Recon's own survey data into the conversation, the digests keep us top of mind with our clients and create a valuable feedback loop where we're all operating from the same perspective on industry developments each morning. Most importantly, clients can now spend less time keeping up with what's happening and more time making critical strategic decisions about how to react, transforming our relationship from research provider to indispensable daily strategic partner.
Recon Analytics · PresidentAs employee #2 at a telecom consultancy, I witnessed major telecom clients making critical strategic decisions based on research data that was 3-6 months old. Traditional survey methods couldn't keep pace with rapid market changes, while syndicated research lacked specificity and custom research was too slow to be actionable. I saw an opportunity to bridge this gap by creating a hybrid solution that combined the speed of syndicated research with the precision of custom insights. I led the development of Recon Pulse from concept to platform, managing a small team responsible for survey programming, data cleansing, ETL pipeline design, and visualization. The core innovation was building a syndicated research framework that allowed clients to 'pulse' their custom proprietary questions on top of it, delivering fresh intelligence weekly rather than quarterly. I personally defined the data specifications flowing through our pipeline and designed the translation process that converted thousands of weekly survey responses into interactive Tableau dashboards, enabling clients to interrogate data and generate insights independently. The biggest technical challenge was maintaining data quality while processing at speed—I iteratively redesigned our ETL flow nearly a dozen times and leveraged my GIS background to build a unique validation system using highly granular broadband provider footprints to verify respondent accuracy, a level of precision unavailable in any competing telecom research solution. The impact was substantial and sustained. Within the first two years, we sold Recon Pulse to 2 of the 3 largest wireless telecom providers and expanded into complementary products for home internet and B2B markets. Five years after launch, we now serve 7 of the top 8 telecom providers in the United States and have successfully launched pulse products across AI, Aviation, Spanish Language Telecom, and Financial Services verticals. Recon Pulse is now recognized as the fastest large-sample custom research option in telecom, directly informing major business decisions around pricing, promotion, and product development across the industry.
Nielsen · Director, Research & Insights - Telecom Practice · June 2013As an analyst covering the mobile operating system market at Nielsen, I noticed the smartphone landscape was shifting fast: RIM was steadily losing ground to Android and Apple, while Windows Mobile and Windows Phone 7 struggled to gain traction. Rather than report these shifts as isolated data points, I saw an opportunity to frame them as an ongoing competitive story that readers would want to follow month after month. I worked closely with corporate communications to pull share data and turn it into bite-sized, easy-to-digest stories built around what we dubbed the "Smartphone Wars." I published these updates monthly, tracking the back-and-forth between platforms and consistently pointing back to our paid data products as the source. The horse-race framing gave the tech press a recurring hook, and outlets kept coming back for the next installment. By 2013, the approach had taken hold: 7 of the top 10 most-read stories on the Nielsen Wire, our corporate blog, were "Smartphone Wars" mobile stories I had collaborated on. The series became one of the most reliable drivers of press coverage and traffic for Nielsen's telecom research during that period.
Recon Analytics · President · January 2023Survey data quality is a constant fight - respondents speed through surveys or give nonsensical answers just to collect an incentive, and that noise flows straight into the analysis if nothing catches it. I set out to build an automated way to flag implausible responses before they ever reached our reporting, without relying on personally identifiable location data. I used the FCC's broadband availability data, which is published using H3, the open-source hierarchical hexagonal spatial index originally built by Uber for location analytics. At the resolution the FCC releases wireless and broadband data, that's over 80 million hexagons across the continental US. I worked out a methodology to rationalize those hexagons down to zip code, the most granular location detail we can use without touching PII, then mapped each zip code to its plausible broadband providers. Every survey response gets checked against that table: if a respondent claims a broadband provider that doesn't actually serve their zip code, it's flagged as low quality. Since there are over 3,000 broadband providers nationally and we can't list them all as fixed survey options, we also let respondents write in their provider as a verbatim, then used AI to match those free-text entries against the plausible provider list for their zip code. That gave us a long-tail view into smaller regional providers beyond the usual top 10-20 names, and another layer to filter on. Once the initial zip-to-hex methodology was built, refreshing it only takes updating with new FCC source data every six months, and the whole check runs automatically as part of our normal ETL process. The result was a straightforward, automated way to catch up to 20% of raw respondents giving bad answers, all without adding manual review steps. Because of our contractual terms with our survey provider, being able to point to specific, provably erroneous responses meant we could push back on paying for that bad sample, lowering our effective cost per usable response. Higher data quality and lower sample cost, both baked into the standard pipeline instead of requiring extra work.
Recon Analytics · President · May 2026As our research / analytics stack scaled, we kept loading in more heterogeneous data - panels, survey codes, public series like FCC, BLS, and Census data, competitive feeds, and each one was structured and curated differently. Giving an LLM enough context to query all of it correctly, without just dumping the entire catalog into every prompt, was becoming a real risk to both speed and accuracy as we on-boarded new sources. I prototyped, and the team built a governed metadata layer that sits under the AI engine: same governed layer under interactive reporting and AI query. For every new or reprocessed data source, a human sets the rules before any model runs against it: what the source is, what questions it can answer, how it joins to the rest of the warehouse, its trust tier, and its known gotchas. That becomes a confirmed instruction pack. Documentation can be auto-drafted, but automation never promotes itself into the live path; a person has to confirm it first. At query time, a deterministic step resolves which sources are actually relevant to the question, then loads only those sources' confirmed instructions. The model never sees the full catalog. It writes SQL inside that narrow frame, and we keep the SQL so anyone can check exactly what ran. This keeps prompts small (so queries stay fast) and keeps the model from ever inventing rules for a source we haven't vetted (so answers stay accurate). The system now governs over 250 independent data sources spanning more than 14 billion rows. Using deterministic tool calling alongside on-demand canonical query examples for each source cut model calls by up to 50% compared to blind queries against the same data.
Vertis Media / TNN · April 2026When Vertis was transitioning a major client to our in-house media buying platform, we faced a critical technical challenge: we needed to assign non-overlapping trade areas to each store location, ensuring that every ZIP code where media was purchased mapped to a single, specific store. With over 10,000 ZIP codes nationally to process, manually working through store-to-ZIP relationships would have taken weeks and threatened our timeline for the platform launch. I developed an automated geospatial solution using an iterative scripting approach. The script created buffers around each point location (store) and counted outward in mileage increments until it encountered the boundary of another store's territory. When a new ZIP code was encountered, it was logged in our database and removed from the next iterative query. This intelligent, cascading approach systematically mapped the entire national territory without overlap or manual intervention. We successfully mapped all 10,000+ ZIP-to-store relationships in less than 3 days—a process that would have taken weeks manually. As a result, we transitioned to the new internal system in less than half the originally budgeted time. Media buyers and analysts were freed up to focus on value-add client activities rather than foundational data mapping, since the critical store-to-territory relationships had been established. The client team was highly appreciative of the work, and I received a promotion shortly after the project's completion.