Market Sizing for Mid-Market Teams: Governing Demand Data Across Markets

At a company of a hundred to a thousand people, the first question is not what your site should say. It is how much real demand exists for what you sell, market by market, and how much of it you can realistically win. You almost certainly have that number somewhere. The problem is you have it seven times, in seven spreadsheets, owned by nobody, and no two of them agree. This is how a mid-market marketing team turns demand sizing from a scattered guess into a governed data asset the whole org runs on, integrated with the stack you already own and defensible to the leadership team that funds it.

#The plain-English version

Your addressable market is the set of buyers who, this quarter, are actively looking for what you offer. Not everyone who could ever use you. The ones searching now, across every region and product line you cover. At mid-market scale that pool is large enough to matter to the board and specific enough that a good model can bound it. Your job is not to estimate it once. It is to own one shared version of that number that your demand-gen team, your regional leads, and your finance partner all read the same way.

This matters because it changes what marketing is arguing about. When every market has its own demand estimate, budget meetings turn into a fight over whose spreadsheet is right. When there is one governed model, the argument moves to the real question: given this demand, where do we invest first. That is the shift from marketing on instinct to marketing you can defend. The method underneath is the same one we lay out in the market-sizing method. What changes at your size is governance, ownership, and scale.

#The three circles, run across every market you cover

Marketers have a formula for this and it is worth borrowing in plain words. Picture three circles, one inside the next. The biggest is everyone who could ever buy what you sell. Inside it is everyone in your served markets who is buying it this year. Inside that is the slice you can realistically win top placement and mention for. The jargon, if procurement asks for it: total addressable, serviceable addressable, and serviceable obtainable market. At mid-market scale the discipline is not just working inward once. It is running that same inward logic consistently across every region, segment, and product line, so the numbers roll up into one comparable picture instead of a pile of one-off estimates.

Everyone who could buy. The all-markets, all-time number. It shows up in board decks and it is useful for narrative, but it is not a planning input. Never budget against it.
Everyone in your served markets buying this year. Scoped to the regions and segments you actually cover, in a real time window. This is the pool worth modeling, market by market.
Everyone you can actually win. The high-intent demand happening now that you can realistically rank for, be named in, and convert. This is the number every downstream plan is built to grow, and the one that has to be comparable across markets.

The failure mode at your size is not ignoring the smallest circle. It is sizing it differently in every market, so a region that looks strong is really just measured generously. Governance is what makes the smallest circle mean the same thing everywhere. That is the whole point of owning one model instead of federating the guesswork.

#The technical version: one demand model, four passes

The model is built in four passes, the same four for every market so the outputs are comparable. Standardize the passes, and a regional lead can run one for a new market and have it roll straight into the shared picture.

Seed the intent terms. Start from how buyers describe the need, not how your product marketing describes the offering. People search for the problem, not your category name. Pull these from real sales-call and support language, from the People Also Ask box, and from your Search Console queries, which at your scale is a large and honest sample.
Attach volume and geography. Each term gets a demand estimate scoped to the specific market, not a global figure. Global volume is a board-deck number; the per-market figure is the one that funds a plan.
Score winnability. A term you can realistically win in a couple of quarters beats a head term a national brand will hold for years. Grade each by who currently holds the top organic results, the map or product packs, and the AI answers.
Add the invisible demand. A growing share of intent now resolves inside an AI answer or an AI Overview with no click. That demand is not lost, it moves to whether the model names you. How to capture it is a governed workstream of its own.

The difference from a small team running this once is that at mid-market scale each pass is a documented, repeatable procedure with an owner, not a thing one analyst does in their head. That is what lets the model survive a reorg, an agency change, or the analyst leaving. Write the passes down as a standard, then run them as one.

#A worked example: a multi-region service business

Numbers make this concrete, so here is the four-pass method run across markets. The figures are illustrative, not a real client, but the shape is what a governed demand model looks like when it spans regions. Say a mid-market company selling a considered service across eight metros, with a central marketing team and a lead in each region.

Seed the intent terms, once, as a shared library. Group the real phrasings buyers use into a master term set that every market inherits, plus room for local variants. Deduped, say that is roughly 120 meaningful terms in the shared library, with each metro adding a handful of its own.
Attach volume and geography per metro. Run each metro against the same term set. Say the eight metros come in between 400 and 1,800 high-intent searches a month each, for a served pool on the order of 7,000 to 9,000 a month across all of them. The global figure for the head term alone is in the hundreds of thousands, and it is irrelevant to any regional plan.
Score winnability per metro, on the same rubric. National aggregators hold the head terms everywhere, so those are a multi-year fight in every market. But the winnable share differs by metro: in three you already have brand and reviews, so obtainable placement covers maybe 55 percent of the pool; in the five newer markets it is closer to 25 percent until reputation builds. Same rubric, honest per-market answers.
Add the invisible demand, tracked as its own column. Roughly a quarter of that intent now resolves inside an AI answer with no click, and that share is rising. Model it as a separate, governed line so leadership can see the classic-click pool and the answer-engine pool move independently.

The punchline: this company's serviceable obtainable market is not one number, it is eight comparable numbers that roll up to one, and the rollup tells you exactly where the next dollar of budget earns the most. Three mature metros with high winnable share and five newer ones where reputation is the constraint is a completely different plan than an even spend, and you can only see it because the model is governed to be comparable. The measurement that turns this into spend decisions is the same set of numbers we cover in the downstream layers.

#Where the numbers come from, and who owns each source

A demand model is only as honest as its sources, and no single source is enough. Triangulate across six, and at your scale assign a clear owner to each so nobody assumes someone else is keeping it current. Weight the sources grounded in real behavior over the ones that are modeled estimates.

Your own Search Console, across every property. At mid-market scale this is a large, honest sample of real queries real buyers typed to find you, with impressions and position attached. It is the truest source you have. Governance here means one team owns the export and the definitions so every market reads it the same way. Google's own documentation on how search works is a useful shared reference for that team: Google Search Central.
Keyword volume tools, scoped per market, for terms you do not rank for yet. Treat their numbers as ranges, not gospel; two tools rarely agree. Vendor-manage this deliberately, because at your scale you are probably paying for two or three overlapping platforms and should consolidate.
The People Also Ask box and autocomplete, which show how buyers in each market actually phrase the need, not how your category names it.
Business-profile and map-pack or marketplace insights, which surface local and category intent that organic keyword tools never show.
Public and census data for the addressable population per market: how many households, businesses, or filings in each region plausibly need what you sell in a year. This is the Public Data layer putting something real underneath the search volume. The primary source is the U.S. Census Bureau, and owning one clean pull of it beats every market sourcing population figures its own way.
The live results themselves per market: who holds the top organic spots, the packs, and the AI answers for each term. That is the raw input to the winnability score, and it is the one an off-the-shelf tool never checks for you.

Where the sources disagree, say so in the model rather than averaging them into false precision. A demand model that documents its own uncertainty is worth far more to a leadership team than one that pretends to be exact, because the honest one survives the first hard question in the budget review and the confident-but-wrong one does not.

#Integrating the model with the stack you already own

At mid-market scale the demand model does not live in a standalone spreadsheet. It has to plug into the systems your team already runs, or it becomes shelf-ware the day the analyst who built it moves teams. Do not buy a new platform for this; wire it into the stack you have.

Land the model where planning already happens. If your team plans in a BI tool or a data warehouse, the demand model is a governed table there, not a file on someone's drive. One source, versioned, with change history.
Feed it from your existing sources, not manual re-entry. Search Console, your keyword platform, and your CRM should flow into the model on a schedule. Manual copy-paste is where drift and staleness enter.
Expose it to the people who need it, read-only. Regional leads and finance see the same numbers you do, without the ability to quietly re-slice them into a friendlier version.
Keep the definitions in one place. What counts as a served market, a high-intent term, a winnable share: these are governance decisions that live in the model's documentation, not in each analyst's memory.

#Governance and ownership: who owns the number

The single most valuable thing you can do at mid-market scale is name one owner for the demand model. Not one owner per market, one owner of the model. That person owns the definitions, the passes, the source refresh, and the rollup. Regional leads contribute their markets; the owner guarantees the whole thing is comparable and current. Without a named owner, a demand model degrades into the exact seven-spreadsheet problem it was supposed to solve, just with nicer formatting.

Clear ownership also settles the recurring cross-team fights before they start. When regional and central marketing disagree about a market's potential, the model owner arbitrates against the documented rubric, not against whoever argues hardest. When finance questions a number, there is one person who can explain how it was built and one place the assumptions are written down. Governance is not bureaucracy here. It is the thing that lets a large team move on one set of facts instead of relitigating the facts every planning cycle.

#The mistakes that blow up a mid-market demand model

Most bad demand estimates fail the same handful of ways, and at scale each failure multiplies across markets instead of hitting one plan.

Using global volume for a regional plan. The head-term number is a narrative figure. The per-market figure is the one that turns into pipeline, and it is a fraction of the size. Budget against the market, not the category.
Sizing markets on inconsistent rubrics. If each region grades winnability its own way, the rollup is meaningless and the strongest-looking market is often just the most generously measured one. Standardize the passes.
Chasing head terms you will lose for years. A cluster of winnable terms you can own in a couple of quarters beats a single trophy keyword a national aggregator will hold indefinitely, in every market.
Counting everyone who could buy instead of everyone buying now. Latent need is not demand you can capture this quarter. Size the active pool per market, not the theoretical one, or the model over-promises and marketing loses credibility with finance.
Confusing traffic with intent. A campaign that pulls fifty thousand curious readers is worth less than one that pulls five hundred buyers ready to purchase. Count the intent, not the visits, and hold every market to that.
Ignoring the invisible pool. If you only count classic blue-link clicks you undercount real demand and miss the answer-engine shift entirely, and at scale that blind spot compounds across every market at once.

#Risk, compliance, and defending the model to leadership

At mid-market scale a demand model is not just a planning tool, it is a document leadership and sometimes legal will lean on. Treat it that way. Two things protect you.

Data provenance you can show. Every number should trace to a named source and a refresh date. When a board member asks where a market figure came from, the answer is a documented source, not "our agency gave us that." Provenance is also what keeps a public-facing market claim defensible if anyone ever challenges it.
Honest ranges over false precision. Leadership trusts a model that says a market is 6,000 to 8,000 high-intent searches a month more than one that claims exactly 7,143. The range is more accurate and more credible, and it survives scrutiny. A model that pretends to a precision it does not have loses the room the first time a number turns out soft.

Defending the program to leadership is easier when the demand model is the anchor of the story. You are not asking for budget on faith. You are showing a governed count of winnable demand per market, the share you can realistically take, and the plan to take it. That is a business case, not a marketing pitch, and it is exactly what a mid-market leadership team funds.

#Demand is a moving number, so govern the re-count

A market is not a figure you size once and file. Demand moves with seasonality, local events, the fiscal calendar, and competitor entry, and at your scale different markets move on different clocks. A model you built in January can be meaningfully wrong by the third quarter in half your regions. Put the re-count on a governed cadence: refresh every market quarterly, and re-run any market the moment it gains a product line, an office, or a major competitor. Because the passes are standardized and the sources are wired in, the re-count is a scheduled job the model owner runs, not a fire drill. The whole point of governing the model is that keeping it current is routine instead of heroic.

Once the org can see one honest, comparable picture of winnable demand across every market, every downstream layer has a target it can trust: what to build, how the site should read, and which numbers to hold each market to. The same discipline retold for other kinds of operators is worth reading alongside this: the agencies version for teams sizing demand on behalf of clients, the micro businesses version for owner-operators doing it themselves, and the SMEs version for a small internal team building its first repeatable process.

Want a governed demand model built across your markets, integrated with your stack and ready to defend to leadership? Run the estimator to see the first cut, look at what we ship and how the pieces fit, or read how we work with professional services teams end to end. Talk to us when you want to scope it with a human.