Somebody has told you that you need data, and three vendors have already found your inbox. All of them can show you a good-looking dashboard. All of them are right that you’re flying half blind. And most of what they’re selling won’t change a single decision you make this year.
We’re gmware, a software and data engineering firm in Austin, TX, with delivery centers in Bangalore and Mohali, India. We also run Shield Suite, retail intelligence for beverage-alcohol brands across 60,000+ storefronts. So we sell one of the things on this page. We’d rather sell it to you in year three than mis-sell it to you now, which is why this post is about sequencing, including the stages where the correct answer is buy nothing.
The mistake almost every emerging brand makes is starting from the tool. You end up with a subscription that produces charts nobody uses, because it was never wired to a decision. Start from the other end. At your size there are about five decisions that actually matter, and every line item in your stack exists to move one of them.
| Decision you’re actually making | What moves it |
|---|---|
| Which markets to push next | Account counts vs licensed universe, sell-through by market |
| Which accounts to keep servicing | Reorder gaps at account level, not market rollups |
| Whether your price is holding | Real shelf price by store, not your price list |
| Which SKUs to cut | Velocity per account, plus your own DTC mix |
| Whether the distributor is working the brand | Authorized accounts with zero shipments ever |
Notice how many of those need account-level detail and how few need national category share. Hold onto that. It’s the whole argument for the sequence below.
Stage 0: the data you already have and probably aren’t joining
Before a dollar leaves the building, you have four or five sources sitting in different inboxes. Nobody has joined them, so nobody trusts any of them. Fix that first.
Start with the distributor’s depletion report. Cases shipped from their warehouse to accounts. Not retail sales, and the difference bites people, which we covered in detail in depletion data explained. At this stage it’s still the most useful file you own, because it’s the only one with account numbers on it.
Then your own invoices to the distributor. What you shipped in, versus what they shipped out. The gap is inventory sitting in their warehouse, and knowing that number stops you from reading a depletion dip as a demand dip.
The authorized-account list is the one you’ll have to chase. It’s every account cleared to buy your product in that market, and nobody publishes it. You get it by asking your distributor, then asking again every month, because it changes and nobody will tell you when it does. Ask for the full street address and the account number, not just the account name. The name alone is useless to you later.
Then the free state licence files. Several states publish the licensed universe for nothing. California ABC posts a daily bulk export of all pending and active licences as a zipped CSV refreshed each business morning, plus reports by city, zip code, county and census tract, plus a list of licensed importers. New York’s State Liquor Authority publishes Current Liquor Authority Active Licenses on the state open-data portal, updated daily, with premises name, legal name, licence class, street address, county, expiry date and geocoordinates. Washington’s Liquor and Cannabis Board posts separate off-premise and on-premise licensee lists along with monthly beer and wine sales-to-distributor files.
Two honest caveats on those licence files. They tell you who holds a licence, never who stocks you or what they charge. And the usage terms differ by state: Washington’s page notes that records obtained under its Public Records Act may not be used for commercial purposes under RCW 42.56.070(8), and currently flags known errors from a data transfer issue. Read the terms for your states, and if you’re planning to build a product on top of them, ask a lawyer rather than a blog.
Last, your DTC and tasting-room data, if you have any. It’s the only place you watch a human being choose your product at a known price. Terrible for market decisions, genuinely useful for SKU decisions.
The join, spelled out
Three tabs. That’s the whole build.
accounts gets one row per retail account: distributor account number, account name, street, city, state, zip, channel, chain flag, sales rep. shipments gets one row per account number, SKU and month, with cases. licences gets one row per licensed premises from the state file.
Join shipments to accounts on the distributor account number. Never on the account name. Account names are spelled four ways across two reports and you will silently double-count.
Joining accounts to licences is the part with no shared key, and it’s also where the value is. Normalize both sides the same way: uppercase everything, strip punctuation, collapse street suffixes so ST and STREET agree, take the five-digit zip. Then match on zip plus the street number and the first several characters of the street name. Hand-check whatever doesn’t match. At a few hundred accounts that’s an afternoon of work, not a project.
The Stage 0 build, in one spreadsheet
Three things fall out of that, and all three are free.
Accounts on the authorized list that have never taken a single case. Those are distribution voids, and they’re the cheapest sales conversation you will ever have, because the paperwork is already done. Licensed premises in your priority zips that aren’t on the authorized list at all: that’s your ask for your distributor, with addresses attached. And a real denominator, so “we’re in 240 accounts” becomes “we’re in 240 of the 1,900 off-premise licences in this county,” which is a different sentence entirely when you say it to an investor.
One rule that decides whether any of this survives: the accounts tab is a single file, with a single owner, and everything you buy later plugs into it. Brands that skip this end up paying a vendor to build an account hierarchy, then discovering it doesn’t match the one their sales team uses.
Stage 1: where the first paid dollar should go
Once the spreadsheet exists you’ll hit its ceiling fast, and the ceiling is always the same. You know who’s authorized and what shipped. You have no idea what’s actually on the shelf.
Our position, and it’s a judgement call rather than a law: at this scale, knowing whether your authorized accounts are actually stocking you and at what price beats buying national share data. The reason is decision-shaped, not technical. Every one of the five decisions in that table above is answered at an account or market level. National category share tells you the weather in a country you can’t fly to yet. If you learn your brand holds a fraction of a percent of a national category, nothing you do on Monday changes.
The counter-case is real and you should know if you’re in it. A brand already sitting in national chain distribution, with a category review on the calendar, needs syndicated data first. Chain category managers reason in panel terms, and showing up to a review without the numbers they use is a lost meeting regardless of how well you know your independents. If that’s you, the Circana vs NielsenIQ vs store-level buyer’s guide is the more useful read.
For everyone else, the coverage question decides it. If a meaningful chunk of your volume runs through independent liquor stores, syndicated panels are the wrong instrument, which we get into in where your coverage goes blind. Buying a measurement that structurally can’t see your accounts is the most expensive kind of wrong.
Four stages, and the trigger for each
Stages 2 and 3, and how you’ll know you’re there
Revenue thresholds for this stuff get quoted constantly and nobody publishes a defensible one, so use triggers you can actually observe in your own company.
| Stage | The trigger you’ll notice | What you add |
|---|---|---|
| 1 | You can’t name your top 50 accounts from memory | Account-level stocking and shelf price |
| 2 | Two people maintain conflicting account lists | Shared product and account master |
| 3 | You’re funding programs you can’t verify | Execution checks, competitor shelf, scorecards |
Stage 1’s trigger has a sharper version: you’ve stopped being able to tell whether a soft market is soft because demand is soft, or because nobody has walked into those accounts in four months. When those two look identical from your desk, you need shelf-level detail, and a distribution void report at store level is usually where brands start.
Stage 2 is a plumbing stage and it’s the least glamorous money you’ll spend. The trigger is disagreement: your number and the distributor’s number differ and nobody in the room can say which is right. That happens because the same SKU carries different item codes in different markets and the same chain shows up under five spellings. Until there’s one product master and one account master, every new data source you add makes the disagreement worse rather than better. This is also when price tracking stops being optional, because by then you’re in enough retailers that shelf price varies across them and your price list has stopped describing reality.
Stage 3 is about accountability, and both directions of it. You’re spending real money on displays and features, so you need to know they actually happened on the floor. And you’re now big enough that your distributor conversations need to be evidence-based rather than vibe-based, which is what a distributor scorecard is for. Competitive shelf tracking lands here too, because it only becomes actionable once you have enough distribution that losing a facing to a competitor costs you something measurable.
Build, buy, or hire
Every founder hits this fork and most pick the wrong tine, usually because hiring feels more decisive than buying.
Three ways to get a first data capability
The analyst hire is the one we see go wrong most often. A good analyst joins, asks for the data, discovers there isn’t any in a usable state, and spends two quarters doing record matching and file wrangling. That’s data-engineering work, they usually didn’t sign up for it, and it’s an expensive way to buy an ETL script. If you hire, hire for whoever will own the account master and enjoy owning it, whatever their title says.
Tooling first fails quieter. The dashboard goes live, looks great in the board deck, and is answering a question nobody asked because the underlying account hierarchy came from somewhere other than your sales team. Outsourcing the feed avoids both traps and introduces one of its own: if you never look at the raw files, you inherit someone else’s definitions of a chain, a market and an active account. Ask for the raw joined data alongside whatever interface you’re given.
What to skip for now
Some things feel urgent at this stage and aren’t.
- National category share dashboards: interesting reading, zero decisions. Revisit them when a chain review shows up on the calendar.
- Predictive forecasting. A forecast needs history and clean masters. Build those first, then read our demand forecasting guide once the inputs exist.
- A data warehouse. With a handful of monthly files, a warehouse is a maintenance burden wearing a maturity costume. Files in cloud storage plus one spreadsheet holds up longer than anyone admits.
- Per-seat BI licences for a team of three. You’ll outgrow this. You haven’t.
- Consumer research panels. Right idea, wrong order. You don’t have the distribution yet to act on what they’d tell you.
- A second data source bought to check a first one you don’t trust. Fix the trust problem instead. Two unreconciled sources is the same problem twice.
What we’d recommend
Spend your first month on Stage 0 and nothing else. Get the four files in one place, do the matching by hand, and produce two lists: authorized accounts that have never ordered, and licensed premises in your priority zips that aren’t authorized at all. That work costs you a weekend and it will change what you do next quarter more than any subscription would.
Then, when you can no longer hold your top accounts in your head, buy visibility into the shelf where your volume actually lives. For most emerging brands that means account-level stocking and real shelf price across the accounts you already have, not national share across a category you don’t yet influence. If you’re the exception, already in chains with a review on the calendar, invert that and buy the panel data first.
Whichever stage you’re in, the account master is the thing worth being precious about. It’s the one asset every later purchase plugs into, and it’s the one nobody can sell you. If you want a second opinion on where you actually are, tell us what files you get today and how many accounts you’re tracking, and we’ll give you a straight answer, including if the answer is that you shouldn’t buy anything yet. When the plumbing does become the bottleneck, that’s what our data analytics and business intelligence practice does for beverage-alcohol brands all day.