ISSUE 012
What You Can Actually Find Out About Your Competitors
A thing I have watched happen more than once: somebody pays for a tool that tells them how much traffic a competitor's website gets, reads the number out loud in a meeting, and nobody asks where it came from. It came from a model. Meanwhile there is a pile of material about your competitors that is free, literally true, and almost never opened, some of it sitting on a Missouri state website in a plain text file. This entry is about the difference between those two things.
An idea worth thinking about
Nobody can tell you how much traffic your competitor gets
In 2016 an SEO agency called Screaming Frog ran the test everybody should run and almost nobody does. They took 25 websites they had analytics access to, pulled the real organic visit numbers out of Google Analytics for February, March and April, and compared them against what SimilarWeb, Ahrefs and SEMrush said those sites were getting. Across all 25, SimilarWeb was closest, estimating 15.7 million visits against an actual 13.4 million — a 17 percent overestimate in total. The other two came in under.
The useful bit is the small sites. There were three sites with under 10,000 organic visits, and in their words, almost every estimation by each tool was at least 40 percent incorrect, either over or under. The largest single overestimate in the whole test was 128 percent, the largest underestimate 94 percent. Your competitor in Poplar Bluff is almost certainly a small site.
Three disclosures before anyone treats this as settled. Screaming Frog sells SEO software and services, which is adjacent to these tools rather than opposed to them. The test is ten years old and the tools have been rebuilt since. And their own closing line is more modest than the headline suggests: our test is by no means a thorough mathematical or scientific experiment, merely a quick test to try and gauge the accuracy of such tools. The under-10,000 finding rests on three websites. I am citing it because it is the only comparison I found where someone held the real numbers, not because the sample is adequate.
A newer comparison points the other way on size and the same way on reliability. Omniconvert, a company that sells conversion-rate software, compared SimilarWeb against Google Analytics for 1,787 e-commerce sites that had given it read-only access to their analytics, and reports SimilarWeb showing roughly 94 percent more sessions than Analytics recorded. That is almost double, where Screaming Frog found 17 percent over. I am not going to reconcile them, because they are not measuring the same thing: one counted UK organic visits to a root domain in 2016, the other counted total sessions on self-selected e-commerce sites at an unstated date, and both authors sell something in the neighbourhood. The honest summary is that two conflicted parties with real data both found the estimates wrong, and could not agree on which direction.
A tool I disappeared into
Free federal records of what your competitors decided to protect
The US Patent and Trademark Office runs three public search tools. None of them needs an account, although the site suggests signing in when traffic is heavy, and I was never asked to pay for a search — the fee schedule the page links to is about filing. The Trademark Search system looks up marks. TSDR, which stands for Trademark Status and Document Retrieval, pulls the status of a particular application and the documents filed with it. Assignment Center searches recorded changes of ownership.
What makes this better than a traffic estimate is that every field in it is a fact somebody swore to. A filing date tells you when a competitor decided a name was worth money. The goods and services description tells you, in their own drafting, what categories of business they intended to be in — often broader than what they currently sell, which is as close to a public statement of intent as you will get. The specimen filed with a use-based application is an actual photograph of how they were using the mark in commerce. And an assignment record tells you when a brand changed hands, which is sometimes the first public sign that a firm has been sold.
The limit is large and it is the same limit as every registry: it only contains the people who registered. Most two-to-twenty person businesses in southeast Missouri have never filed a federal trademark, so searching for your nearest rival will often return nothing at all, and nothing is not evidence of anything. It is also a federal register, so a Missouri state registration or a plain unregistered common-law name will not be in it.
Where I would actually use it: when a competitor starts behaving oddly — a new name on a van, a sudden rebrand, a company from two counties over advertising in yours. Searching the name takes ninety seconds and occasionally tells you that somebody filed for it eighteen months ago across a list of services much wider than the one they are currently offering.
Learn something
You can ask a Missouri public body for its records, and most people never do
Missouri's Sunshine Law is Chapter 610 of the state statutes, and the Attorney General's office keeps an overview page for it with links to the statute, a longer guide, sample request language, an FAQ and a complaint form. It is written for citizens rather than for lawyers, and it is worth half an hour even if you never file anything.
Two practical points from the page. Every public governmental body in the state is required to have a custodian of records who is responsible for responding to public records requests, and the law does not require that an open records request be submitted in a specific format — although the page encourages putting it in writing so you have a record of what you asked and when. If a body refuses you, the complaint goes to the Attorney General, though the page is clear that the AG enforces the law rather than acting as your lawyer, and that its office is not the repository for all public records in the state. You ask the body that holds the record.
The page itself does not set out response deadlines or fee rules, so I am not going to quote you any; those are in the statute and the longer guide, and you should read them there before you rely on them.
What I would request, and this is my suggestion rather than the Attorney General's: bid tabulations. When a city, county, school district or public agency puts work out to bid, the tabulation of who bid what is typically a public record, and it is the only legal way most of us will ever see a competitor's price for a defined scope of work. Building permits are the other one — they tell you which contractor is doing what, where, and at what declared value. Neither of these is in any intelligence subscription, and both are sitting in a filing system eight miles from your office.
What we’re seeing
Twenty-one plumbing, heating and air conditioning locations in the whole county
The Missouri Department of Revenue publishes a plain text file that counts every business location registered for sales or use tax, broken down by county and by four-digit industry code. It is free, it is a direct download, and it takes about four seconds to open. In Cape Girardeau County, under code 1710, Plumbing, Heating, and Air Conditioning, it shows 12 sales tax locations and 9 use tax locations — 21 in total. The whole county contains 7,490 registered business locations. Statewide, code 1710 has 1,518, out of 594,142 locations in Missouri.
Before anyone acts on that: the file's own header says PRINTED: 20180105. It was produced on 5 January 2018 and it is eight years old. It sits on a page where the Department updates other reports through the current quarter, which makes it dangerously easy to assume this one is current too. It is not, and I have no way to tell you how much the county has changed since.
The second caution is about what a location is. This counts tax registrations, not firms. A business with two branches appears twice. A sole trader with no sales tax liability does not appear at all. And the split between the two columns matters — Missouri vendor's use tax is generally collected by sellers based outside the state, so my reading, and it is my reading rather than anything the Department says, is that several of those 9 use tax registrations are out-of-state sellers rather than rivals you could drive to. Which would put the local count nearer 12 than 21.
Even with both cautions, look at what it is. Not an estimate, not a model, not a panel projected up to a national figure. A count, from the agency that collects the tax, of how many businesses registered under your industry code in your county. Nobody sells this. It is on a .gov page with a filename like DT60871_SIC_cntytots.txt and it has been there for years.
The contrast with the first half of this entry is the whole point. A subscription will quote you your competitor's monthly traffic to three significant figures, and the one test I could find where somebody held the real numbers found almost every small-site estimate wrong by at least 40 percent — on a sample of three sites, which is itself the problem. The state will tell you exactly how many businesses registered in your code in your county, for nothing, and tell you on its face how old the file is. One of those is insight and the other is a confident interface.
So: open the file, find your county, find your four-digit code, and write the number down. Then do the part the file cannot do — drive the county and count how many of them you can actually name. If your list is shorter than the state's count, you have competitors you have not noticed. If it is longer, some of them are not registered, which is its own piece of information. Either way you now have a denominator, and almost nobody in your industry in your county has one.
One question
How many direct competitors do you have in your county?
Your number first, then check it against the state's. I would like to know which way the gap runs.
More in this pillar as it's written.
— Shana
Sources link to the original publisher, never to a summary.