I Asked an AI the Five Questions Every Print Customer Asks. Tiverton's Shops Answered Zero of Five.

Somebody in Tiverton needs 250 double-sided flyers by Friday. It is Tuesday night. They open ChatGPT and type the question they would otherwise have asked a friend: is there a print shop near Tiverton that can turn around 250 flyers in three days, and what do they need from me?

The assistant does not drive down Main Road. It does not call anybody. It reads whatever text a crawler collected from print shop websites, and it answers out of that text or it does not answer at all.

So the useful question for a print shop owner is not "do I have a website." It is: when a machine reads my homepage, can it answer the five questions my customers actually ask?

Last night I measured that on eight real print, sign and copy shops in Tiverton and the towns immediately around it — Fall River, Newport, Middletown. These are working shops. Two of them have been running presses in this region longer than I have been alive, and one of them has a hand-painted sign out front that is better design work than most of what I see on the internet. Nothing below is a knock on their printing. It is a knock on a class of problem almost nobody has been told about yet, because until about eighteen months ago it did not cost anybody anything.

I have anonymised every shop as Site A through Site H. They did not ask to be graded in public.

The five questions#

I did not invent these. They are the five things a person asks before they hand a print shop money, in roughly the order they ask them:

1. Turnaround. How fast? Can you do Friday?
2. File formats. What do you need from me — a PDF? Can you use a Canva file? A photo off my phone?
3. Minimum order. Will you do 25, or do I have to buy 500?
4. Price anchor. Roughly what am I looking at? Not a quote. A number to think with.
5. Hours. Can I come in Saturday?

A crawler can only relay facts that exist as text. So for each shop I fetched the homepage the way GPTBot fetches it — raw HTML, no JavaScript executed — stripped it to visible text, and searched for whether each answer was literally in there.

What came back#

Four things in that table are worth your attention.

Nobody publishes a price. Zero of six. Not a price list, not a "starting at," not a single anchor number. I want to be careful here, because there are real reasons a print shop does not post prices — stock, quantity and finishing move the number too much to publish honestly. But "no number anywhere" is a different decision from "no price list," and it is the one that gets you skipped. Site E manages it with a line about no minimums; nobody manages it with money.

The only shop that answered three of five is a national franchise's location page — and it answered them with corporate-supplied FAQ boilerplate, not with anything the local owner wrote. Its file-format answer is verbatim: "We can work with just about any digital file type; however, PDFs or Adobe Design Files (.ai, .id, and .ps) are preferred." That is one paragraph. It is doing more work for that business than its entire homepage design.

Site C's turnaround "pass" is the weakest pass in the set, and I am counting it as a pass only because my check was mechanical. The literal text is "the absolute best products with the quickest possible turn around time." No number. A machine asked "can they do Friday?" learns nothing from that sentence, and neither does a human.

Two of the eight have no fetchable website at all. Site G fails the TLS handshake outright — every AI crawler user-agent I tried got tlsv1 alert internal error, and plain HTTP returns a 403. Site H returns HTTP 404 on its own root domain, on both https:// and http://, with and without www. Directories and search results still list those domains as their websites. Both of those businesses are almost certainly still answering the phone and printing beautifully, with no idea their web address is dead.

The one that made me swear#

Separately from the five questions, I checked whether the servers even let the crawlers in. Two of the six reachable sites returned HTTP 403 to GPTBot and ClaudeBot, and HTTP 200 to PerplexityBot, from the same origin. Both of those sites' own robots.txt files explicitly permit all AI crawlers.

Read that again. The owner's file says "come in." Something above that file — a firewall, a CDN, a security plugin somebody switched on in 2023 — says "no." Nobody chose this. Nobody knows it is happening. You cannot see it from your own browser, because your browser is not GPTBot.

This is the part where I want to say something honest about why I bothered to write a whole article about it.

Unfortunately, there is no general manual to tell you the shape of the Jagged Frontier of AI abilities, which are constantly evolving. Trial and error, and sharing information with peers, is vital to figuring this out.

Nobody is going to hand you a chart of what the machine will and will not get right about your shop. There is no vendor who can tell you, and there is certainly no consultant in Fall River who can tell you without looking. You go and find out, on your own material, in twenty minutes. That is the whole argument.

The walkthrough: run this tonight#

Free. No account, no tool to buy. A laptop, a phone and twenty minutes.

Step 1 — Read your homepage the way a machine reads it. (4 minutes)

Open your homepage in Chrome. Press Ctrl+U on Windows or Cmd+Option+U on a Mac. That is View Source — the raw file your server sent, before a single line of JavaScript ran. It is the closest thing to a crawler's-eye view you can get without writing code.

Now press Ctrl+F inside that source view and search, one at a time:

• business day — then turnaround, then next day, then 24 hour
• PDF — then vector, then 300 dpi, then CMYK
• minimum
• $ — then starting at
• Monday — then Saturday, then am

Every search that returns zero hits is a question an AI assistant cannot answer about your business. Write down which ones came back empty. That list is your homework, and it is more useful than any SEO report you have ever been sold.

Step 2 — Now actually ask the machine. (6 minutes)

Copy the visible text of your homepage. Open ChatGPT, Claude or Gemini — the free tier is fine — and paste this:

Below is the entire text of a print shop's homepage. Using only this text and nothing you know from elsewhere, answer these five questions as a customer would need them answered. If the text does not answer a question, say "not stated" — do not guess or fill in from general knowledge.

1. How fast can they turn around a job?
2. What file formats do they accept?
3. Is there a minimum order?
4. What does anything cost, even roughly?
5. When are they open?

[paste your homepage text here]

The "do not guess" instruction is the entire point of the exercise, and it is the step most people skip. Without it the model will pad the gaps with plausible printing-industry generalities and you will come away reassured for no reason. With it, you get an honest count of what is missing.

Do it twice, once in two different assistants. They disagree more than you would expect.

Step 3 — Check nobody is turning the robots away. (3 minutes)

Type your domain followed by /robots.txt in the address bar — yourshop.com/robots.txt. Read what comes back.

You are looking for any line naming GPTBot, ClaudeBot, anthropic-ai, PerplexityBot, CCBot or Google-Extended followed by Disallow: /. That combination bans that crawler from your whole site. Also check whether User-agent: * is followed by Disallow: /, which bans everything.

A 404 here is fine. No robots.txt means everything is allowed by default. Absence is not a block.

Step 4 — Ask a real customer question, cold. (7 minutes)

Open a fresh chat, with no prior context, and type what an actual customer would type: "print shop near Tiverton RI that can do 250 flyers by Friday" — then ask a follow-up: "what do they need from me to start the job?"

You are not looking to see whether you are recommended. That varies per user, per session, and anybody who promises you a ranking there is selling something. You are looking at what the machine is able to say about your shop at all — and, more usefully, at what it says about the shop three doors down.

The fixes, cheapest first#

Free, 15 minutes — write the five answers as plain text in your footer. Not in an image of your storefront sign. Not inside a booking widget. Typed into the page, on every page:

Typical turnaround: 3–5 business days; rush available. Files: PDF preferred, also .ai and .eps; 300 dpi minimum for photos. No minimum order. Business cards from $X per 500. Open Mon–Fri 8am–5pm, Sat 9am–noon.

That single paragraph would move both Tiverton shops from zero of five to five of five tonight. It is the fix I most want you to take even if you never speak to me. It also happens to be the fix that helps the sixty-eight-year-old customer squinting at your site on a phone, which is the part everybody forgets: there is no version of this where you do the work and only the robots benefit.

Free, 20 minutes — name your towns and your jobs in words. "Serving Tiverton, Portsmouth, Little Compton, Westport and Fall River" beats "the East Bay." "Banners, yard signs, NCR forms, wide-format, business cards, vinyl lettering" beats "full-service printing since 1974." An assistant can only match words that are on your page.

Free, 10 minutes — check your website is actually alive. Two of eight in my sample were not. Type your own domain into a browser you have never used, on a phone that is not on your shop's wi-fi. If it does not load, nothing else on this list matters.

One phone call — chase a 403. If your robots.txt permits AI crawlers but the bots are being refused anyway, the answer is in your host's firewall or CDN settings, not in your website's content. Most hosts will tell you in a single support ticket whether they have AI-crawler blocking switched on by default. Quite a few now do, quietly, without ever mentioning it.

Half a day, and worth it — get LocalBusiness schema on the site. Address, phone, openingHoursSpecification, service area. Only one of six sites I tested publishes machine-readable hours. It is the difference between a machine inferring your shop and a machine knowing it.

Are you late to this? No.#

The U.S. Census Bureau's Business Trends and Outlook Survey found that firms with one to four employees went from 4.6% to 5.8% AI use between September 2023 and August 2024. Most small businesses in this country are not using AI at all yet. You are not behind.

But your customers are using it, and that is a different curve. The twenty minutes above is not an AI strategy. It is a check that the machine your customers are already typing into can find out you exist and what you charge. When I audited fourteen print shops in New Bedford in July, nine of thirteen had no LocalBusiness schema and the median site took 16.1 seconds to show anything on a phone. This is the same story arriving through a different door.

Most of your work still comes from referrals, repeat business and the sign over the door. That is not stopping. But the check costs twenty minutes and nothing, and the top fix costs fifteen minutes of typing. That is a very cheap insurance policy against somebody asking a machine about you on a Tuesday night and getting silence back.

---

If you run this and find something you cannot explain — a bot getting refused, a website that will not load from outside your own shop, a homepage that says nothing a machine can use — send me the URL and I will tell you what I see. No charge, no pitch. I am in Fall River, twenty minutes up the road, and I would rather this region's print shops turn up when somebody asks.

Joshua Amado, Avalon Partner · 774.559.8992 · Joshua.Amado@AvalonPartner.com

Method notes: eight print, sign and copy shops in Tiverton RI and adjacent towns, measured 2 September 2026. Homepages only, fetched over HTTPS using the published user-agent strings for GPTBot, ClaudeBot and PerplexityBot, with no JavaScript executed. A question counts as answered only if the words appear in the raw HTML the crawler receives. Two sites could not be fetched and are reported as not measured, never as zero. Sites A and F refused GPTBot and ClaudeBot, so their scores describe what PerplexityBot received. A "fail" means not answerable from the homepage, not absent from the entire site — I did not crawl interior pages. Site E's opening hours exist in JSON-LD structured data but not in visible text; I scored the visible text and am telling you both. I did not measure page speed, whether any assistant recommends these shops, or whether their Google Business Profiles carry hours. Anyone can repeat this: View Source on any homepage and search it.