The education film · copy, tone and running order (draft 1, 15 Sep 2026)
The proposition. Not a pitch. Vpress has run B2B web to print for 25 years. Over the last two we have put AI inside it and watched what happened. This film shares what we found, in four places: speed, errors, the customer’s experience, and integration. Every claim is either a cited industry figure or a finding from our own portals, marked as such. The viewer should leave thinking “that is how I should be using it”, not “I should buy that”.
Voice. First person plural, plain. “We” is Vpress, a company that has been at this since 2001. We never say “revolutionary”, “seamless”, “unlock” or “powerful”. We say what happened. Sentence case, no capitals, British spelling. Short lines on the left of the stage, the object on the right doing what the line says, exactly as the sales film does. Numbers live on their own line, never inside a sentence, and every number carries its source in small type at the foot of the frame.
Shape. Same spine as the sales film so the scenes re-cut under the new copy: an opening finding, then four chapters (design, brand, connect, quote), each built as what we saw · what we did · what changed, each capped by a findings grid instead of a feature grid, then a close that hands over to the people on the stand. Five to six minutes.
Two kinds of figure. [cited] is a published number with the source in the citations list at the end. [Vpress] is a number for you to confirm from our own data before it goes on screen; the figure written is a placeholder in square brackets, never to be shown as is.
0 · Opening · What we found
Title card: Twenty-five years in. / What we found when we put AI inside web to print.
| words | sub | what is on the stage |
|---|---|---|
| Twenty-five years in. | We started in 2001. This is what changed in the last two. | the years map, as the sales film ends: it opens this one |
| Only three in a hundred. | Printers who say they are fully automated. Most have started; almost nobody has finished. [cited 4] | a single figure, large, source line under it |
| One in four sells online. | For those that do, it is now a quarter of turnover. [cited 20] | the figure |
| We wanted to know why. | So we watched our own portals. Thousands of orders a year, hundreds of brands. [Vpress: orders a year, portals live] | the four marks arrive, the four chapters named |
| Four things moved. | Speed. Errors. The customer’s day. The systems around it. | the four words |
Note for the ident: keep the “asked” phone chat if you like it, but the questions change. “Does AI actually help a print business?” / “It has helped ours. Four ways.” / “Show me.” / “Watch.”
1 · Speed · From brief to press
Title card: Speed. / Where the time was going, and where it went.
What we saw
| words | sub | stage |
|---|---|---|
| The press was never the slow part. | Approval was. Three to six rounds, and more than twenty people with a say. [cited 2] | the figure |
| Content demand is heading up fivefold. | By 2027, marketing teams expect five times the content with the same people. [cited 1] | the figure |
| Sixty-four percent want it faster. | Print buyers asking their printer for a quicker turnaround. [cited 3] | the figure |
What we did
| words | sub | stage |
|---|---|---|
| We put the chat beside the page. | Say the change. It makes it. Nothing else on the page moves. | Demo 1A: the poster edited by asking |
| We let the sheet do the typing. | Drop the spreadsheet. Twelve cards, in seconds. | Demo 2: the set built from the staff list |
| We let it flow. | Ten rows, eight fit. It tells you, you decide, page two arrives. | Demo 3: the menu, the note travelling to page two |
What changed [Vpress figures to confirm]
| words | sub | stage |
|---|---|---|
| Minutes, not days. | A poster from brief to ready to order: [x] minutes on our portals, against [y] days by email. | the before and after on one line |
| Nobody re-keyed a name. | Twelve cards from one sheet, [n] seconds. Zero re-typing. | the set |
Findings grid · Speed (six tiles, the grid design as the feature grids, the kicker “What we found”) 1. The slow part is approval, not print. 2. A change asked for is a change made, in the same minute. 3. A spreadsheet is a brief. Treat it as one. 4. Overflow is a decision, not an error. Ask, then flow. 5. Say what, not how. The rules do the how. 6. Speed came from fewer hands, not faster hands.
2 · Errors · Checked before it prints
Title card: Errors. / Where they came from, and what caught them.
What we saw
| words | sub | stage |
|---|---|---|
| Nearly half of printers say one PDF in ten arrives broken. | A quarter say half of them do. [cited 7] | the figure |
| Reprints run at close to three percent of jobs. | The best in-plants hold it to one. [cited 6] | the figure |
| Four in five companies still ship off-brand work. | The ones that don’t grow faster, by their own account. [cited 8] | the figure |
What we did
| words | sub | stage |
|---|---|---|
| Six checks, every piece, every time. | On brand. Fonts. Bleed. Resolution. Type sizes. Safe area. They run when you order, not when you remember. | the checks ticking |
| The brand rules travel with the template. | Logo, palette, fonts: locked by the rules, not by you. | Demo 4: a hand edit, the checks re-running on their own |
| A fail says why. | The hours line under nine point. Fix it, by asking. | Demo 1: the amber ring, the fix |
| Only the library, not the internet. | Approved once. Rights cleared. Colour-managed for print. | the library sheet with its folders |
What changed [Vpress figures to confirm]
| words | sub | stage |
|---|---|---|
| [x] percent fewer reprints. | Across [n] portals, [period]. | the figure |
| Zero off-brand orders. | Not because people were careful. Because they could not. | the checks pill, green |
Findings grid · Errors 1. Check at order time, not at prepress. 2. A refusal is information. Show the rule. 3. Lock the brand, free the words. 4. A closed library beats a good eye. 5. Fix by asking beats fix by hand. It logs itself. 6. The cheapest reprint is the one the check refused.
3 · The customer’s day · Ask, like a colleague
Title card: The customer. / What they asked for, and what they stopped doing.
What we saw
| words | sub | stage |
|---|---|---|
| Two-thirds of buyers would rather not talk to a rep. | Sixty-seven percent, and rising every year. [cited 12] | the figure |
| Nearly all of them already ask an AI. | Ninety-four percent use it somewhere in the buying process. [cited 15] | the figure |
| Eighty-two percent want to order online. | Two in a hundred printers have a chatbot to answer them. [cited 11, 18] | the two figures side by side |
| One bad answer and they are gone. | Sixty-eight percent will not use a chatbot again after a bad experience. [cited 13] | the figure |
What we did
| words | sub | stage |
|---|---|---|
| Ask, like a colleague. | No menus. No catalogue hunting. It knows your catalogue and your prices. | Ask Vpress / Quote Assist: the newsletter by asking |
| It asks only what matters. | Pages, stock, how many. Not thirty fields. | the chips in the thread |
| Quote, proof and order in one thread. | Nothing copied. Nothing mistyped. | the receipt card |
| A human confirms. | Every order, every fix: a person taps the last button. [cited 14] | the confirm modal |
What changed [Vpress figures to confirm]
| words | sub | stage |
|---|---|---|
| [x] percent of orders now start with a question. | Not a form. | the figure |
| [x] fewer emails per order. | The thread holds what the inbox used to. | the figure |
Findings grid · The customer 1. Buyers want self-service. Give them a colleague, not a form. 2. Ask three things. Never thirty. 3. Keep the history on the screen. It is the audit trail. 4. Let the AI draft. Let the person confirm. 5. One wrong answer costs the relationship. Answer from the catalogue only. 6. The receipt belongs in the conversation.
4 · The systems · Your data, already in
Title card: Integration. / What connected, and what it took.
What we saw
| words | sub | stage |
|---|---|---|
| Seventy-seven percent are automating. | Three percent are done. The gap is the integration. [cited 4] | the figure |
| The top planned investment is AI. | Then workflow, then web to print, then the CRM. [cited 21] | the ranked list |
| Every re-keyed field is a future error. | Sixty-nine percent of buyers have caught a supplier contradicting itself. [cited 9] | the figure |
What we did
| words | sub | stage |
|---|---|---|
| Hand it the spec. | An OpenAPI file, a CSV, a form export. No project. No statement of work. | Connect: the spec dropped |
| It reads. It maps. It builds. | Your field names matched to ours, one by one. The translation, written for you. | Connect: the phases |
| Something is wrong. It notices. | A 400 on one field. Fixed inside the rules, then tested again. | Connect: the error beat |
| Your data, already in. | CRM, spreadsheets, web forms, APIs. Forty-two stores from your CRM, nobody re-keyed a thing. | Connect: used |
What changed [Vpress figures to confirm]
| words | sub | stage |
|---|---|---|
| [x] integrations, [y] days each. | Against [z] weeks the old way. | the figure |
| A field mapped once. | Fills every version, every time. | the fill |
Findings grid · Integration 1. Start from the spec you already have. 2. Mapping is the work. Let it be read, not typed. 3. Test in a sandbox before the first real order. 4. An error found in mapping is an error not found on press. 5. Connect the data, not the department. 6. The last mile is a sign-off, not a project.
5 · Close · The findings, and the people
| words | sub | stage |
|---|---|---|
| Four things moved. | Speed. Errors. The customer’s day. The systems around it. | the four marks strip |
| Eighty-five percent say AI is critical. | The barriers they name are skills and unclear use cases. Not money. [cited 19] | the figure |
| That is what this was for. | Four use cases, shown, not described. | the marks |
| Ask the web to print experts. | On the stand, any question. Twenty-five years of them. | the recap strip close, as now |
Outro: the lockup.
Tone rules for whoever cuts this
- A figure is never inside a sentence. It is the line. The sub says what it means. The source sits under the frame in small type: “NAPCO Research, 2024”.
- Never round a figure to make it neater. Say “close to three percent”, not “three percent”, when the source says 2.8.
- Never present a Vpress figure until it is confirmed. If none is available for a “what changed” beat, cut the beat, do not soften it into a claim.
- “We” says what we did and what we saw. “We” never says what you should buy.
- The demos are evidence, not features. Introduce each one with the finding it proves, never with its name.
- No superlatives. No “seamless”, “powerful”, “revolutionary”, “unlock”, “supercharge”. No exclamation marks.
- Every chapter ends on a plain finding a print owner could repeat to a colleague.
What is reused from the sales film
Every object already exists: the poster editor with the chat (Demo 1, 1A), the staff card set (Demo 2), the menu (Demo 3), the merch catalogue (Demo 4), the newsletter, banner and stock orders (Quote Assist), the portal (Brand Assist, optional here), the spec drop and the flow (Connect). New pieces to build: the figure frame (one number, a sub, a source line), the findings grid (the feature grid with the kicker “What we found” and sentences instead of features), and the before and after line for the “what changed” beats.
Where the Brand chapter went
Brand Assist (the portal themed from a URL) does not carry a finding of its own in this cut; it belongs under Errors as the reason the rules exist (“hand it your brand once, every portal follows”). If you want it as a fifth chapter, its finding is: the brand was never written down before; now the portal is the brand book.
Citations
Primary or independent research is marked P. Vendor-commissioned studies, still credible, are marked V. Figures the researcher could not verify at source are not used above.
- Adobe, Australian marketing teams brace for more than 5x growth in content demand by 2027, September 2025 (part of a 2,800-person global study). V. https://news.adobe.com/news/downloads/pdfs/2025/09/20250916-australian-marketing-teams-brace-for-more-than-5x-growth-in-content-demand-by-2027.pdf
- Adobe, same study: more than 20 approvers, three to six rounds of approval, 46% of marketers spend over 40% of their time on admin, reviews and approvals. V.
- NAPCO Research, Who Buys Wide-Format Display Graphics, 2023: 64% of providers report rising demand for faster turnaround; 75% of buyers call online ordering essential. P. https://www.wideformatimpressions.com/article/three-key-takeaways-from-napco-researchs-who-buys-wide-format-printed-display-graphics/
- NAPCO Research, Annual Commercial Print Trends and Strategy Service, 2024: 77% automating, 3% fully automated. P. https://www.thinkforum.com/blog/production-digital-printing-2025-setting-course-thrive-year-ahead-0
- PRINTING United Alliance and NAPCO Research, In-Plant Printing KPI Report, summer 2024, n=134: reprints 2.8% of jobs on average; 52% hold it to 1% or less. P. https://www.rocsoft.com/blogs/partner-and-dealer/selling-to-in-plants-the-napco-researchprinting-united-alliance-in-plant-printing-kpi-report-can-help-you-better-understand-this-critical-market/
- Ghent Workgroup, PDF survey, 2017, n=1,109: 48% of printers say at least one PDF in ten contains errors; 23% say half or more. P, the oldest figure here. https://www.prepressure.com/pdf/pdf-troubleshooter
- Lucidpress (now Marq), State of Brand Consistency, 2019: consistent branding lifts revenue by up to 33%; 81% experience off-brand content. V. https://www.prnewswire.com/news-releases/study-finds-companies-with-consistent-branding-can-see-up-to-33-increase-in-revenue-300967219.html
- Gartner, sales survey, July 2025, n=632: 69% of B2B buyers report inconsistencies between the website and the rep. P. https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-sales-survey-finds-61-percent-of-b2b-buyers-prefer-a-rep-free-buying-experience
- NAPCO Research, survey of 167 communication buyers, 2024: 82% prefer providers with online ordering. P. https://www.piworld.com/article/the-smart-investments-driving-printing-success/
- Gartner, March 2026, n=646: 67% of B2B buyers prefer a rep-free experience; 45% used AI in a recent purchase. P. https://www.gartner.com/en/newsroom/press-releases/2026-03-09-gartner-sales-survey-finds-67-percent-of-b2b-buyers-prefer-a-rep-free-experience
- Salesforce, State of the Connected Customer, sixth edition, 2023, n=14,300: 61% prefer self-service for simple issues; 68% will not use a chatbot again after one bad experience. V, large sample. https://www.paypalobjects.com/marketing/web23/shared/campaign/enterprise/think-forward-2024/State-of-the-Connected-Customer.pdf
- Salesforce, same report: 80% say it is important for a human to validate AI output. V.
- Forrester, Buyers’ Journey Survey 2025, published January 2026: 94% of business buyers use AI in the buying process. P. https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/
- PRINTING United Alliance, State of the Industry survey, Q1 2024, n=124: 2.4% of printers use AI for customer service, 25.8% plan to within a year. P. https://www.printingunited.com/wp-content/uploads/2024/09/2024-PrU-Expo-Breakfast-Slides-FINAL.pdf
- PRINTING United Alliance, AI Adoption in the Printing Industry, September 2025, 300+ companies: 85% believe AI is critical to competitiveness; barriers are skills, unclear use cases, culture. P. https://www.printing.org/content/2025/09/23/printing-united-alliance-releases-next-report-in-its-groundbreaking-series-on-ai-adoption-in-the-printing-industry
- drupa, 9th Global Trends Report, November 2023: 25% of printers run web to print; for them its share of turnover rose from 17% (2019) to 26% (2023). P, verify in the PDF before use. https://www.drupa.com/cgi-bin/md_drupa/lib/pub/object/downloadfile.cgi/9th_Global_Trends_Main_Report_final.pdf?oid=50679&lang=2
- PRINTING United Alliance, State of the Industry, 2024: planned investment, AI 37.1%, workflow software 30%, web to print 28.6%, ERP/CRM/MIS 22.9%. P. Same slides as 18.
Not used, because they could not be verified at source: “manual job ticketing adds 18 hours” (attributed to Ricoh, only on an uncited market-report page); “rework consumes up to 10% of revenue” (attributed to Markzware, not on Markzware’s page); any JDF adoption percentage (none published since the 2000s); any survey of staff time spent on order entry.