Sign Making Software Problems: The 5 That Cost Real Money, and How to Avoid Them

Why does the estimate keep living in one person's spreadsheet?

Because sign estimating is not a price list and every tool in the category treats it like one. A four foot channel letter set in five inch depth with white returns, acrylic faces, trim cap, remote power supplies and a raceway is a bill of materials plus labour stages plus a mounting method plus an electrical connection. Your estimator knows the module spacing changes when the letter drops below eight inches. He knows a mount above twenty feet prices differently because it needs a bucket truck rather than a lift, and that downtown adds a lane closure permit and a flagger.

Cyrious Control lets you build assemblies and part items, and Shopvox and Corebridge give you a product builder with fields. None of them give you conditional rules, so the moment your pricing depends on a combination of job attributes the estimator hand adjusts every quote, and the workbook that has been passed down through two owners becomes the actual company. When that person takes vacation, quotes go out at guessed numbers.

The model that works is a configurable product per sign type with a real bill of materials, material yield calculated off actual sheet sizes rather than a flat waste percentage, labour priced per stage covering design, machining, fabrication, paint, wiring, assembly and install, and rules that fire on job attributes. Feed it the letter set and it returns material cost, hours by stage, an install method derived from mounting height and access, and a flag when permit or engineering is triggered. The estimator reviews and overrides, and every override is logged. Six months later the override log tells you exactly which rules are wrong and by how much, which is the closest thing to free margin in this business.

What goes wrong when you migrate out of Cyrious?

The honest answer is that some of it does not come out cleanly, and any developer promising a lossless migration has not opened the schema.

Customers, contacts, part items and price lists map over well with database level access. Open orders move with effort. Where it gets difficult is historic job detail, particularly assemblies and the custom pricing overrides that carry the reasoning behind what you charged, because those were stored to serve the application rather than to be exported. Plan for a portion of your history landing as read only archive that is searchable but not structured, and decide deliberately how far back you actually need structured data. Most shops need three years for costing analysis and want twenty for the ability to answer a warranty question.

The second trap is that your part items are not as clean as you think. The same substrate appears under four names entered by four estimators over a decade, with different units and different vendors. Deduplicating that is analyst work rather than engineering work, and it has to happen before the estimating rules are built, because the rules reference the materials. Start it before development does.

The third is job costing history. Time entries against the wrong job code, install crews logging a full day for a two hour service call, and material issued at purchase order rather than at pull all mean your historic actuals are directionally useful and precisely wrong.

Why do the accounting, design and production integrations break after launch?

Three boundaries matter and each fails differently.

Accounting is the first and the rule is simple: do not try to replace it. Keep the accounting system as the financial record, push invoices out and pull payment status back, and keep the per job profit and loss in the custom system where the stage level detail lives. What breaks is duplication, when somebody edits an invoice in accounting and the job record never learns, so pick one authority per field and show which system a value came from.

The design and production boundary is the second and it is fiddlier than anyone expects. Pulling nesting yields out of a routing package, or reading specifications off artwork, is awkward work and the file formats change with updates you do not control. Getting actual yield for your three highest volume substrates is worth real money. A general purpose bridge to every design tool your studio touches is a project with no end.

The third is the mobile survey and it fails on connectivity. Half your sites have no signal inside a shell building, so the survey has to work fully offline and sync later, and photographs are large. Build the offline path first rather than adding it after a surveyor loses a morning of measurements in a car park.

What happens when permits and listing compliance are not covered?

Permitting is the critical path in this business and every off the shelf sign tool treats it as a checkbox with a date. Real permitting is a dependency graph. The city needs the landlord authorisation, the landlord needs the rendering, the rendering needs the survey, the variance needs the engineering seal, and the seal needs the wind load zone and the mounting substrate. Some jurisdictions accept electronic submission, some want a paper set walked in, and some will not tell you they rejected it unless you call.

Because the permit is a note rather than an entity, nothing can hold production, so the shop keeps building against an approval that does not exist. The fix is to make permits first class records with jurisdiction profiles you build once per city and reuse: required document set, typical review time, fee schedule, submission method and contact. Each permit gets a real status flow from survey needed through documents assembling, submitted, in review, corrections requested, approved and expired, and it gates the production stage so fabrication cannot release without it.

Listing and inspection compliance is the second half. If you fabricate electric signs, listing and label tracking gives each unit a serial and a label chain, and electrical inspection scheduling attaches to the permit record. If you take municipal or school district work you will meet certified payroll and prevailing wage reporting. None of this is exotic in this industry, and a developer who has not heard of it will discover it in week ten of your build at your expense.

Should you build custom or configure what you already own?

If you are a single location shop doing mostly repeat banners, decals, vinyl and small interior sets, configure. Cyrious or Shopvox already exceeds what you need, and the discipline of using one properly will improve you more than a build will. Spend the money on production equipment.

Before pricing anything, check whether your complaint is capability or configuration. Many shops describe a system as unable to do something it can do, because it was set up by someone who left and nothing has been revisited since. Corebridge and Signtracker are both closer to production than people expect, and a week with an implementation consultant is cheap.

Build when these appear together. Your estimator is a single point of failure and you cannot hire a second because the knowledge is not writable in any tool you own. You run multi site rollouts where a permit slip cascades into fabricated inventory and cancelled crane days. You have two or more production locations quoting the same job differently. And the clearest signal: you have already bought a second and third tool to patch the first, so the permit tracker is a spreadsheet, the survey tool is a text thread, the schedule is a whiteboard and job costing is a quarterly guess. That stack is not cheaper than a build, it is just billed to you in margin rather than in invoices.

How do hidden costs get into the quote?

Sign type count is the first and largest. A shop doing channel letters, monuments, wayfinding, accessibility signage, vehicle wraps and digital displays is six product engines rather than one with six settings, and each carries its own bill of materials structure, labour stages and rules. Scope two for the first release and add the rest once the pattern is proven.

Jurisdiction count is second. Each city profile is fast to build and slow to populate, and a national rollout shop may need forty of them. That is your permit coordinator's time rather than the developer's, and it belongs in the plan as work rather than as an assumption.

Third, the Cyrious extraction, which is priced honestly only after somebody has opened your database. Fourth, listing, inspection and certified payroll compliance, which is cheap to build early and expensive to retrofit. Fifth, shop floor time capture hardware, since a tablet at each station with a job scan is the only way costing numbers become real, and a costing module fed by memory is worse than none.

What separates a build that works from one that fails here?

The shops that succeed put an estimator in the room for three or four weeks before code is written. The estimating rules have never been documented anywhere, and extracting them is a conversation, not a specification exercise. Every rule that comes out of that room is margin you were previously guessing at. Shops that skip this stage get an invoicing application with a nicer interface.

Second, gate production on the permit from day one, even if the permit module is thin at first. The single control that stops fabrication releasing against an unapproved permit prevents more loss than everything else in the build combined, and it is not technically hard. It is only ever missing because nobody modelled the permit as an entity.

Third, close the loop on costing quickly. Capture time where the work happens, on a shop floor tablet with a job scan and on the installer's phone with arrival and departure, and issue material at pull rather than at purchase order. Then quoted against actual by stage tells you which sign types make money, and it feeds the estimating rules directly. In the shops we have built this for, the leak is almost never where the owner guessed.

Finally, settle ownership before kickoff: source code, database, cloud accounts in your name, and a repository you control from the first commit rather than a handover at the end. At Digital Heroes that is standard. Your estimating rules are twenty years of learning things the hard way, and they should not sit in somebody else's account under a licence you renew.