Modernization Should Start Where the Work Gets Stuck

You have seen the modernization proposal. It replaces the ERP or the claims platform, it runs about two years, and the number on the last slide sends the conversation to next year's budget. Next year it moves again.

Meanwhile the exports keep getting built and the status calls keep getting answered, because the work does not pause for a budget cycle. Modernization should start at the workflow your team stalls on every week, which is rarely the oldest platform on the list, and that project is small enough to approve this quarter.

‍

Why do modernization projects start with the biggest system?

Modernization starts at the biggest system because a platform comes with a vendor and a business case someone else will write for you. A handoff between two departments comes with neither.

Birmingham City Council estimated £19 million to move onto Oracle. The project began in 2018, go-live slipped from December 2020 to April 2022, and the total cost of the failure has now reached £216.5 million. The council could not produce auditable accounts against a £3 billion revenue base, declared effective bankruptcy in 2023, and spent more than £5 million on manual workarounds while the new system was being repaired.

A program bought to remove manual work generated £5 million of it, and the people stuck doing that work were the same people who had been asking for help before the project started.

‍

Where does work get stuck in manufacturing and insurance operations?

Work gets stuck wherever a person has to wait, re-enter, or go looking before the next step can happen.

The pattern looks the same on a plant floor and inside a health plan:

• Somebody waits on an approval that has no queue behind it
• Somebody exports and cleans a file before a decision can be made
• Somebody checks three systems to answer one question
• Somebody routes a request by hand because no rule covers it
• Somebody explains a status the customer should have been able to see
• Somebody reconciles two numbers that were supposed to match
• Somebody remembers the exception, because it was never written down

In manufacturing that lands in scheduling, quality holds, maintenance requests, and customer order updates. In health insurance it lands in prior authorization follow-up, claims exceptions, provider data changes, and document intake. A plant scheduler rebuilding the week in a spreadsheet and a claims analyst keeping an exceptions tab are doing the same job, which is holding together a seam no system owns.

‍

Should you start modernization with your team's biggest complaint?

You should not start modernization with the biggest complaint, because the loudest one is usually the one that irritates a senior person occasionally, and the expensive one is usually the one a junior person absorbed years ago and stopped raising.

An executive who waits twenty minutes for a number once a month will mention it in every meeting until somebody fixes it. A scheduler who spends three hours rebuilding the same spreadsheet every week has not mentioned it since their first year, because by now it is the job. Three hours a week comes to about thirty-nine hours a quarter, plus a single point of failure, against the executive's twenty minutes.

Rank the candidates by what they cost rather than by how often you hear about them:

• How often does it happen, and how many people does it touch
• How much time does it consume in a week
• Does it delay revenue, a shipment, or a decision a customer is waiting on
• Does it create compliance or reporting exposure
• Does it generate rework somewhere downstream
• Does it depend on one person's memory
• Would fixing it make two other workflows easier

There is a real version of the objection. An executive raising the same problem every month is telling you what leadership will fund, and a first project with no visible sponsor tends to lose its budget by the second quarter. Take the loud problem when it also scores well on that list, and take it anyway when it is cheap and ignoring it costs you political capital you need later. The failure mode is letting complaint volume do the ranking, because the people absorbing the worst of it stopped complaining years ago. You get roughly one of these before people stop believing modernization produces anything, which is why the first pick carries more weight than the second.

‍

What makes a good first modernization project?

A good first modernization project is frequent, crosses at least two teams, and changes something a person outside the department can see.

Frequency gives you the arithmetic. A task that takes forty minutes and happens twice a day eats close to seven hours a week, which is most of a workday that never appears in anyone's plan. The cross-team requirement keeps you honest, since a problem contained inside one department is usually a preference and a problem that crosses two is usually a handoff. As for visibility, a customer or a provider noticing the difference will do more for your next budget than an internal time saving only operations can feel.

Leland Little runs multiple online auctions a week in a business where, in their words, every second a problem lasts is a lost bid and lost revenue. We started at checkout, because that is where the money stopped. We modernized the platform by adding new integrations, implementing a completely new checkout process, and upgrading systems to support their transition to a complete online solution. Their Director of Operations, Jessi Yousko, says they value "the level of expertise, service, and support they continue to provide," and Big Pixel is their technical backbone now, supporting all major systems.

We believe that business is built on transparency and trust, and that good software is built the same way, which on a first project means choosing the workflow your team can feel inside a month.

‍

What can custom software fix without replacing the system?

Custom software can fix a handoff without replacing the system of record underneath it, usually with one tool that absorbs the step people are doing by hand.

Depending on where the work stalls, that looks like:

• A status view that answers the question customers, members, or providers keep calling about
• An intake queue that shows what arrived, what is missing, and who owns the review
• An approval workflow with ownership and escalation, replacing an email thread
• An integration between two systems that already hold the same record
• An exception queue for the cases the core platform has no status for
• A reporting layer that removes the manual cleanup before a number is trusted

The math on this is already public in health insurance. The 2025 CAQH Index put $258 billion in administrative costs avoided through automation in 2024, with $21 billion still on the table in transactions that remain manual or partly manual. That remaining money is spread across individual transactions rather than concentrated in one platform, which makes it exactly the kind of problem you solve one workflow at a time. CAQH chief executive Sarah Ahmad framed the pressure as reducing administrative costs "without disrupting care," and replacing a platform is the most disruptive way available to go after it.

‍

What should you do before funding a modernization project?

Before funding a modernization project, pick one workflow and follow a single record through it end to end, writing down every wait, export, re-key, and phone call it takes to get from the first system to the last.

That walk usually produces a shorter list than expected, and the first item on it is normally a handoff rather than a platform. Fix that one, measure what it returns, and use the result to fund the next. A modernization roadmap built from three completed projects is a different conversation with your board than one built from a vendor's estimate.

AI belongs after this work. A model can summarize a claim, classify a document, or suggest a route, and all of that depends on a source of truth, defined owners, and statuses the business has actually agreed on. A workflow still held together by exports cannot supply any of the three, and adding AI to it produces faster versions of the same confusion.

Your team already knows which workflow to pick. Ask the person who has been doing the workaround longest, and believe the answer.