Back to all articlesBusiness software decisions

Decide Whether to Replace Software or Fix the Buying Assumptions

Alactic Editorial10 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

A team asks for new software because reporting takes too long, records are incomplete, and users keep working in spreadsheets. The request may be justified. But those complaints do not yet establish that another product would solve the problem.

Replace business software when a documented, important requirement cannot reasonably be met by keeping, reconfiguring, or narrowing the current system—and when a credible replacement path justifies the switching effort. Defer procurement when the main causes remain uncertain or would travel with the team into a new product.

The useful question is not simply whether people dislike the current tool. It is: which failure would disappear after a switch, what evidence supports that belief, and what would it cost to find out? A replacement decision worksheet makes those assumptions visible before a buying process gathers momentum.

1. Separate the complaint from the replacement claim

A complaint describes an experience. A replacement claim asserts a cause and a remedy. Treating them as the same thing is how a difficult week becomes a procurement project.

For each complaint, write down the task, the observed failure, its consequence, and the evidence available. Then state the replacement claim separately.

Hypothetical example: “Reporting is terrible” becomes “The operations analyst manually reconciles account ownership before producing the weekly pipeline report, delaying its release.” The replacement claim is “A different reporting product would remove that reconciliation.”

That claim might be true. It might also be false because account ownership is recorded inconsistently, regardless of reporting software. The next step is not to dismiss the complaint; it is to distinguish the explanations.

  • Product constraint: A necessary task is unsupported or cannot be completed acceptably.
  • Configuration constraint: The current setup obstructs a task the product can support.
  • Scope mismatch: The system has been assigned work beyond what the business actually needs it to do.
  • Operating assumption: The purchase depended on inputs, ownership, access, or decisions that never materialised.

These are hypotheses, not verdicts. More than one can explain the same complaint. A product limitation can coexist with poor inputs; fixing either alone may leave the result unsatisfactory.

2. Recover the assumptions behind the original purchase

Before judging the software, recover what the buying decision expected to be true. Use the original requirements if they exist, but do not treat them as permanently correct. Interview the people who now depend on the system and record where their needs differ.

Look especially for assumptions that were never assigned an owner:

  • Someone would supply complete, consistent source data.
  • Different teams would use the same definition of a customer, owner, or completed task.
  • The required capability would be available in the purchased plan.
  • A connected system would expose the information needed.
  • Users would need only a small subset of the features purchased.
  • Reporting would reflect decisions the business had not yet standardised.

Check current capability and plan availability rather than relying on an old sales conversation or a general feature page. Equally, do not assume a feature’s existence proves it fits the task. Availability, suitability, and successful use are separate questions.

The output should be a short list of assumptions marked confirmed, contradicted, or untested. If a replacement proposal relies on the same untested assumption, flag it before comparing vendors.

3. Use a filled worksheet, not a list of grievances

Fictional worked example: The following company, observations, quantities, thresholds, and decisions are hypothetical. They illustrate a decision method, not customer results or product capabilities.

A business services team wants to replace its customer-management system. Its decision owner is the operations lead. The bounded decision is whether to begin procurement, not which vendor to buy. The team gives itself ten working days to test the strongest complaints.

Its hypothetical essential requirements are a usable weekly pipeline view, dependable retrieval of active customer records, and visibility of renewal ownership. It prefers broad activity reporting but has not established that this is essential.

Worksheet entry A: “The weekly report cannot be trusted”

  • Observed failure and consequence: The analyst spends four hours reconciling the weekly pipeline report before managers use it.
  • Evidence: A review of twenty disputed records finds twelve with inconsistent ownership or stage entries and eight whose source is still unexplained.
  • Replacement claim: A new reporting engine will eliminate reconciliation.
  • Alternative explanation: Inconsistent source entries may be driving much of the disagreement. The unexplained records may still reveal a reporting defect.
  • Low-cost validation: Agree the expected result for the twenty-record sample, correct only the known inconsistencies in a safe test copy, and compare the existing report with that expected result.
  • Hypothetical test result: Eighteen records reconcile; two still differ and require a bounded capability check.
  • Decision: Defer procurement on this complaint. Test the remaining discrepancy before calling it a product constraint.

Worksheet entry B: “Finding a customer is too slow”

  • Observed failure and consequence: Staff repeatedly leave the system to consult a separate customer index.
  • Evidence: In a fictional sample of ten retrieval tasks, six begin with an unfamiliar default view. Four involve incomplete customer identifiers.
  • Replacement claim: Another interface will make every customer easy to find.
  • Alternative explanation: View configuration may explain some failures; incomplete identifiers may explain others.
  • Low-cost validation: Check whether the current workspace and plan allow a suitable view. If available, test that view against the same retrieval tasks, keeping the underlying records unchanged.
  • Hypothetical test result: Eight tasks succeed without the separate index. Two remain blocked by incomplete identifiers.
  • Decision: Reconfigure rather than replace for this requirement. Record the remaining input problem separately.

Worksheet entry C: “Renewal tracking needs another spreadsheet”

  • Observed failure and consequence: A coordinator copies renewal dates into a spreadsheet to identify the responsible owner.
  • Evidence: The team can store dates, but its capability check finds no supported way in the current product to produce the required ownership view.
  • Replacement claim: A different system could remove the recurring copy-and-check task.
  • Alternative explanation: The requirement may be narrower than a full replacement, or ownership may still be ambiguous even with a better view.
  • Low-cost validation: Define the expected output using sample records with agreed owners. Confirm the current limitation and use that same sample as a candidate acceptance task.
  • Hypothetical test result: Ownership is clear in the sample, but the current product cannot produce the essential view.
  • Decision: Proceed to bounded procurement research for this requirement, while keeping a narrower solution in consideration.

Worksheet entry D: “We paid for activity reporting nobody uses”

  • Observed failure and consequence: Managers request detailed activity summaries but make no identified decision from them.
  • Evidence: The requirement review cannot identify a necessary action that depends on the summary.
  • Replacement claim: Easier dashboards would make the reporting useful.
  • Alternative explanation: The requested output may not be needed.
  • Low-cost validation: Ask the report sponsor to name the decision, required detail, and consequence of not producing it.
  • Hypothetical test result: A basic exception list meets the stated need.
  • Decision: Reduce scope. Do not carry the unused reporting requirement into a new purchase.

The worksheet produces a mixed answer: reconfigure one area, reduce another, investigate a residual issue, and begin limited procurement research for one confirmed gap. That is more useful than forcing every complaint into a single keep-or-replace verdict.

4. Compare four options against the same failures

Evaluate keeping, reconfiguring, reducing scope, and replacing using the same essential requirements. Do not compare the current system’s observed weaknesses with an imagined replacement that has none.

  • Keep: Appropriate when the task is workable and the residual burden is tolerable. Its cost includes recurring manual work and known limitations, not just the subscription.
  • Reconfigure: Appropriate when a bounded, available change addresses the failure. Include setup, verification, maintenance, and the chance that the change affects another task.
  • Reduce scope: Appropriate when an expensive or troublesome requirement is not essential. State what the business will stop expecting; otherwise the discarded requirement will return later.
  • Replace: Appropriate when an important gap survives validation and a candidate can demonstrate a credible path to meeting it. Include transition effort and requirements that could regress.

In the fictional worksheet, the customer-retrieval complaint supports reconfiguration. The renewal gap supports procurement research. Neither finding proves that replacing the whole system is preferable. The next comparison must establish whether solving renewal ownership separately creates less burden than moving everything.

5. Make switching effort visible before authorising a search

A replacement request often compares current subscription cost with a prospective subscription. That leaves out the work required to reach a usable state.

Create an effort inventory without pretending it is already a precise estimate:

  • Data: What must be exported, cleaned, mapped, imported, and checked? Which history or attachments need preservation?
  • Dependencies: Which reports, connections, forms, or downstream tasks rely on the current system?
  • Access: Who needs access to what, and who will verify that the new arrangement is appropriate?
  • Overlap: Will both systems run temporarily? Who resolves conflicting updates?
  • Acceptance: Who checks that essential tasks work with representative records?
  • Exit and recovery: What must remain available if the move stalls or a candidate fails?

For each item, record an owner, an estimate range, and confidence. Mark unknowns explicitly. Uncertainty is not a reason to invent a reassuring total.

In the hypothetical example, the team estimates eight to twelve staff-days for a full switch, excluding unresolved connection work. It estimates one to two staff-days for the retrieval-view change and its checks. These fictional ranges do not settle the decision, but they show why a confirmed improvement in one area does not automatically justify replacing everything.

6. Set the procurement gate before seeing demonstrations

Write the gate while attention is still on the business problem. Otherwise an appealing demonstration can quietly replace the original requirement.

A proceed decision should identify:

  1. The essential failure that remains after low-cost validation.
  2. The evidence that keeping, reconfiguring, or reducing scope is insufficient.
  3. The representative task a candidate must complete.
  4. The acceptable switching-effort range and unresolved dependencies.
  5. The condition that would stop or narrow the search.

Hypothetical decision record: “Proceed with a bounded search for renewal ownership visibility. Do not authorise a full-system replacement yet. Candidates must produce the agreed ownership view from representative sample records. Compare a narrow solution with full replacement. Stop the full-replacement route if it cannot preserve essential retrieval and pipeline tasks within the agreed effort range.”

A defer decision also needs specificity: the unresolved question, its owner, the evidence required, and the review date. “Wait until people use the system properly” is not a testable decision.

7. Handle inconclusive tests without turning them into approval

A small test can fail for reasons unrelated to the hypothesis. The sample may be unrepresentative, the expected output may be disputed, or the person running it may lack access. Record these as test limitations rather than proof that either product is good or bad.

If reconfiguration fails, preserve the original setup where feasible, record what changed, and return to the complaint evidence. Do not expand into an open-ended troubleshooting exercise. The procurement decision needs a bounded answer about feasibility, not an exhaustive repair programme.

If the current product meets the requirement but dissatisfaction remains, investigate whether the requirement captured the real task. If a candidate fails the acceptance task, revise the shortlist or defer; do not quietly weaken an essential requirement to preserve buying momentum.

Urgency can justify a temporary workaround while evidence is gathered. Give it an owner and an expiry condition. Otherwise a temporary measure can conceal the failure indefinitely, making both keeping and replacing harder to evaluate.

8. End with a decision someone can revisit

The final record should name the chosen route, rejected alternatives, supporting evidence, remaining uncertainty, and the event that would reopen the decision. A decision can be sound without being permanent.

For the fictional team, the immediate outcome is to retain the current system, make the validated view change, drop unnecessary reporting scope, and authorise limited research into the confirmed renewal gap. A full replacement remains conditional rather than approved by accumulated frustration.

For your next replacement request, select the most consequential complaint and complete one worksheet entry before scheduling vendor demonstrations. Assign an evidence owner, a bounded validation step, a review date, and a pass-or-fail condition. At review, measure whether the stated task improved and whether the proposed cause was supported. That creates a defensible reason to begin procurement—or a concrete reason not to begin it yet.

Software procurementBusiness software decisionsReplacement planningDecision worksheets

Put the next conversation to work.

Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.

Explore Communications