How to Build an OMO Strategy: Customer Needs, Pilots, Costs, and Evaluation
Shusaku Yosa
An OMO strategy selects a customer problem across touchpoints, tests a manageable solution, and expands investment only when the evidence supports it. Start with the task, baseline, owner, cost, and evaluation method rather than choosing an app or customer-data platform. Installation is not the outcome; reliable completion and sustainable operations are.
Separate observed needs from assumptions
Collect difficulties such as uncertain availability or unclear collection status through observation, interviews, and service records. Label staff assumptions as hypotheses. The UK government’s guidance on learning user needs illustrates understanding people’s behavior and difficulties before designing a service. It is a research reference, not an OMO certification standard.
Consider customers who need assisted or non-digital routes. A single journey need not serve everyone, and more integration is not always justified when the existing problem is small compared with ongoing costs.
Complete an investment worksheet
Decision | Record | Evidence |
|---|---|---|
Audience and problem | Who struggles, where, and with what task? | Interviews, observation, service records |
Baseline | Time, failures, questions, costs | Consistent population and period |
Change | Stores, products, journey, owner | Smallest workable scope |
Success and limits | Completion, quality, workload, cost | Denominators and review date set in advance |
Risk and fallback | Shortages, outages, incorrect information | Owner, contact, stop and manual procedures |
Expansion or stop | Conditions for the next decision | Pilot records and remaining uncertainty |
Use retail and service OMO examples to explore options, then adapt them to capacity, staff time, and geography. Set criteria from user needs and business constraints rather than treating a generic benchmark as mandatory.
Use the minimum necessary data and technology
An order reference and existing register may be sufficient for an early test. Decide what information is needed, when it changes, where the authoritative record lives, and who can access it. CDPs, CRMs, and dedicated apps are options, not prerequisites for OMO.
When personal information is used in Japan, review purpose, safeguards, and sharing against the Personal Information Protection Commission’s general guidelines. Do not merge different-purpose records merely because integration is convenient, or infer identity from superficial matches. Decide what happens to test records afterward.
Assign decisions at each stage
Stage | Work | Responsibility and gate |
|---|---|---|
Prepare | Baseline, terms, measurement, exceptions | Accountable lead checks readiness |
Limited pilot | Restricted locations, products, and dates | Staff record failures and workload |
Evaluate | Completion, quality, costs, unresolved problems | Operations, planning, and information owners review |
Expand or revise | Add one scope element or restore the earlier process | Monitor new workload and recurring problems |
Without fallback handling, a screen can accept an order the store cannot fulfill. Define decision handoffs as more teams join. For daily information and fulfillment responsibilities, see omnichannel operating design.
Keep a fictional budget separate from ROI
Suppose a fictional two-month pilot has JPY 300,000 of setup and production costs and JPY 50,000 of monthly operations. Its budget is 300,000 + 50,000 × 2 = JPY 400,000. State whether internal labor is included and check training, maintenance, rework, shared equipment, and exit costs.
That amount is a budget, not ROI. Without verified benefits or savings, you cannot establish payback. Record an expected increase in repeat visits as a hypothesis and specify what would have to be observed before translating it into money.
Review comparable conditions before expanding
- Match populations, dates, units, and completion conditions.
- Log seasonality, holidays, pricing, advertising, product mix, and staffing.
- Use a feasible comparison, but do not claim causality from unlike groups.
- Inspect delays, cancellations, questions, and staff time alongside completion.
- Resolve safety and quality issues first; extend or reduce a test when value remains unclear.
A shared journey objective does not eliminate channel-level diagnostics. Keep measures that reveal entry-point failures, store capacity, or delivery problems, while avoiding local optimization that harms overall completion. Use what the pilot establishes, including its limits, to decide the next investment.