Retailers are juggling customer expectations, stock accuracy and tighter operating budgets at the same time. Oblytech provides IT services that improve the customer experience, simplify daily operations and support profitable growth.
Build or improve your online store with dependable e-commerce platforms, reliable payment gateways, and easy-to-use interfaces that help turn visitors into buyers.
Implement and integrate modern POS systems to streamline transactions, manage inventory, and gain valuable customer insights.
Utilize data analytics and AI to optimize inventory levels, streamline your supply chain, and reduce carrying costs.
Migrate your retail applications and data to the cloud for enhanced scalability, agility, and cost-effectiveness.
Build strong customer relationships, personalize marketing campaigns, and improve customer loyalty with Salesforce or custom CRM solutions.
Harness the power of your data to understand customer behavior, identify trends, and make informed decisions that drive sales and profitability.
Give shoppers a consistent experience across online, in-store and mobile channels, using integrated technology.
A retail operations director or an e-commerce lead usually reaches this question from the store floor or the fulfilment desk. A store reports that the system shows stock for a line that the shelf does not have. A customer returns a parcel bought online and posted back to a shop that cannot find the order. A supplier disputes a deduction for a delivery that arrived short, and the receiving record is in a different system from the invoice. These problems look like they belong to the commerce platform, but many of them originate in store execution: receiving, counting, returns handling and exception follow-up.
The commerce platform, the point-of-sale estate and the stock system each have a clear job. The commerce platform takes orders and manages the storefront. The point-of-sale estate records transactions in store. The stock system holds the official quantities and locations. A service workflow sits across them and owns the human response when those records disagree or when a process is interrupted.
| Process | Where it usually breaks in peak trading | Owner that should be named |
|---|---|---|
| Receiving and goods-in | Short or damaged deliveries are accepted without a discrepancy record | Store manager, with a head office receiving contact |
| Stock count and adjustment | Counts are adjusted in one system and not the other | Inventory controller for the estate |
| Online order collection and returns | An order cannot be found at the counter and the customer waits | Store duty lead with an e-commerce escalation route |
| Supplier deductions | A chargeback is raised without the receiving evidence | Buyer or supplier finance contact |
Retail processes are shaped by a simple constraint: the work happens across many sites at once, and many of those sites run with a small team that is also serving customers. A stock adjustment that takes real effort at a quiet counter is an interruption during a Saturday rush. A returns exception that requires a call to head office is a call that nobody wants to make at five in the afternoon. Designs that assume a calm office workflow tend to produce a record that is written up later, from memory, by a manager who has already moved on. The more useful question is what the minimum record is that a store can complete between customers, and what the head office does with it.
That minimum record usually has four parts: what happened, where, when, and what the store needs from someone else. Anything beyond those four is usually a refinement that can wait. A design that asks for a full root-cause analysis at the point of a short delivery is likely to be ignored, and a design that asks for nothing is likely to leave the head office unable to recover the money or the stock.
A stock discrepancy is one of the most useful cases to design first because it touches every channel. When a store’s count disagrees with the stock system, the inventory controller needs to know whether the count or the system is wrong, and whether the difference is caused by a recent sale, a delivery, a transfer or a theft. The workflow should record the count, the system figure, the time of each and the evidence attached, and it should route the case to a single owner who can approve the adjustment.
A common mistake is to let the store adjust the system directly, which removes the trail. The reverse mistake is to route every small difference to head office, which floods the controller. A sensible rule sets a tolerance below which the store can adjust with a reason code, and above which a case is raised. The tolerance should be agreed with the finance team and reviewed when the pattern of differences changes.
Returns are where the online and physical channels meet, and where the customer notices a failure fastest. A parcel returned to a store has to be matched to an order that may have been fulfilled from a different location, with a refund that depends on the payment method and the return reason. Each step has a different owner, and the customer’s patience is shortest at the counter. A workflow that gives the store a single case reference, shows the customer what will happen next and tells the head office what is outstanding is likely to remove much of the friction.
The exception categories should be kept few. A lost parcel, a damaged item, a mismatched order and an unauthorized return need different owners and different evidence. A design that sends all of them to one generic queue is likely to be slow on the cases that cost the most.
A useful conversation with a provider centers on how a store would use the workflow on a difficult day, not on the dashboard. The questions below are a practical set.
A first release works best in a cluster of stores with similar formats and one head office team that will own the outcome. It should cover one exception type, most often receiving discrepancies or stock adjustments, and it should run through a full trading cycle before anything else is added. The sequence below can be adapted to formats of different sizes.
The effort depends on the number of stores, the quality of the product and location master data, and the number of external suppliers whose deductions must be handled.
Loyalty and customer data are valuable to marketing, but the operations design rarely needs most of it. A return case needs the order reference and the customer’s contact method, not the full purchase history. A complaint needs the product and the store, not the loyalty tier. Keeping the operational record lean reduces both the data-protection exposure and the chance of a customer being contacted through a channel they did not choose. Where a customer relationship platform is part of the wider estate, the Salesforce service workflow design page is the related pointer for service case context and ownership. Keeping service cases distinct from the marketing record is a design choice worth making early.
Three ambitions are worth postponing. A full personalisation engine depends on clean customer data and on consent decisions that should not be made in the middle of a trading cycle. A rebuild of the stock system before the discrepancy process is trusted is likely to carry the discrepancies into a new environment. A move of the commerce platform to a new hosting arrangement is a separate decision, which belongs with the cloud consulting and migration route rather than with this first release. Each can follow once the store-level process is reliable, and deferring them keeps the first release from competing with the peak season.
No. Those platforms keep their roles. A service workflow manages the human response to exceptions and keeps the evidence trail, and it references the commerce and point-of-sale records rather than replacing them.
An inventory controller with authority over the adjustment should own it, with the store responsible for the initial count and evidence. Ownership without the authority to adjust is a common failure, because the case stalls waiting for someone to act.
The design should cut the minimum record to the essentials for the peak period, route urgent cases to a duty lead, and allow the store to defer non-urgent cases to a clear review slot. The rules should be agreed in advance, not improvised during the rush.
Not always. Loyalty and marketing needs can often be met by the existing commerce or loyalty tool. A customer relationship platform becomes worth considering when service cases, marketing history and sales activity need to be seen together for the same customer.
Fix first the area where staff or customers currently work around the system every day, because that is where money and trust leak fastest. For many retailers that is one of four areas: the online storefront, point-of-sale, stock accuracy or the links between them.
A simple decision rule helps. Ask what the last three complaints from stores or customers had in common, and match them to the list below.
| What people report | Usual area | First step |
|---|---|---|
| Orders or payments fail, pages are slow, baskets are abandoned | Storefront and payment | Trace one failed order end to end before changing anything |
| Queues at the till, prices differing between channels | Point-of-sale and pricing data | Agree one source for prices and promotions |
| Shelf and system disagree, stock is oversold | Stock records and counting routine | Fix receiving and count discipline before buying software |
| Data is right in each system but wrong between them | Integration | Map which system owns each field, then connect them |
The last row is the one that is often misdiagnosed. A new platform rarely cures two systems that disagree about who owns the stock figure. The cure is an ownership decision, then an interface built around it. Our API development and integration page describes how that connection work is usually scoped.
Analytics and forecasting typically pay off mainly when the underlying records are trusted. If a store cannot tell you what it holds today, a demand model is likely to produce confident-looking mistakes. Settle data ownership first; the data strategy consulting route then has something solid to build on.
Repair when the platform does what you need but the process around it is weak. Rebuild usually only when the platform cannot support a requirement you have written down, such as a channel you must sell through. Replacement feels decisive but often carries the same bad data across. If the choice is genuinely unclear, a short IT strategy and planning piece of work can compare the options against your own trading calendar.
Stock accuracy and returns handling, in most implementations, because they touch every channel and the cost of getting them wrong is visible within a week.
A retail case workflow is working when store staff raise cases at the moment the problem appears, the named head office owner acts within the time the store was told to expect, and the same discrepancy does not return in the same store the following week. Measure those three things across one full trading cycle before you add another exception type.
Start with the time from submission to the first action by the owner. Pull it separately for each exception category. A pooled average hides the cases that cost the most, such as a short delivery that is only queried after a supplier deduction has already been applied. Ask the owner to list the cases they picked up late and why, and record that reason on the case so the pattern can be seen later.
The second measure is how often a store closes a case without a call to head office. If stores raise a case for every small difference, the tolerance agreed with finance is probably too tight or has not been explained at the counter. If a store raises nothing for a whole month, compare its goods-in log with the case list before you assume the process is clean.
The third measure is repeat discrepancies. The same line showing a difference in the same store over several counts usually points to a receiving habit, a wrong location or a returns step that writes to the wrong record. One case rarely shows this. A monthly view by store, line and exception type does.
Review these measures in a short meeting at the end of each peak week. Date and record every change to the tolerance, the reason codes and the routing rules, so that a later difference can be traced to the change behind it. Check each field name, status value and report on your own instance and release before you build on them.
For the reporting side, Performance Analytics is one option, and service level management is where a time-to-action target is usually written down.