Skip to main content
The price approval workflow is where proposed prices become live prices. Both Dynamic Pricing and Price Optimization write their results into the grid as proposals - they don’t change your live prices on their own. You review each proposal here, then publish the ones you’re happy with. This keeps you in control: an agent does the heavy lifting across thousands of SKUs, and you stay the final decision-maker on every price that goes out.

​
When to use it

Reach for the approval workflow after any agent run that produced prices you intend to act on:
  • A Dynamic Pricing run wrote a new Rules Based Price column.
  • A Price Optimization run wrote Proposed Price recommendations.
  • You want a deliberate review-and-publish step rather than treating the agent output as final.

​
Open the review queue

After an agent run completes, switch to the Price Approval tab in the grid. Each row is one proposed change, with the product, its current price, the Proposed Price, the price change in money and percent, the margin impact, and the forecasted sales impact, so you can judge each change in context. Status chips across the top count the queue and filter it in one click:
  • All - every row in the grid.
  • To Review - proposals still waiting on you. This is where you work.
  • Approved - proposals you’ve accepted, staged and ready to publish.
  • Rejected - proposals you’ve discarded.
The Status column on each row tells you where that row stands: To Review, Accepted, Rejected, or No Change for a row where the agent proposed the price the product already has. No Change rows need nothing from you.

​
Sales-impact columns

Each row also shows the forecasted impact of the proposed price as the Δ Sales Units, Δ Sales Value, Δ Gross Profit, and Δ Gross Margin % columns, over a horizon you pick from the Forecast horizon control (7d, 30d, 90d, or 180d). Change the horizon and the columns re-compute and re-label to match. Use it to weigh a change on its financial consequences, not just the size of the move: a small cut that lifts forecasted gross profit can be the better call over a larger one that doesn’t. When you override a proposed price, the forecasted impact recomputes against your number, so you’re weighing the effect of your own edit rather than the agent’s original. See Sales forecast & price impact for how the forecast is built.

​
Run-level Impact Summary

Above the queue, an Impact Summary band rolls the whole run up into one view, so you can judge the batch before you work through it row by row. It reports the products in scope and how the run splits them - prices raised, cut, and unchanged - alongside the average change and the forecasted effect on gross profit, revenue, and units over the selected horizon. The band reflects the current filters and forecast horizon: filter the queue to a category or a confidence band and the summary re-computes for just that slice, so you can size the impact of the segment you’re about to approve. Use it as the sanity check before a bulk-approve - a batch that reads fine row by row can still add up to a larger move on gross profit than you intended.

​
Per-row actions

For each proposed change you can:
  • Approve - accept the proposed price as-is.
  • Reject - discard the proposal; the live price is left unchanged.
  • Override - replace the proposed price with a manual price of your own, then approve that.
  • Comment - leave a note on a row (why you rejected, what you overrode and why) so the decision is auditable later.

​
Filter the queue

Work through a large queue in deliberate passes rather than row by row. Start with the To Review chip, then narrow further with the grid’s own column filters, which work here like anywhere else in the grid:
  • Sort or filter on Price Change % to take the biggest moves first.
  • Filter on New Unit Margin % pp change to catch changes that push margin the wrong way.
  • Filter on product attributes (category, brand, store) to review one segment at a time.
  • Filter on Confidence to separate the recommendations the model is sure about from the thin ones.
Save a pass you repeat as a View, so next month’s queue opens on the same filtered slice.

​
Bulk-approve

Once a filtered set looks right, bulk-approve the whole selection in one action instead of clicking each row. A common pattern: filter to small, in-policy changes, bulk-approve them, then review the large or unusual moves individually. To let scheduled runs clear in-policy changes without a manual pass at all, turn on auto-approval in the Automation tab - it approves and queues the proposals that pass the margin and price-change limits you set, and holds everything else here for review.

​
Publish

When your approvals are ready, click Publish Prices. The button carries the count of staged approvals, so it reads Publish Prices (12) when twelve are ready to go. Publishing applies the approved prices as the live prices on those products. From that point the new prices flow through the rest of Retailgrid - metrics recompute against them, and they appear in exports and the Public API. Rejected and un-reviewed proposals are not published; only the changes you approved go live.

​
Common pitfalls

  • Forgetting to publish - approving a row stages it; nothing goes live until you publish. If prices didn’t change, check whether you published.
  • Bulk-approving without a filter - bulk-approve acts on the current selection. Filter first so you only approve the set you actually reviewed.
  • Overrides skip the agent’s guardrails - a manual override is your number, not the agent’s. Double-check it against your margin floor before publishing.