Skip to main content
Subscribe to product updates via RSS, or watch the What’s new feed inside Retailgrid.
FeaturesImprovementsBug Fixes
Cycle 24
Summary: A cycle about closing the loop and making the new Home dashboard fast and dependable. Price Optimization starts measuring whether approved prices actually hit their goal, the Home dashboard loads progressively and survives a competitor-data outage, the Runs view gets summary cards you can filter by, and Price Approval gets a cleaner actions layout.

Features

  • Did the price work? Goal attainment on approved prices — after an approved price goes live, Retailgrid now measures whether it hit its profit or revenue goal. It’s the first step toward a closed suggest, execute, measure, learn loop for Price Optimization.

Improvements

  • Faster, more resilient Home dashboard — the grids list appears immediately while the dashboard loads in the background from a prepared snapshot, and the dashboard still shows your sales and margin KPIs when competitor data (Price Radar) is temporarily unavailable. See Home dashboard.
  • Richer Runs view — the Runs section gains summary cards that double as filters (click the failed-processes card to see only failed runs), plus pagination. See Runs.
  • Clearer Price Approval actions — the actions area in Price Approval is redesigned for a cleaner, less cramped layout. See Review and approve price changes.

Bug Fixes

  • More trustworthy base forecast — the sales forecast no longer accepts demand levels far above an item’s own sales history, so a handful of items can’t be projected unrealistically high. See Sales forecast.
  • Fixed overlap in Price Approval — the section title no longer overlaps the Impact button.
  • Clearer dashboard error message — when the Home dashboard can’t load, it now shows a specific, readable message instead of a generic one.
FeaturesImprovementsBug Fixes
Cycle 23
Summary: A cycle about seeing your position at a glance and running price monitoring as a managed service. A new Home dashboard rolls pricing health across every grid into one view, Competitive Position breaks your price index down by competitor and by product role, price monitoring moves to a request-from-Settings setup, and competitor-match review gets sharper filters.

Features

  • Home dashboard — opening Retailgrid now leads with a home view: pricing-health KPIs, a revenue trend, and your grids beneath, so you can tell where to spend your attention before opening a grid. See Home dashboard.
  • Price index by competitor and by product role — the Competitive Position dashboard now breaks your index down per competitor and per product role, so a healthy blended CPI can’t hide a weak position on your KVIs. See Competitive Position.
  • Recent price-change flags — three grid columns flag whether a product’s regular price changed in the last 7, 14, or 30 days, so you can avoid repricing the same SKU too often. See Metric columns.
  • Push competitor data to Price Radar over the Public API — send raw or already-matched competitor data to Price Radar programmatically, then poll the publication job for counts and row refusals. See Publish competitor data to Price Radar.
  • Google Cloud Marketplace — subscribe to Retailgrid through the Google Cloud Marketplace and bill it against your existing Google Cloud commitment. See Billing.

Improvements

  • Request price monitoring from Settings — competitor price monitoring is now set up for you by the Retailgrid team. Describe what you want watched in Settings → Price monitoring instead of configuring it per grid, and track your request’s status. See Request price monitoring setup.
  • Base Price Performance moved to Modelling — BPP settings now live under Settings → Modelling with the other account-level model parameters, instead of in the grid Agents view; they still recalculate automatically after every model update. See Modelling.
  • Maximize Revenue drops the built-in cost floor — the Maximize Revenue goal no longer forces prices to stay above unit cost, so on a very elastic SKU it can propose a price below cost; add a margin-floor rule if you need a hard lower bound. Maximize Profit still respects margin by design. See Price Optimization.
  • Pick a confidence floor when reviewing matches — the competitor-match review queue’s confidence filter now takes a minimum you choose (for example 90% and above), not just fixed bands. See Review and approve competitor matches.
  • Competitor matching filters default to All — the Products, Cost, Price, and Stock filters now open on All on every tab, so the figures describe your whole catalog until you choose to narrow them. See Competitor matching.

Bug Fixes

  • Spanish UI fully translated on Competitor matching — the Competitor matching page no longer shows English labels when the interface language is Spanish.
  • Keyboard focus restored — controls that had lost their focus outline show it again, and an open dialog keeps keyboard focus contained.
FeaturesImprovementsBug Fixes
Cycle 22
Summary: A cycle about inventory-aware decisions and faster review. Price Optimization learns to balance margin against how fast stock turns, a new forecasted days-of-supply metric makes that visible in the grid, competitor matches move from a read-only dashboard to an actionable review queue, and the Public API gets two integration-friendly endpoints.

Features

  • Inventory-aware pricing with GMROI — Price Optimization adds a GMROI goal that balances margin against inventory velocity, with Days of Supply min/max guardrails (default 7 and 30 days) and optional end-of-life markdowns; it falls back to profit-only where inventory isn’t configured. See Price Optimization.
  • Forecasted days of supply metric — add a column showing how many days current stock lasts at today’s price, from the seasonal demand forecast. See Metrics glossary.
  • Batch product updates over the Public API — update many products in one call, each with only its changed fields, with explicit per-item outcomes. See Products API.
  • Latest-transaction lookup — a new GET /v1/transactions/latest endpoint returns the most recent transaction by business date, optionally within one store. See Latest transaction.

Improvements

  • Review competitor matches, don’t just read them — the Competitor matching page gains a review workspace: approve, reject, reopen or rematch individual matches, bulk-approve everything in the current filters, and set admin-level auto-approval. See Competitor matching.
  • Clearer Price Analysis — the recommended price reads Proposed Price throughout, and the panel now shows Regular Price as a reference. See Price Optimization.
  • Leave the Upload data page without the browser back button — the page now has an in-app back/close control.

Bug Fixes

  • Rematch no longer preselects “Direct” for a match whose type is not stated, so you choose direct vs. substitute deliberately.
FeaturesImprovementsBug FixesAPI
Cycle 21
Summary: A cycle about seeing more at a glance and configuring with less friction. The Grids home page now carries live performance, forecast, impact and approval numbers per grid, so you can triage without opening each one. The pricing agents share one cleaner configuration layout, new grids open lean, and your grid setup now follows the grid everywhere. Around that: supplier columns in grids, bigger and correctly documented transaction API pages, and clearer sign-in and password flows.

Features

  • Grids home shows live metrics — each grid row now shows historical sales, forecast at proposed prices, the impact if all proposals are approved, and its price-approval counts, with a metrics picker, timeframe control, sorting and a review filter. See Grids overview.
  • Supplier, Supplier Code and Manufacturer Code in grids — these product fields are now available as grid columns (leading zeros preserved), on existing grids after refresh and over the MCP column listing. See Configure grid and the Products data spec.
  • Store-aware imports over the public MCP — product/store-price and transaction imports now flow through the public MCP interface with canonical store identities, resolving to your default store when none is given. See AI connections.

Improvements

  • Price Optimization config, reorganized — the dialog now shows four always-visible tabs (General, Parameters, Rules, Automation) in a two-column layout with Config by AI on the right, and no Workflow panel. Rules are editable up front as guardrails on the recommendation. See Price Optimization.
  • Consistent agent dialogs — Dynamic Pricing and Price Monitoring now use the same two-column layout as Price Optimization. See Dynamic Pricing.
  • Leaner new grids — a new grid opens with a focused default column set for price review; everything else stays one click away in Columns. See Configure grid.
  • Your grid setup follows the grid — column layout, sorting, grouping and filters now save as one durable configuration in the backend, so they survive scope changes and appear the same across browsers, when shared, and when duplicated. See Configure grid.
  • Larger transaction API pages — the List transactions endpoint now serves up to 10,000 rows per page (default 1,000), and the reference docs were corrected to match the live contract.
  • Password composition rules with live checklists — every password-setting flow now shows a live checklist and enforces the same rules on submit. See Account.
  • Company name at signup — the company name used on generated artifacts is now captured during signup/onboarding. See Organization.
  • Clearer elasticity confidence — a SKU’s confidence label and elasticity source now describe the coefficient actually used; prices that fell back to a category/group prior are reported as such instead of as an item fit. See Price Optimization.

Bug Fixes

  • Grid layout no longer resets after a scope change — changing a grid’s scope keeps your configured columns instead of turning everything visible.
  • Rounding applies to optimizer-matched prices — a proposed price that exactly matches the optimizer target is now rounded to your rule’s endings unless a hard constraint (such as a margin floor) makes that impossible, and rows where rounding can’t apply are flagged in Price Approval.
ImprovementsBug Fixes
Cycle 20
Summary: A refinement and reliability cycle. The Forecast chart now explains sales through price and lets you compare scenarios, Price Approval tells you when and how each proposal was calculated, and the demand model treats slow sellers and out-of-stock weeks far more sensibly. Around that, one shared usage budget now governs every API and MCP call, and several MCP and data-integrity bugs are fixed.

Improvements

  • A Forecast chart that shows price and scenarios — the Forecast view now plots the SKU’s price alongside its sales, marks each price change with its direction and size, and compares the forecast at your current price against the forecast at the proposed price. See Sales forecast & price impact.
  • See when and how a proposal was calculated — the Price Approval header now shows when the current proposals were calculated and which agent produced them. See Review and approve price changes.
  • One shared usage budget across API and MCP — every API request and MCP tool call now draws on a single, attributable organization budget, with clear limit and reset information when you hit it. See Usage & limits.
  • Steadier forecasts for slow and out-of-stock items — the demand model no longer reads out-of-stock days as zero demand, slow sellers keep a sensible base forecast instead of collapsing to zero after a quiet stretch, and promo lift now covers more events and SKUs.
  • Clearer AI-assistant tool names — every Retailgrid MCP tool now carries a human-readable title, so it reads clearly in your AI client’s tool list. See AI connections.

Bug Fixes

  • No more duplicate sales from re-sent transactions — re-sending the same transaction lines no longer double-counts sales.
  • MCP optimization summary after a run — the optimization summary and its impact are available again over MCP once a run completes. See AI connections.
  • MCP Product Roles on a new grid — running Product Roles over MCP on a brand-new grid no longer fails with a bare “conflict.”
  • Base Price Performance after a model build — BPP is no longer blocked after a completed model build. See Forecast accuracy & BPP.
FeaturesImprovementsBug Fixes
Cycle 19
Summary: Last cycle your AI assistant could read your grids; this cycle it can act on them — run a Price Optimization, build and edit grids, and approve and publish prices, always and only after you confirm each step. Around that, approved prices gained a proper lifecycle with amendable effective dates, Price Radar now matches a raw competitor export so you don’t have to link rows first, and the pricing explanations got a lot more to say — per-rule corridor cards and the full price ladder behind a Price Relations price.

Features

  • Act on your grids from an AI assistant — a connected MCP client can now run or cancel a Price Optimization, create or duplicate a grid, add or edit columns and set cells, and approve, override, or publish prices — each only after you confirm it, with your own permissions. See AI connections and Connect your AI assistant.
  • Approved-price lifecycle and effective dates — follow each registered price through its lifecycle, see exactly which version a consumer received, and amend an effective date without rewriting what was already delivered. See Approved prices: lifecycle, history & delivery.
  • Price Radar matches a raw competitor export — publish a raw competitor file and Price Radar matches it to your catalog for you, so you no longer have to link every row to an item_id first. See Competitor matching.
  • Forecast accuracy & Base Price Performance — track how accurate the forecast has been week to week and the realized benefit of the price changes you already made. See Forecast accuracy & BPP.

Improvements

  • Price Analysis explains each rule — the panel now shows a compact card per applied rule with its priority and price step, and an expandable price corridor chart per rule, plus the elasticity behind an optimized price. See Dynamic Pricing.
  • The full price ladder in a Price Relations explanation — “Why this price” now shows every rung of the SKU’s ladder with current and proposed prices, flags inversions, and marks any rung held by another rule. See the Pricing rules guide.
  • Price Match can apply a formula to the reference price — multiply or shift the reference before matching, so you can price against a derived value like “80% of Regular Price when stock is low.” See the Pricing rules guide.
  • Price Relations compares within several attributes — keep only genuinely comparable products in one ladder by matching on extra attributes (model, storage, colour…), allow a gap range rather than one exact step, and preview the ladder before you run. See the Pricing rules guide.
  • Group by dynamic columns — group the grid by AI-derived columns such as Product Role, not just imported columns. See Configure grid.
  • Progress on the “Updating data” screen — a full grid update now shows a progress bar instead of an open-ended spinner.

Bug Fixes

  • Unchanged prices stay out of approval — a proposal that matches the current price no longer lands in the To Review queue across the remaining paths that still sent it there. See Review and approve price changes.
  • Stores competitor visibility restored — the Stores Competitors tab reads competitors and their store/zone/channel scope from current data again, fixing empty lists and stale scope counts. See Assign competitors to stores, zones, and channels.
FeaturesImprovementsBug Fixes
Cycle 18
Summary: This cycle was about reach and unattended trust. Retailgrid now answers to your AI assistant over a read-only connector, grids can be shared with teammates instead of rebuilt, and scheduled pricing runs can approve and queue prices on their own — but only inside limits you set, and they email you what happened. Alongside that, grid creation got a faster scope step with a rolling competitor window, and you can now remove a competitor price that shouldn’t count.

Features

  • Query your grids from an AI assistant — connect Claude, ChatGPT, or another MCP client to Retailgrid over a read-only connection and ask about your catalog and grids in plain language, no exporting. See Connect your AI assistant and manage connected clients in AI connections.
  • Share grids with teammates — grant named teammates in your account view-only or edit access to a grid, change it, or revoke it; shared grids appear alongside your own. See Share a grid.
  • Auto-approve prices in scheduled runs — let a scheduled Price Optimization or Dynamic Pricing run automatically approve and queue the prices that clear a minimum unit margin and a maximum price change you set, while everything else waits in Price Approval. See Schedule and automate pricing runs.
  • Email summary when a scheduled run finishes — opt in per schedule to get one email when a run completes, with what changed and a link straight to the results in Grid. See Schedule and automate pricing runs.

Improvements

  • Rebuilt grid-creation scope step — a pinned header keeps Create grid in reach, sections collapse behind live summaries, category/brand/competitor pickers search instead of listing everything, quick filters cover Active / in-stock / recently-selling, and live cards count SKUs, stores, competitors, and competitor coverage. See Create grid.
  • Rolling competitor observation window — set the competitor window as a rolling Last 7 / 14 / 30 / custom days that re-evaluates on every refresh, instead of fixed dates that go stale. See Create grid.
  • Remove a competitor price — delete a wrong or irrelevant competitor price straight from the Price Comparison popover, so it stops feeding the price range, market-data metrics, and optimization. The removal is permanent, and price monitoring re-collects the competitor on its next run if it’s still configured. See Remove a competitor price.
  • Four new product fields — Long Description, Supplier, Manufacturer Code, and Supplier Code are now first-class optional fields in the product data model, over CSV and the API. See Products.

Bug Fixes

  • Grid editing locks behave for private grids and same-user sessions — concurrent-edit locks no longer block your own second session on a private grid, and the editing lease was shortened to one minute so a stale lock frees quickly.
FeaturesImprovementsBug Fixes
Cycle 17
Summary: This cycle was about putting settings where they belong. Demand model parameters are account-level, not per-run, so they now live in their own Modelling tab instead of the Price Optimization dialog. The Datasets page stops pretending data only flows one way and splits into Inbound and Outbound. And duplicating a grid now carries the setup you spent time on, instead of just the rows.

Features

  • Seven inventory and pricing metrics — Value of Inventory, Cost of Inventory, GMROI, Sell-through %, Inventory Turnover, Average Selling Price, and Realized Discount % joined the curated metrics library, in every timeframe. See Metric columns.

Improvements

  • Modelling settings, admin only — confidence thresholds, clustering, recommendation corridors, seasonality, GLM regularization, data filtering, and model freshness moved out of the Price Optimization Advanced tab into Settings → Modelling. They belong to the account’s demand model rather than to one run, so a change applies to every future run and scheduled recalculation, and only Admins can make it. You can also rebuild the model on demand from there. See Modelling.
  • Datasets grouped by direction — the Datasets page now separates Inbound data you load from Outbound data Retailgrid publishes back, and names the Connector behind each dataset (CSV, API, or Shopify) so it’s obvious where a dataset came from. See Datasets.
  • Duplicate a grid with its full configuration — duplicating now copies the whole setup (added columns, formulas, metric columns, layout, saved views) onto a product scope you choose, instead of leaving you to rebuild it. See Duplicate a grid.

Bug Fixes

  • Matched-data upload reports what it refused — publishing a competitor scan no longer hides rejected rows or repeats work it had already finished. A delivery that is refused says so, and a publish already in flight blocks a second one instead of racing it. See Competitor matching.
  • Selling items no longer vanish from the forecast without a reason — an item left out of the forecast now reports why (no recommendation, low confidence, missing inputs, or no recent sales) instead of coming back blank. See Sales forecast & price impact.
FeaturesImprovementsBug Fixes
Cycle 16
Summary: This cycle was about trusting what the grid tells you. A new pinned totals row rolls every calculated metric up over your exact filtered set — summing what should sum, averaging what should average, and recomputing ratios like margin and CPI correctly. Three pricing behaviors that quietly produced no output — a Minimum change limit, a one-by-one Price Relations ladder, and a Maximize Revenue run — now do what the rules say.

Features

  • Grid totals row — a pinned row at the bottom of the grid aggregates every calculated metric over the exact rows your filters and scope select, recomputing as you filter or change the forecast horizon. Sums, averages, sales-weighted margins, and recomputed ratios (Gross Margin %, ASP, GMROI, CPI) each roll up correctly, and you can collapse it with a click. See Totals row.

Improvements

  • Cleaner grid and totals styling — colored cell backgrounds and red/green text are gone from regular data rows; color accents now live only in the totals row, which reads as visually distinct from the last product row.

Bug Fixes

  • Price Change Limit (Minimum) now acts as a floor, not a block — a Minimum-type limit applies after the target price is set, so it caps small moves instead of suppressing every price change. See the Pricing rules guide.
  • Price Relations (one by one) produces changes again — a one-by-one ladder now emits gap-preserving proposals in production instead of returning zero changes. See the Pricing rules guide.
  • Maximize Revenue stays on the right side of the curve — the revenue goal no longer returns a cost-floored price whose own forecast lowers revenue; price safety and the revenue optimum are handled separately. See Price Optimization.
  • Honest no-data on inventory metrics — inventory metrics show a no-data state instead of a confident 0.00 or 100% when a product’s stock history is missing. See the Metrics glossary.
FeaturesImprovementsBug FixesTechnical Updates
Cycle 15
Summary: This cycle was about control and trust. You can now split your team into Admin and User roles, tell Retailgrid when your prices already exclude tax so margin and markup come out right, and rely on forecast and price-impact numbers that no longer blow up on thin-history items. The Product Role agent also got the clean, editable result view every other agent already has.

Features

  • Team roles — Admin and User — invite teammates as Admin (manage users, Organization defaults, and datasets) or User (build grids, run agents, work with data), change a role, or remove someone, all from the Users table. See Organization.
  • Prices exclude tax — a setting that tells Retailgrid your product prices already exclude tax, so margin and markup use the price as supplied instead of deriving a net-of-tax value. Built for US and Canadian retailers whose feeds are tax-exclusive. See Account.

Improvements

  • Cleaner Product Role results — the Product Role Classification agent now writes the role as a colored single-select badge you can edit in place, and hides the technical columns (Role Confidence, Role Source, Product Role Details) by default, matching every other agent flow.

Bug Fixes

  • Elasticity kept inside a plausible range in forecasts — extreme price-response coefficients no longer inflate the forecasted units, value, or gross-profit uplift shown in Price Approval and the grid export. See Sales forecast & price impact.
  • Accurate baseline on no-change items — an unchanged price now forecasts close to the item’s recent actual sales instead of overstating them, and dormant SKUs are no longer force-forecast. See Sales forecast & price impact.

Technical Updates

  • Decoupled pricing-model parameter calculation from the Price Optimization agent, so model builds run independently for more reliable agent runs. No user-visible change beyond steadier runs.
ImprovementsBug FixesTechnical Updates
Cycle 14
Summary: This cycle was about trusting the numbers at scale. The elasticity behind Dynamic Pricing and Price Optimization now leans on recent, real selling behavior and refuses to propose a price below cost or outside an item’s proven trading range. Underneath, the Market Data grid, metrics, and imports moved onto BigQuery, so everything stays responsive on large catalogs.

Improvements

  • Smarter, safer elasticity — the model now gives more weight to recent price changes and to how long each price actually sold, and every recommendation is bounded to the item’s observed price range and kept above cost. Dynamic Pricing and Price Optimization stop proposing increases a product has already proven will erode profit, and loss-leader pricing stays a deliberate choice rather than a side effect.
  • The grid and pricing models now run on BigQuery — the Market Data grid, metrics, and CSV / API imports run on BigQuery, so grids and imports stay responsive on large catalogs, and the pricing model input ceiling was raised to 10 million rows. See Datasets.
  • Group names in Price Analysis — the Dynamic Pricing Price Analysis panel now shows your category and group names instead of internal IDs.
  • More items earn a confidence tier — lowered the evidence threshold so more product groups clear the High and Medium confidence gates in Price Optimization.

Bug Fixes

  • Fixed newly-shown columns showing empty until you refresh or sort — a column’s values now appear immediately when you make a previously hidden column visible in the Market Data grid. See Manage columns.
  • Fixed the grid not refreshing after a price approval — the grid and the Price Approval queue now update after you approve and when you return to the tab.

Technical Updates

  • Moved grid, metrics, transaction reads, and imports onto BigQuery, and replaced the pricing workflow heartbeat with durable, event-driven job dispatch for more reliable agent runs. No user-visible change beyond the improvements above.
FeaturesImprovementsBug Fixes
Cycle 13
Summary: This cycle put more decision context in front of you before you commit a price. Every SKU can now be auto-classified into a product role, price approvals show the forecasted sales and profit impact of each change, and the app wears Retailgrid’s new trademark brand. Under the hood, demand and elasticity accuracy got a round of fixes so those forecasts are trustworthy.

Features

  • Product Role Classification — automatically tag every SKU with an explainable product role (KVI, Traffic Driver, Profit Driver, Niche, Tail, New) so pricing, promo, and monitoring can treat the SKUs that matter differently, without hand-maintained tags. See Product roles.
  • Sales-impact forecasts in Price Approval — each proposed price now shows the forecasted units, value, gross profit, and margin over a 7 / 30 / 90 / 180-day horizon, so you can judge the financial consequences before you approve. See Review and approve price changes and Sales forecast & price impact.

Improvements

  • New trademark brand across the app — the application now shows Retailgrid’s refreshed, trademark-clean logo everywhere the brand appears.

Bug Fixes

  • Fixed top-selling products getting no elasticity at any tier, so price optimization produces recommendations where they have the biggest profit impact.
  • Recalibrated base demand at item and group level so forecasts reflect each item’s actual recent sales rate.
  • Fixed the homepage grid row count not matching the row count shown inside the grid.
FeaturesImprovementsBug FixesAPI
Cycle 12
Summary: This cycle was about connecting Retailgrid to the systems around it and tightening store-level control. Shopify is now a first-class data source — connect a store and your products and sales sync in and refresh automatically, no CSV uploads. Store setup got bulk CSV import/export, competitors can be scoped to the exact stores, zones, and channels where they actually compete, and approved prices are now reachable through the public API. Alongside that, the pricing approval flow, the grid, and long-running jobs all got more livable.

Features

  • Shopify integration — connect a Shopify store to import your products and sales history, with automatic scheduled and event-driven sync so data stays fresh without manual CSV uploads. See Connect Shopify.
  • Import and export stores via CSV — create and update stores in bulk and download them back out, instead of adding each store by hand. See Manage stores.
  • Scope competitors to stores, zones, and channels — assign each competitor to the locations where it actually competes, so pricing rules and gap metrics compare against the right set per store. See Assign competitors to stores, zones, and channels.
  • Quick filters, saved in Views — show, configure, add, and remove filters from a single Filters menu, and keep them as part of a saved View.

Improvements

  • Override proposed prices in Price Approval — replace an agent’s proposed price with your own number inline, while still seeing the original suggestion and its margin impact. See Review and approve price changes.
  • Per-zone price relations — set the relationship between each pair of zones one by one in the Dynamic Pricing rule UI, instead of a single global gap.
  • Cancel running jobs — stop a long-running import, agent run, or grid fill from the Runs view.
  • Recall generated Pricing Strategy decks — generated decks now persist across sessions, so a refresh no longer loses the link. See Pricing Strategy.
  • Persistent column widths — columns keep the widths you set across reloads, and auto-size is now an explicit action rather than forced on every load.

API

  • Fetch approved prices via the public API — pull finalized pricing decisions into your ERP, webshop, or POS without a manual export. See the API changelog.

Bug Fixes

  • Fixed CSV / Excel export capped at 100 rows — exports now include the full filtered set. See Export data.
  • Fixed CSV export failing on the Datasets page. See Datasets.
  • Fixed Config by AI errors in the Agents and Formula columns. See Agents overview.
FeaturesImprovementsBug Fixes
Cycle 11
Summary: This cycle put pricing managers firmly in control. Agent-proposed prices now flow through a dedicated price approval workflow — review, adjust, bulk-approve, and publish in one click — and Price Optimization’s output was redesigned around that flow. The grid got more livable too: search any product by any attribute, and stop worrying about losing work, because layout changes autosave automatically.

Features

  • Added a price approval workflow — review, approve, reject, or override every price proposed by Dynamic Pricing and Price Optimization, with comments, filters, and bulk-approve, then publish in one click. See Review and approve price changes.
  • Search by any attribute in the grid — find a product by SKU, name, brand, EAN, supplier, category, or any custom field, without building a filter. See Configure grid.

Improvements

  • Redesigned Price Optimization output — recommendations now appear one value per cell, with a price type and effective date you set per recommendation and a clear explanation of why each price was proposed. See Price Optimization.
  • Grid autosave — filtering, sorting, grouping, and column changes save automatically with a status-line confirmation; the “Leave site?” warning is gone. See Configure grid.
  • Store-level pricing refinements — Sync Prices respects a chosen group column at the store-product level, and Price Relations can hold a structure across store attributes like pricing zones. See Dynamic Pricing.
  • Price Optimization clusterization — sparse-history SKUs borrow demand signal from clusters of similar items instead of dropping out.
  • More room for the grid — reclaimed vertical space so more product rows are visible at once.
  • Aligned data model and API docs on the Product vs SKU distinction — item_id is the SKU-level key, product_id groups related items. See Datasets.

Bug Fixes

  • Fixed grid list sorting on the home page — the list now sorts predictably with a created-date default. See Grids overview.
  • Fixed sorting and filtering on the image column, with a placeholder for products that have no image.
FeaturesImprovements
Cycle 10
Summary: This cycle shipped store-level pricing end to end. You can now define your stores, markets, and channels in a dedicated Stores section, load per-store prices and costs by CSV or API, work in a SKU-Store grid view, and run pricing rules that respect each store’s own competitor set and costs. Alongside it, Dashboards grew into a full editor — manual building and AI assistance on the same configuration — and the grid gained a time-series analytics panel that shows how any SKU’s price, competitors, sales, and profit have moved over time.

Features

  • Added Stores management — create and organize stores with country, currency, sales channel, and pricing zone, and group stores into pricing zones that share pricing logic. See Manage stores.
  • Added a SKU-Store grid view — toggle between Product view (one row per product) and SKU-Store view (one row per product-store combination) with store columns, store-aware filters, and metrics computed per store. See Product and SKU-Store grid views.
  • Dynamic Pricing now runs at the SKU-Store level — one recommended price per product-store combination, using each store’s own competitor prices and costs, with rules targetable by country, channel, or pricing zone.
  • Store-level data via the Public API — ingest endpoints accept store_id alongside product_id, with (product_id, store_id) upsert semantics and full backward compatibility for product-level payloads. See the API changelog.
  • Added a SKU time-series analytics panel in the grid — select a SKU to see Competitor Prices, Sales & Price, and Profit & Price over time (4–52 week windows, daily or weekly), with promo and below-cost markers and a profit-maximizing price view. See SKU time-series analytics panel.
  • Added the dashboard editor — one place to build and modify dashboards, combining a manual builder (searchable widget picker, drag & drop layout, auto-save) with a Dashboard Assistant that adds, removes, or replaces widgets from a plain-language prompt. See Dashboard editor.

Improvements

  • Rules Based Pricing is now Dynamic Pricing — same agent, same engine, clearer name. Existing configurations, runs, and output columns are unaffected, and old documentation links redirect.
  • Refreshed the Dashboards home page — built-in and custom dashboards in one list, with create, rename, and delete; built-in dashboards are now editable and can be reset to their defaults. See Dashboards.
  • Added global dashboard filters — Period, Category, Brand, and Competitor — applied consistently to every widget on the dashboard.
  • Metrics are now computed at the SKU-Store grain in BigQuery, so margin, price index, and competitor-gap figures reflect each store’s own price, cost, and competitor set, with a product-level rollup for Product view. See the Metrics glossary.
FeaturesImprovementsBug Fixes
Cycle 8
Summary: This cycle made Retailgrid feel more programmable and more analytical at the same time. The Public API gained outbound read endpoints — Products, Transactions, and a full Grid export — closing the inbound/outbound asymmetry that integrators have asked about since v1. Dashboards picked up an AI creation flow so analysts can describe what they want to see in plain language. And the grid grouping experience finally feels sensible out of the box: each column type now picks a meaningful default aggregation, and all aggregation runs on the backend so the numbers are correct even at full grid scale.

Features

  • Shipped read endpoints on the Public API — GET /v1/products, GET /v1/transactions, GET /v1/grids, and GET /v1/grids/{grid_id}/rows — for ERP and BI ingestion. The grid export includes canonical, custom, and formula/calculated columns. See the API changelog for the full list.
  • Added AI-driven dashboard creation — describe the dashboard you want in plain language from the Create with AI entry point on the Dashboards page, review the example prompts, and save the generated view. (v1 — single dashboard per prompt.)
  • Added Conditional Color Highlighting to grid columns — drive cell color from a formula result or value condition (e.g. IF({Margin} > 20, "High Margin", "Low Margin")), much like Excel conditional formatting. Configure it from the column header menu in Add Column.

Improvements

  • The Runs list’s Type column now shows the agent name (e.g. Pricing Strategy) instead of the raw action name. Easier to scan, matches how you think about your runs.
  • Smart default aggregation per column type in the grid — sales sum, prices and margins average, identifiers count. Override per column from the column menu when you need a different rollup.
  • Agents UI in Rules Based Pricing — separated Config by AI from the configuration tabs so it has its own surface above the tabs, and gave the Price Analysis panel a polish pass (consistent 2-decimal formatting, cleaner labels, a margin-impact card trio).

Bug Fixes

  • Grid grouping aggregation is now computed on the backend, so rolled-up values (Sum, Average, Count, Min, Max) are correct and consistent at any grid size — including large grids where the previous frontend implementation was unreliable.
FeaturesImprovementsBug FixesTechnical Updates
Cycle 7
Summary: This cycle made Retailgrid feel more responsive across three axes — language (a working English / Finnish switcher and a fully localized AG Grid chrome), data freshness (every grid now refreshes itself overnight, no manual replace needed), and observability (the in-app loader is now hard to miss). Under the hood, the static metrics that used to live in Cloud SQL now flow through BigQuery, so every metric — not just sales — accumulates a daily history.

Features

  • Shipped multi-language UI powered by Crowdin. A language switcher in Settings → Account lets you pick English or Finnish; the entire app, including AG Grid headers, filter menus, and pagination, follows your selection. Your own data (product names, categories) is never translated.

Improvements

  • Nightly grid refresh: every grid now refreshes automatically overnight with the latest data from your Datasets. Replaced datasets are picked up the next morning without any manual action.
  • The in-app processing indicator is now centered and more prominent — no more clicking around while a grid upload is still running. Only the agents that are actually in flight are shown in the loader.
  • Pricing Strategy runs now surface a Download button on the run progress card as soon as the deck is ready, so you can grab the Gamma link without drilling into Runs.
  • Static metrics like Margin and Current Unit Margin now keep a daily history in BigQuery alongside time-series metrics — enabling future analytics that compare how every metric has moved over time, not just sales.

Bug Fixes

  • In the Grid creation flow, pressing Back after a mapping error now returns you to the Mapping step with your selections preserved, instead of dropping you on the legacy upload screen. The legacy /upload route was also removed.

Technical Updates

  • All transactional tables now stream from Cloud SQL into BigQuery on an ongoing basis, consolidating analytics on a single warehouse. No user-visible change today; this is the foundation for upcoming time-series dashboarding.
FeaturesImprovements
Cycle 6
Summary: This cycle promoted the Pricing Strategy agent from prototype to product (real grid data, real account binding, full timeframe ladder), added Single Select / Status as a first-class grid column type, and tightened the Runs view with a default 7-day window and an active-jobs badge.

Features

  • Promoted the Pricing Strategy agent to product: it now reads your Account company name and runs on real Grid data via BigQuery — no more placeholder numbers — and supports seven timeframes (last 7 / 30 / 90 days, QTD, YTD, T12M, Custom).
  • Added Single Select / Status as a column type in Add Column, with auto-mapped colored pills — ideal for product roles, workflow states, and risk tiers.

Improvements

  • The Runs view now shows the last 7 days by default, with Last 30 days and All time controls when you need a wider window.
  • Added an active-runs badge to the Runs sidebar icon — a quick at-a-glance signal that something is still in flight.
  • Polished the Price Analysis panel in Rules Based Pricing: cleaner labels (Start Price, Applied Rules), consistent 2-decimal formatting, and a margin-impact card trio (new unit margin, pp change, relative change).
FeaturesImprovementsBug FixesAPI
Cycle 5
Summary: This cycle shipped the public REST API for data ingestion, surfaced pricing-rule traceability in the right Insights Panel, added richer column-type support to the grid, introduced the Pricing Strategy agent, and cleaned up the MVP UI ahead of the v1 demo.

Features

  • Shipped the public REST API v1 for Products and Transactions data ingestion, so retailers can load and maintain data programmatically without manual CSV uploads.
  • Added a Rules Based Price Insights Panel — click any calculated rules-based price in the grid to see the pricing trace, applied rules, and any violated rules for that row.
  • Added Analytical Metric as a third column type in the Add Column flow, so you can pick from a curated list of pre-built retail metrics (margin, revenue, price index, elasticity).
  • Added the Pricing Strategy agent — generates an AI-powered pricing strategy presentation directly from your retail data.

Improvements

  • Refreshed the Agents selection UI: removed Credits / Buy Credits placeholders and per-agent pricing, regrouped agents to match the new structure.
  • Removed credits, billing, and Team placeholder sections from UI Settings ahead of MVP.

Bug Fixes

  • Fixed the Group column disappearing from view after applying Group By in the grid.
  • Fixed columns (notably Product URL and Image URL) disappearing from the grid after running the Rules-Based Pricing agent.
FeaturesImprovementsBug FixesTechnical Updates
Cycle 4
Summary: This cycle focused on making Rules-Based Pricing faster, more reliable, and easier to use. We shipped AI-assisted rule setup, reusable action configs, several UX improvements, and fixes for large-grid and session reliability issues.

Features

  • Added AI-powered rule creation, so you can describe a pricing rule in plain language and review the generated setup before saving.
  • Added save and load for action configurations, making it easier to reuse pricing setups.
  • Improved rule setup with context-aware column selection and search, so only relevant fields are shown for each rule type.

Improvements

  • Refreshed the Actions UI to better match the Retailgrid design system and improve consistency across the flow.
  • Moved advanced rule parameters behind a cleaner expandable section to reduce UI clutter.
  • Improved app availability so the product stays responsive instead of needing to wake up after inactivity.

Bug Fixes

  • Fixed grid creation filters so they apply correctly and update previews as expected.
  • Fixed failures when working with very large grids, including datasets around 200k rows.
  • Fixed Rules-Based Pricing execution being blocked by another job.
  • Fixed rounding behavior in Rules-Based Pricing so rule order and rounding work together more predictably.
  • Added automatic logout after long inactivity for better session security and more reliable timeout behavior.

Technical Updates

  • Reduced memory issues in Rules-Based Pricing runs on large datasets.
  • Prevented database connection exhaustion in production to improve backend stability.
FeaturesImprovementsBug FixesTechnical Updates
Cycle 3
Summary: This cycle focused on making competitor pricing workflows much more usable and trustworthy. We shipped major Market Data UI improvements, added clearer pricing visuals, and fixed several pricing and data accuracy issues.

Features

  • Added a new competitor pricing view in the grid, so you can compare competitor prices directly in-row and open source links faster.
  • Added a visual Price Range column showing min/max competitor prices and where the current price sits in the market range.
  • Improved the Columns panel with grouped sections, search, expand/collapse, and bulk show/hide controls for faster navigation in large datasets.
  • Added column type icons to make editable, formula, and AI-generated columns easier to distinguish.

Improvements

  • Moved the Export button to a more accessible position in the toolbar.
  • Renamed “Load Data” to “Upload Data” for clearer terminology.
  • Improved competitor matching by prioritizing Product ID before falling back to SKU.

Bug Fixes

  • Fixed incorrect competitor price indicator colors so market position is shown accurately.
  • Fixed stale Zooplus competitor prices so the latest value is displayed.
  • Corrected competitor gap and undercut percentage calculations.
  • Fixed missing “Last updated” values in Datasets.
  • Improved rule-based pricing rounding so rounded prices stay within the feasible range.

Technical Updates

  • Migrated production infrastructure to Google Cloud.
  • Set up profiling and telemetry foundations to improve performance analysis and debugging.
FeaturesImprovementsBug Fixes
Cycle 2
Summary: This cycle shipped major usability improvements across pricing, data management, and grid workflows. You can now configure pricing rules visually, manage datasets directly in the app, and work with clearer grid controls and data signals.

Features

  • Rules-based pricing: Shipped a full rules configuration flow with support for multiple rule types, rule ordering, conditions, enable/disable controls, and action execution.
  • Datasets management: Added a dedicated datasets page for uploading, viewing, downloading, replacing, and deleting imported files, with status tracking and row counts.

Improvements

  • Added a derived promo flag (is_promo) to support filtering, rules, and downstream workflows involving promotional items.
  • Updated the grid to show filtered vs total row count, making it easier to understand active filters at a glance.
  • Made Add Column easier to access from the main menu.
  • Improved CSV import behavior by removing automatic grid creation, giving users more control during import flows.

Bug Fixes

  • Fixed import behavior so CSV uploads no longer create grids automatically in the background.