MCP App Store
by KeenEthics
ECDB icon

ECDB MCP Server for Claude and ChatGPT

By ECDB GmbH

Overview

ECDB gives you instant access to e-commerce market intelligence based on billions of real transactions and thousands of other sources. Query retailer revenues, market sizes, category rankings, and conversion KPIs across 100+ countries — directly in Claude. No dashboard switching, no manual exports. Just ask, and get the data you need to benchmark competitors, analyze markets, and identify growth opportunities. Powered by real transaction data, updated monthly.

How to connect

  1. Claude: Open the ECDB listing in the Claude connectors directory and choose Connect. Claude listing ECDB exposes 12 tools here.
  2. ChatGPT: Open the ECDB listing in the ChatGPT app directory and choose Connect. ChatGPT listing ECDB exposes 27 tools here.

Tools

aggregate_entities

ChatGPT
Aggregate one yearly metric over a set of entities of ONE type: sum, threshold filter, sort / top-N, or a weighted share. REQUIRED: entity and metric. Give the set either as identifiers (domains, slugs or region codes) or omit them and the tool takes the top_n (default 10, max 50) largest entities from the ranking for region / category — free, and the ranking's universe size ("4,482 stores ranked for France") comes back as summary.universe_size. Operations: - sum (default): combined value; summary.aggregate, one series per entity, ranked. "Europe minus Russia": entity=market, identifiers=[...], exclude=["RUS"]. Only amounts and counts add up (gmv, revenue, buyers, orders); shares, growth rates and indices are refused — combine shares with weighted_share, rank or screen rates with sort / filter. - filter: keep entities with value >= threshold; summary lists below_threshold and, for amounts and counts, the sum of the kept ones. - sort: descending by the metric, optional top_n cut, rank per entity. - weighted_share: the combined value of a share metric over the set, e.g. the online share of DEU+FRA+GBR as one market. The weight is the share's denominator as ECDB defines it and the tool picks it itself: total market for online_share (derived as revenue / share), visits for conversion_rate (derived as orders / rate), the order value for discount_rate (orders x aov_net_before) and return_rate (orders x (aov_net_before - discounts_per_order)) because both rates are value-based, population for internet_penetration; for add_to_cart_rate and cart_abandonment_rate orders stand in and the response carries the disclaimer "weighted with orders as an approximation" — say it. weight_metric only to override. Every weight metric is fetched per entity, and the answer names the weight ("weighted by order value, not a plain average"; summary.weight has the wording). Within a market ("stores in Austria above 100 m", "the German fashion GMV of the top 10"): metric filtered_gmv is the GMV allocated to the region / category filter (the figure the filtered ranking is based on; 1 credit plus one 4-credit share item per filter and entity). With a plain metric and a region / category filter the values are worldwide totals and the filter only picks the candidates; summary.applies_to says which of the two the sum, threshold or sort worked on — repeat it. Costs: every value is one API call (1 credit), so the set is capped at 50 entities — for "all stores in a market" take the top_n largest and say the sum covers the top N of universe_size, do not page through hundreds. exclude is applied before fetching. Values are for one year (default: current data year), one currency, one basis; entities that fail are listed in not_available and left out of the aggregate. Above the credit threshold the tool returns "Confirmation required" with the estimate instead of data — ask the customer, then repeat the call with confirm=true. ONE entity type per call. A retailer's figure already contains its stores, so never combine retailer and store values in one sum; compare them in two calls instead. For companies the tool reads the relations first (one item call per company) and leaves a subsidiary out when its parent is in the same sum (summary.subsidiaries_left_out says which): PDD Holdings plus Whaleco is PDD Holdings alone. A list drawn from the ranking carries summary.ranking_measured_as, the sentence that says what the ranking measured. Response is an output contract (JSON) followed by "Explore more" links; present the figures, not the JSON or the formulas (calculation_path is for checking and for customers who ask).

benchmark_entity

ChatGPT
Compare one metric of a store, retailer or shopper segment with a reference in one year — the tool for "is X above / below the market", "how does zalando.de compare with Zalando overall", "do Gen Z customers buy more often than all customers of the store". REQUIRED: entity, identifier, metric, reference. One metric per call (a KPI table is several calls). References: - market (entity store or retailer): the market of region x category (default WOR, all). Store-level metrics: aov (gross incl. VAT, compared with the market's aov_gross), conversion_rate; retailer: aov_gross, conversion_rate. purchase_frequency is not compared with the market (the market counts orders per buyer across all shops, a store its own orders per own customer): a store's purchase_frequency goes against the shopper panel, country required. Shopper-panel metrics of a store (retention_rate, repeat_rate, churn_rate, new_buyers, lifetime_revenue, purchase_frequency, purchase_interval, ...) are compared with the average across the panel's stores in the country (domain "all", not the market): country required. - parent_retailer (entity store): the store vs the retailer it belongs to, aov (vs the retailer's aov_gross, both gross), conversion_rate or purchase_frequency. - store_total (entity shopper_segment, identifier = store domain, country required, one of age_group / income_group): the segment vs all customers of that store in that country, any shopper-panel metric. A generation spans several ECDB age groups: one call per group. Returns both values, delta_abs, delta_pct (0-1 decimal), delta_pp (shares and rates only) and index (value / reference, 1.0 = reference level). year defaults to the latest year with a value on both sides (the shopper panel lags the current data year; summary.year_note says so) — name that year in the answer. Amounts and headcounts (gmv, buyers, orders) against the market are not available in this form; the tool says so. Costs: 1 credit per value and side (2-6 per call). Response is an output contract (JSON) followed by "Explore more" links; present the two figures and the delta, not the JSON (calculation_path is for customers who ask).

build_kpi_table

ChatGPT
Build a KPI table: entities as rows, several metrics as columns, one year — the tool for "give me a table of X, Y and Z with GMV, buyers and basket size", "the top 10 German fashion stores with their key figures", "DEU, FRA and GBR side by side". REQUIRED: entity and metrics (yearly series of the entity's get_* tool: gmv, aov, buyers, orders_total, purchase_frequency, conversion_rate, revenue, online_share, ...; max 10). Rows (source): - given (default when identifiers are passed): the customer's list, max 50. - ranking (default otherwise): the top top_n of the ranking of region x category (default worldwide, all, 10 rows); rows carry their ranking position. - competitors (stores only): identifiers = one store; the table is that store plus its top_n most similar stores. sort_by orders the rows by one column, descending; otherwise the source order is kept. Metric columns are always the entity's TOTALS (worldwide, all categories). A ranking filtered by region or category is ordered by the GMV allocated to that filter, so the table then adds the column filtered_gmv by itself (ECDB's filtered GMV, the same figure as calculate_gmv_breakdown) and summary.columns_note says the other columns are totals — say both in the answer. filtered_gmv can also be requested as a metric for a given list together with region / category. Returns kind "table": columns (one per metric with its own unit, currency and basis — the table mixes euros, counts and shares, read each column by its unit) and rows (values per metric, null where the entity has no value, listed in not_available; rank per column among the rows with a value; for rates and AOV columns the rows above five times the column median in columns[].outliers — amounts and headcounts carry none, a size ranking spreads by design). The table computes nothing else — no sums, no growth; for those use aggregate_entities / calculate_growth. Costs: 1 credit per cell (rows x metrics); filtered_gmv costs 5 (one filter) or 9 (region and category) per row because each share split is a 4-credit item, and it shares the gmv fetch with a gmv column; plus 4 for a competitors list (filtered_gmv works with every source). 50 rows x 10 metrics is the confirmation threshold. Response is an output contract (JSON) followed by "Explore more" links; present the table with units in the header, not the JSON.

calculate_gmv_breakdown

ChatGPT
Calculate absolute GMV for a specific country/region and/or category. Use this tool when users ask for regional or category-specific GMV that isn't directly available. It combines global GMV with share percentages to calculate filtered values — ECDB's "filtered revenue": the part of an entity's global figure allocable to a country and/or category. When to use: - "What's Amazon's GMV in Germany?" → country_filter="DEU" - "What's Amazon's GMV in EU?" → country_filter="E27" - "What's Zalando's fashion GMV?" → category_filter="fashion" - "What's Amazon's fashion GMV in Germany?" → country_filter="DEU", category_filter="fashion" - "Compare fashion retailers in Italy" → identifier=["amazon", "zalando"], country_filter="ITA", category_filter="fashion" Calculation: - Single filter: global_gmv × filter_share - Both filters: global_gmv × country_share × category_share Important: At least one filter (country_filter or category_filter) must be provided. Results are calculated estimates based on share percentages. Always mention this is a calculated estimate. Supports batch queries: pass a list of identifiers to compare multiple entities.

calculate_growth

ChatGPT
Calculate growth for stores, retailers, companies, markets, categories or brands on any timeseries metric — year-over-year, CAGR, indexed, percentage-point change, or quarter / YTD / season totals. REQUIRED: entity, identifier and metric. Use this tool for EVERY growth, CAGR, index or "how did X develop" question instead of computing from get_ values yourself: it returns the window actually used, missing periods and forecast flags. calculation_path documents the formula for your own checking and for customers who ask how a figure was computed — do not print it unprompted. `metric` is any timeseries metric of the matching get_ tool, e.g. store/retailer: gmv, aov, buyers, orders_total, conversion_rate; company: gmv, revenue; market: revenue, online_share, aov, buyers; category: gmv; brand: revenue. Info and breakdown metrics (payment_providers, gmv_by_country, ...) are not series and are rejected with the list of valid ones. Modes: - yoy (default): growth vs the previous year, or vs the same month a year earlier when start/end are YYYY-MM. Unit growth (a decimal, 0.10 = +10 %, 1.3 = +130 %). Metrics that are already growth rates (gmv_growth, revenue_growth, ...) are refused in yoy / cagr / indexed — use the base metric (gmv, revenue) instead; mode pp_change on them gives the change of the rate in percentage points. The year before start is fetched automatically so the first period has a value. summary.values_by_period carries the absolute values behind the rates (no extra get_* call needed for a table); for yearly windows summary also has start_value, end_value, total_growth and abs_change ("since 2020"). Monthly windows omit those totals because they would compare different seasons — pass window_totals=true to get them anyway. - cagr: compound annual growth rate between start and end (yearly only). Result in summary.cagr plus abs_change and yoy_by_period (the year-over-year change of every year, for tables); the series itself is returned in the metric's unit. Refuses if start or end has no value and says which years exist. - indexed: every value relative to base_period (default start) = 100. - pp_change: change of a share metric (online_share, conversion_rate, ...) in percentage points; summary carries values_by_period (the shares) and, for yearly windows, start_value, end_value and total_pp_change. - quarterly / ytd / season: totals of a monthly series (start/end as YYYY-MM); ytd runs January..through for every year, season sums season_months (e.g. [11, 12]). Incomplete periods are flagged partial_period. A window that reaches beyond the data (a series that starts in 2024, a forecast year not yet published) is not refused: the CAGR, index base and window totals use the years that have a value and summary.window_shortened says requested vs used — name the window actually used. Gaps in the data are never bridged: the period after a missing one is null and all gaps are listed in missing_periods. A period that has a value but no prior-year value (e.g. the first month ECDB tracks) is null with flags ["no_prior_period"] — that is not a data gap. The flag already reflects the API's answer for the prior year (the tool fetched it and got nothing), so do not re-check it with get_store/get_market: that call returns empty too. Years after the current data year carry flags ["forecast"]. benchmark_region (cagr and yoy) adds a benchmark block against the market of that region (a store's GMV is compared with market revenue). Defaults: window = ECDB's default year window, region WOR (category), category all (market). currency is required: growth rates of amounts are computed from the values in that currency and differ in another one (yearly average exchange rates; the web tool shows the same rate to a user with that currency setting) — say "in EUR" next to the rate. Supports batch: pass a list of identifiers of one entity type; failures are listed in not_available. Credits: 1 per yearly value, 4 per monthly value, per series (lagged modes fetch one extra year). Above the threshold the tool returns "Confirmation required…

compare_entities

ChatGPT
Compare one store, retailer or company with its peer group on one metric — the tool for "how does X compare with its competitors", "where does X stand among stores of its size", "compare X, Y and Z on their basket size". REQUIRED: entity, identifier, metric. One metric per call (for several metrics call it once per metric). metric is any yearly series of the entity's get_* tool (gmv, aov, buyers, orders_total, purchase_frequency, conversion_rate, ...). Where the peers come from (peers): - competitors (default for a store): ECDB's list of the most similar stores (assortment, audience, regional focus), the peer_count most similar. Stores only. - ranking_neighbours (default for a retailer or company): the entities ranked right above and below the entity in the ranking of region x category (default worldwide, all). summary.peer_source_note tells you that no similarity list exists for retailers / companies — say so and offer that the customer names the peers instead. - given: the customer names them in peer_identifiers. Returns every member's value (the entity first, then the peers by value), the entity's rank in the group ("3 of 11"), the peer median as benchmark with delta_abs, delta_pct and index, min / max, and outliers above five times the group median. summary.peer_source says in words which group was used — name it in the answer. With a single peer the comparison is head-to-head (the peer is the reference, no median); summary.peer_source_note then asks you to offer ranking_neighbours or more peers for a group view. Peers without a value for the year are listed in not_available, not silently dropped. Costs: 1 credit per member and year, plus 4 for the competitors list (about 15 for 10 peers). Response is an output contract (JSON) followed by "Explore more" links; present the ranked figures and the median, not the JSON (calculation_path is for customers who ask).

decompose_metric

ChatGPT
Decompose a headline figure into its drivers with a fixed formula — the tool for "why did it change" and "what is the lever". REQUIRED: entity, identifier, target_metric. Modes: - gmv (entity store or retailer): GMV = buyers x purchase_frequency x aov for year vs base_year (default: the year before). Returns each driver's growth and its share of the identity's growth (D5, shares sum to 1), plus ECDB's reported GMV with its own growth rate. - growth (entity market, identifier = region code): the market's revenue growth split by sub-category (G6); contributions add up to the total growth. - revenue_share (entity shopper_segment, identifier = store domain, country required, one of age_group / income_group): the segment's share of revenue vs its share of buyers and the spend index (D3; 1.0 = spends like the average buyer). A generation spans several ECDB age groups (Gen Z = age_under_20 + age_20_29): one call per group. - channels (entity store, retailer or company): the 1P / 3P split of GMV for year and base_year — share of own sales (1P = net sales + VAT) and of third-party marketplace GMV (3P) per ECDB's definition gmv = 3P + 1P net + 1P VAT (M2), plus each channel's contribution to the GMV growth (G6). The tool for "how much runs via third-party sellers / the marketplace", "1P vs 3P", "which channel drove the growth". What the numbers mean: contributions are shares of an arithmetic identity, NOT causes — say "the growth came from more buyers", never why there were more buyers. In gmv mode compare summary.identity.growth with summary.reported_gmv.growth (growth_gap_pp) before presenting the contributions as the explanation of the GMV growth; level_gap_pct is the distance of the identity from the reported GMV. The tool passes no verdict on the gap. Refuses (with the component named) when a driver has no value in one of the years. Not available: customer lifetime value (ECDB's own method — use get_shopper metric lifetime_revenue for the figure, do not rebuild it from AOV, frequency and retention), growth by country / category for stores (the split is a static snapshot) and brand channel decomposition (pending endpoints). Costs: 1 credit per component and year (gmv 8, growth 4, revenue_share 2, channels 8); above the credit threshold the tool returns "Confirmation required" instead of data — ask the customer, then repeat the call with confirm=true. Response is an output contract (JSON) followed by "Explore more" links; present the contributions and figures, not the JSON or the formulas (calculation_path is for checking and for customers who ask).

describe_brand_metrics

ChatGPT
Get detailed descriptions of available brand metrics (get_brand tool). Use this to understand what each metric measures, its unit, response shape, and credit cost.

describe_company_metrics

ChatGPT
Get detailed descriptions of available company metrics (get_company tool). Use this to understand what each metric measures — in particular reported revenue (online + offline, as published) versus gmv (the sum over the company's online stores, gross).

describe_market_metrics

ChatGPT
Get detailed descriptions of available market metrics. Use this to understand what each metric measures, what's included/excluded, and available granularity.

describe_retailer_metrics

ChatGPT
Get detailed descriptions of available retailer metrics. Use this to understand what each metric measures, what's included/excluded, and available granularity.

describe_shopper_metrics

ChatGPT
Get detailed descriptions of available shopper metrics (get_shopper tool). Use this to understand what each metric measures, its unit, and response shape.

describe_store_metrics

ChatGPT
Get detailed descriptions of available store metrics. Use this to understand what each metric measures, what's included/excluded, and available granularity.

get_brand

ChatGPT
Get brand analytics for one or more product brands: online revenue across every store the brand is sold via (or at one store), country/category mix, and the stores selling it. REQUIRED: both brand and metric must be provided. Do NOT call this tool with metric=null or omitted — pick a specific metric first. A BRAND is a product brand (Nike, Samsung, Coca-Cola) whose goods are sold through many online stores — its own shop, marketplaces, and wholesale partners. It is NOT the same as a retailer: get_retailer("nike") covers nike.com and its country sites; get_brand("nike") covers Nike products sold anywhere online, including amazon.com, tmall.com and zalando.de. Brand slugs come from ECDB's ~1,000-brand catalogue — resolve them via search_entities(query=..., entity_type="brand"); retailer slugs and store domains are not brand slugs. Valid metric values (all yearly; use YYYY for start/end): - info: Profile text — what the brand sells, positioning, channel mix, competitors, and which sub-brands are in/out of scope (4 credits; ignores domain) - revenue, revenue_growth: Online revenue of the brand's products and its YoY growth. domain="all" (default) = across every store; domain="amazon.com" = the brand's sales at that store. Series starts in 2024, so growth exists from 2025. 1 credit per data point. - revenue_breakdown: Country x category grid with both axis totals for one store or all stores; the all-stores variant adds a coverage ratio (4 credits) - revenue_by_country, revenue_by_category: Brand-wide revenue and share by country / by category (4 credits each; no domain) - top_stores: Stores ranked by how much of the brand they sell — the brand's channel ranking (0 credits). Rank only, no revenue; end picks the year, limit the number of stores (default 10, max 100). Follow up with revenue + domain=<store> for amounts. domain applies to revenue, revenue_growth and revenue_breakdown only. A specific domain that does not carry the brand returns "not found". Supports batch queries: pass a list of brands to compare brands, OR a list of domains (with one brand) to compare a brand's sales across stores — not both. Use describe_brand_metrics for detailed metric definitions. Response includes "Explore more" links to ECDB Brand Analytics pages.

get_category

ChatGPT
Get Category Explorer data for one or more product categories. REQUIRED: both category and metric must be provided. Do NOT call this tool with metric=null or omitted — pick a specific metric first. Covers the ~18,000-category Category Explorer taxonomy (levels 1-7), which includes niche product categories like "electric-toothbrushes" or "fashion-waist-packs". Resolve slugs FIRST via search_entities(query=..., entity_type="category") — they cannot be guessed, and values from get_category_filters are a DIFFERENT taxonomy that does not work here. Valid metric values: - gmv: Category GMV (yearly + monthly; use start/end in YYYY or YYYY-MM) - gmv_growth: Category GMV growth rates (yearly + monthly) - avg_price: Average price of products sold in the category (current snapshot, no time params) - num_brands: Number of brands selling in the category (current snapshot) - num_products: Number of products sold in the category (current snapshot) - subcategories: Browse the taxonomy — lists the category's direct children (with per-child child_count for drilling deeper) plus its parent chain. Use for "what X categories are available/exist" questions; costs no credits and ignores region/time/currency. Optional argument defaults (use when user doesn't specify): - region: "WOR" (Worldwide) - don't ask, just use it - currency: "USD" - time window: omit start/end for the default window ending at ECDB's current data year Supports batch queries: pass a list of category slugs to compare multiple categories. Response includes "Explore more" links to ECDB Category Explorer pages.

get_category_filters

ChatGPT
Get the category FILTER hierarchy (~242 categories): the valid category values for get_market, get_ranking, and calculate_gmv_breakdown. For the full ~18,000-category product taxonomy (niche categories like "electric toothbrushes"), use search_entities(entity_type="category") to find a slug and get_category for its metrics — those slugs are a different namespace and do NOT work as category filters here.

get_company

ChatGPT
Get company data for one or more companies. REQUIRED: both company and metric must be provided. Do NOT call this tool with metric=null or omitted — pick a specific metric first. Companies are legal business entities that own retailers. Company slugs are derived from the legal entity name, e.g. "amazon-com-inc", "alibaba-group-holding-ltd", "zalando-se", "nike-inc". The exact slug is hard to guess (legal suffixes, punctuation, regional subsidiaries like "amazon-eu-s-a-r-l") — resolve it first via search_entities(query=..., entity_type="company"). Slugs also appear in get_ranking(entity="company") results. Valid metric values (yearly data only - no monthly granularity): - gmv: Total GMV across all company's eCommerce operations - revenue: Total company revenue - gmv_growth: Year-over-year GMV growth rate - revenue_growth: Year-over-year revenue growth rate - revenue_net_1p: First-party (1P) net sales revenue - ecom_activity_revenue: eCommerce activity revenue — gross 1P sales plus 3P GMV generated on all in-scope online platforms operated by the company or its subsidiaries - gmv_by_category: GMV breakdown by product category (no time params) - gmv_by_country: GMV breakdown by country (no time params) Supports batch queries: pass a list of company slugs to compare multiple companies. Response includes "Explore more" links to ECDB web tool company profiles.

get_credit_balance

ChatGPT
Get the current user's ECDB API credit balance. Use this when the user asks about their remaining credits, account balance.

get_entity_profile

ChatGPT
One-call profile of a store, retailer, company, brand or market — the tool for "give me a profile of X", "what does ECDB know about X", "tell me about X", "overview of the German fashion market". REQUIRED: entity and identifier (domain, retailer slug, company slug, brand slug or region code; resolve companies and brands via search_entities first). Default: every section of the entity; pass sections to narrow. Sections: - headline (all): the key figures of year as a one-row table — store / retailer: gmv, gmv_growth, buyers, aov (retailer: aov_gross), purchase_frequency, conversion_rate; company: gmv, gmv_growth, revenue, revenue_growth, revenue_net_1p, revenue_gross_3p; brand: online revenue and its growth across all stores. - structure (all): country split and category split (top 5 plus "other", shares 0-1), the 1P / 3P split of GMV (gmv_split: share_1p, share_3p computed from ECDB's GMV components, M2); store / retailer also the source reference of the GMV figure (gmv_data_source). - customers (store): share of GMV by age group and by income group. - shoppers (store): the shopper panel in the store's largest country — retention rate, repeat rate, customer lifetime value, and per age and income segment the buyer share, revenue share and spend index (1.0 = spends like the average buyer). Present only when that country has a panel; then simply absent, do not comment on panel coverage. - competitors (store): the 5 most similar stores (ECDB similarity). - brands (store): the product brands with the most revenue on the store. - status (store: online status, latest published year, parent retailer; retailer: name, latest year, stores; company: company type, ultimate parent, subsidiaries, owned stores (count and first 5), headquarter). - info, top_stores (brand): profile text, and the stores selling most of the brand. - market (identifier = region code, category = classic filter category, default all): headline = revenue, revenue_growth, online_share, aov_gross (gross incl. VAT), buyers, orders_total, purchase_frequency, conversion_rate, return_rate, discount_rate of the market; history = revenue of the four years before year to two years after, with the year-over-year rate (G1) and forecast years marked; structure = sub-category shares (top 5); players = the 5 largest stores of the ranking with rank and universe size (amounts on request via build_kpi_table); cross_border = domestic vs import share, import sources and export destinations (whole region only, category all); providers = payment, shipping and shop software penetration (top 5 each); macro = gdp, population, internet_penetration; info = region and category names. Sections that do not exist for the entity are listed in not_available. The profile computes nothing except the spend index; for "is X above the market" use benchmark_entity, for "X vs competitors" compare_entities. Costs: about 45 credits for a full store profile, 20 for a retailer or company, 15 for a brand, 45 for a market. Response is an output contract (JSON, kind table with one row, the sections in summary) followed by "Explore more" links; present it as a structured profile with the units from the columns, not the JSON.

get_market

ChatGPT
Get market data for one or more regions and a category. REQUIRED: metric must be set to one of the values listed below. Do NOT call this tool with metric=null or omitted — pick a specific metric first. Supports batch queries: pass a list of regions to compare multiple markets in one call. Example: region=["DEU", "FRA", "GBR"] returns data for all three countries. Optional argument defaults (use when user doesn't specify): - region: "WOR" (Worldwide) - don't ask, just use it - category: "all" - for aggregate queries - currency: "USD" - time window: omit start/end to get the default window ending at ECDB's current data year; some metrics/entities have data from 2014 and forecasts several years ahead Valid metric values (use start/end in YYYY for yearly, YYYY-MM for monthly where available): - revenue, revenue_growth: Market revenue and growth (yearly + monthly) - aov, aov_gross, aov_growth: Average order value metrics (yearly only) - conversion_rate, add_to_cart_rate, cart_abandonment_rate: Funnel metrics - orders_total, buyers, purchase_frequency: Volume metrics - discount_rate, return_rate: Transaction metrics - subcategory_shares: Revenue split by subcategories (no currency) - online_share: eCommerce share of total retail - gdp, population, internet_penetration, vat: Macro indicators (no category needed) - cross_border_split: Domestic vs import sales percentage (no time params, no category) - import_share: Breakdown of imports by source country (no time params, no category) Use get_regions and get_category_filters first to find valid region/category values. Response includes "Explore more" links to ECDB web tool market reports.

get_ranking

ChatGPT
Get rankings for stores, retailers, companies, countries, categories, or target shopper groups. Entity types: - store: Top online stores (domains). Filter by region AND category. - retailer: Top retailer brands (aggregated across domains). Filter by region AND category. - company: Top eCommerce companies. Filter by region AND category. - country: Largest eCommerce countries. Filter by category only (no region). - category: Largest product categories. Filter by region only (no category). - shopper: Top stores among a target shopper group ("where do 20-29-year-old high-income Germans shop?"). REQUIRES region as an ISO alpha-3 country (no worldwide aggregate); takes age_group/income_group/year, no category/sort. age_group, income_group, and year are ONLY valid for this entity. For broad worldwide rankings, OMIT region instead of passing "WOR". For unfiltered rankings, OMIT category instead of passing "all". Valid sort fields: - store, retailer: filtered_total_gmv, global_total_gmv, global_1p_net_sales, global_3p_gmv, growth - company: filtered_revenue, global_revenue - country, category: filtered_revenue, total_revenue, online_share, growth "global" = the entity's total worldwide figure across all categories; "filtered" = the share of it allocated to the region/category filters you pass (ECDB's "filtered revenue"). With no region and no category the two are identical. Use the default sort when users ask for the "largest" or "top" entities. Do not use sort="revenue" for store/retailer rankings; those entities do not expose a generic revenue sort. Rankings CANNOT filter by: shop software, payment providers, shipping providers, revenue thresholds, or store type (1P/3P/Hybrid). Response includes "Explore more" links to ECDB web tool ranking pages.

get_regions

ChatGPT
Get region/country CODES for use in other tools (get_market, get_ranking, etc.). Returns codes like DEU (Germany), USA, E27 (EU), WOR (Worldwide), APA (Asia-Pacific). IMPORTANT defaults when user doesn't specify: - Use "WOR" (Worldwide) for global/unspecified market queries - Use "E27" for European Union, "APA" for Asia-Pacific, "GE7" for G7 Types: - countries: Individual countries with their group memberships - regions: Region groups (continents, economic blocs) with member counts - all: Both (default)

get_retailer

ChatGPT
Get retailer data for one or more retailers. REQUIRED: both retailer and metric must be provided. Do NOT call this tool with metric=null or omitted — pick a specific metric first. Valid metric values (use start/end in YYYY for yearly, YYYY-MM for monthly where available): - gmv, gmv_growth: Total GMV and growth (yearly + monthly) - aov_gross: Gross average order value (yearly only) - buyers, conversion_rate, purchase_frequency: Customer metrics - orders_total, orders_1p, orders_3p: Order volume metrics - revenue_net_1p, revenue_gross_3p, vat_amount_1p: Revenue breakdown - gmv_source: 1P vs 3P GMV split in one call - gmv_by_category, gmv_by_country: Share percentages by dimension (NOT absolute GMV - use calculate_gmv_breakdown for absolute values) - aov_components: AOV breakdown including discounts, returns, shipping, and VAT (use for discount/return questions) - info, store_type, payment_providers, shipping_providers, mobile_apps: Retailer info (retailer only). Use mobile_apps for app ratings/reviews/downloads. Supports batch queries: pass a list of retailers to compare multiple retailers. Response includes "Explore more" links to ECDB web tool retailer profiles.

get_shopper

ChatGPT
Get shopper-behavior data (loyalty, churn, CLV, buyer overlap, segments) for one or more store domains. REQUIRED: domain, metric, AND country must all be provided. Do NOT call this tool with metric=null or omitted. country is an ISO alpha-3 code and there is NO worldwide aggregate — shopper data is per-country panel data (mostly European coverage). If the user doesn't name a country, use the store's main market (the domain TLD is a good guide: zalando.de -> DEU; get_store metric=main_markets can confirm). A "not found" response usually means no shopper panel for that domain/country combination. domain="all" returns the market-wide benchmark for the country (e.g. average churn rate of German online shoppers). Valid metric values (all yearly, YYYY for start/end): - aov, lifetime_revenue: Monetary per-shopper metrics - retention_rate, churn_rate, repeat_rate, reactivation_rate: Loyalty rates - new_buyers, returning_buyers: New vs returning buyer split (shares, sum to 100%) - stable_share, upgrader_share, downgrader_share: Repeat buyers whose AOV stayed/increased/decreased - buyer_share, spent_share: Buyer/spend split by age and income segment (query with age_group=all, income_group=all for the full breakdown) - revenue_concentration: Spend share per spending decile 1-10 (4 credits) - purchase_frequency, purchase_interval, time_to_second_purchase: Purchase cadence (frequency in orders/buyer, the others in days) - aov_distribution: Order value buckets with shares (4 credits) - cohort_retention: Monthly cohort retention curve, month offset 0-11 (4 credits, ignores start/end) - cohort_retention_annual: Cohort retention triangle per year (4 credits) - overlap: Buyer overlap between domain and domain_b (REQUIRES domain_b). Use for "how many X customers also shop at Y". age_group/income_group filter to one segment; default "all" aggregates. Supports batch queries: pass a list of domains to compare stores. Use describe_shopper_metrics for detailed metric definitions. Response includes "Explore more" links to ECDB web tool store profiles.

get_store

ChatGPT
Get store data for one or more domains. REQUIRED: both domain and metric must be provided. Do NOT call this tool with metric=null or omitted — pick a specific metric first. Valid metric values (use start/end in YYYY for yearly, YYYY-MM for monthly where available): - gmv, gmv_growth: Total GMV and growth (yearly + monthly) - aov: Average order value (yearly only) - buyers, conversion_rate, purchase_frequency: Customer metrics - orders_total, orders_1p, orders_3p: Order volume metrics - revenue_net_1p, revenue_gross_3p, vat_amount_1p: Revenue breakdown - customer_age, customer_income: Customer demographics - gmv_source: 1P vs 3P GMV split in one call - gmv_by_category, gmv_by_country: Share percentages by dimension (NOT absolute GMV - use calculate_gmv_breakdown for absolute values) - aov_components: AOV breakdown including discounts, returns, shipping, and VAT (use for discount/return questions) - store_type, payment_providers, shipping_providers, shop_software: Store info (domain only) - relations, retailer, mobile_apps, main_markets: Additional info (domain only). Use mobile_apps for app ratings/reviews/downloads. - competitors: Most similar stores, ranked by a 0-1 similarity score (domain only). Use for "who competes with X" / "stores like X". Returns whatever number of peers ECDB has for that domain — it cannot be filtered by region or category, and it is not a size ranking (use get_ranking for largest stores). - top_brands: The 10 product brands with the most revenue allocated to the store (domain + currency; no time params). Use for "which brands sell most on amazon.com". Rows carry brand_name, revenue and a brand_slug for follow-up get_brand calls; the API returns no share field. Supports batch queries: pass a list of domains to compare multiple stores. Response includes "Explore more" links to ECDB web tool store profiles.

search_entities

ChatGPT
Search the ECDB catalogue by name or domain to discover entities and their slugs. Use this FIRST when you don't know the exact identifier another tool needs: - retailer slugs for get_retailer (e.g. "nike", "lounge-by-zalando") - store domains for get_store (e.g. "nike.com", "zalando.de") - company slugs for get_company — derived from the legal entity name (e.g. "nike-inc", "amazon-com-inc", "alibaba-group-holding-ltd"), so the exact form is hard to guess; always resolve them here - brand slugs for get_brand (e.g. "nike", "nvidia-geforce", "coca-cola-other-brands") - product-category slugs for get_category — the ~18,000-category Category Explorer taxonomy (e.g. "electric-toothbrushes"), searchable only here Also use it when a get_store/get_retailer/get_company/get_brand call returned "not found". Matching is prefix/substring only, NOT fuzzy: a misspelling returns zero results. If a search comes back empty, retry with a shorter prefix or the domain instead of the brand name. Optional entity_type restricts the search to "retailer", "store", "company", "brand", or "category"; omit it to see all types (useful for disambiguation). Returns up to 20 matches per entity type. Many names exist as BOTH a retailer and a brand (e.g. "nike"): the retailer is the seller and its own shops (get_retailer), the brand is the products sold anywhere online (get_brand). IMPORTANT: category results are Category Explorer slugs for get_category ONLY. They are NOT valid for the category filter of get_market/get_ranking/ calculate_gmv_breakdown — that filter uses the separate get_category_filters taxonomy. Each result includes slug, display name, entity_type, and a link to its ECDB web profile (category results also carry level and parent; brand results carry the brand's category). This tool consumes no API credits.

search_knowledge

ChatGPT
Search ECDB's FAQ and glossary: how ECDB defines, measures, sources and forecasts its data, plus eCommerce and region-group vocabulary. Use this BEFORE answering methodology questions from memory, e.g.: - definitions: GMV, 1P vs 3P, marketplace, hybrid marketplace, pure player, eCommerce activity revenue, global vs filtered revenue, AOV, conversion rate - methodology: data sources (card transaction panels etc.), accuracy, quality indicator (A/B/C), forecasting techniques, why past years' figures change between updates, currency rates and inflation handling, update frequency - region groups: what DACH, Benelux, MENA, EU27, Nordics, LATAM etc. contain query is a term ("GMV", "DACH") or a question ("how accurate is the data"). An exact term match returns that entry (plus a directly related FAQ entry if one exists); otherwise the top 5 matches. Omit query to get the title index (25 FAQ questions by group, ~110 glossary terms) and pick from it. source narrows to "faq" or "glossary" (default "all"). This is NOT for a specific tool metric — use describe_market_metrics / describe_store_metrics / describe_retailer_metrics / describe_shopper_metrics / describe_brand_metrics for those. Costs no API credits. Entries carry a link to the public FAQ or glossary page.

calculate_gmv_breakdown

Claude

describe_market_metrics

Claude

describe_retailer_metrics

Claude

describe_store_metrics

Claude

get_categories

Claude

get_company

Claude

get_credit_balance

Claude

get_market

Claude

get_ranking

Claude

get_regions

Claude

get_retailer

Claude

get_store

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/ecdb"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/ecdb)