RMIQ (Retail Media IQ) is an AI-driven retail media buying platform that runs SKU-level keyword, bid and budget optimization across Walmart Connect, Instacart, Target, Kroger and other retail media networks from one dashboard, on a pay-for-performance model.
Used this tool? Rate it
Used this tool? Rate it
RMIQ, which presents itself publicly as Retail Media IQ, is an advertising execution platform for retail media networks. Its homepage states its purpose in a single line: "AI-DRIVEN RETAIL MEDIA PERFORMANCE ADVERTISING," followed by the qualifier "FOR LEAN TEAMS — Get discovered. Sell more. Improve ROAS. On Instacart, Whole Foods and more." That framing is worth reading carefully, because it defines both the product and its intended buyer. RMIQ is not a general marketing suite, not an e-commerce storefront tool, and not a social advertising platform. It exists to buy and optimize sponsored placements inside retailer-owned ad networks — the ad inventory that appears when a shopper searches for a product on Walmart, Instacart, Target, or Kroger.
A note on naming is necessary before going further. This tool is catalogued under the name "Retail," which is a generic word that tells a reader almost nothing. The registered domain is rmiq.net, and the operator's own branding across the site, its press materials, and third-party industry coverage is consistently RMIQ or Retail Media IQ. The catalogue entry's short description — an AI-powered platform to create, launch, and optimize ad campaigns across retail media networks — is materially accurate and matches what the site actually claims. The naming, however, is not: readers searching for "Retail" will not find this product, and the vendor never uses that bare word as its brand. Throughout this document the product is referred to as RMIQ, which is what the operator calls itself.
Retail media is a specific and relatively young advertising category. Over the past several years, large retailers realized that their own websites and apps constitute valuable advertising inventory, because a shopper browsing a grocery app is far closer to a purchase decision than someone scrolling a social feed. Walmart, Target, Kroger, Instacart, and dozens of others each built their own advertising businesses with their own consoles, their own bidding mechanics, their own reporting definitions, and their own onboarding processes. For a brand selling across five or ten of these retailers, this created a genuine operational problem: five or ten separate logins, five or ten incompatible reports, and no coherent way to compare performance or shift budget between them. RMIQ positions itself squarely at that problem.
For a small-domain product, the question of who is actually behind it deserves a direct answer, and in RMIQ's case the answer is unusually well documented. The company publishes a full team page naming its leadership with verifiable professional histories. Mateusz Drela is listed as CEO and described as "an accomplished ad tech entrepreneur and the former CEO of Zeropark" with "over 15 years of experience in performance marketing." Jacek Chodźko is listed as COO with a fifteen-year consulting background. Jan Żywczak is listed as CTO with prior senior engineering roles at Codewise and Ocado — the latter being a significant name in grocery technology, which is relevant given RMIQ's grocery-heavy network focus. The page also names a CSO, additional senior engineers, and an advisory board.
This is a meaningful signal. A large share of small AI-branded domains are anonymous: no named team, no verifiable histories, no accountable entity. RMIQ is the opposite. Its leadership is named, their prior employers are checkable, and their stated backgrounds are coherent with the product being built — ad tech and performance marketing people building an ad tech performance product. Independent industry coverage corroborates this: the Path to Purchase Institute, a retail industry publication, ran a question-and-answer interview with Drela in which the company is described as "a unified, AI-driven retail media buying platform that simplifies campaign execution across multiple retailer networks." That description, arrived at independently, matches the company's own.
There is, however, a genuine inconsistency in how the operating entity is identified, and it should be stated plainly rather than smoothed over. The privacy policy names the controller as "RMIQ LLC," an entity that "owns and operates the RMIQ business, including the RMIQ platform." A press release distributed through financial news syndication gives a United States address: RMIQ LLC, 16192 Coastal Highway, Lewes, Delaware 19958. But a separate legal page hosted on the same site identifies "RMIQ SPÓŁKA Z OGRANICZONĄ ODPOWIEDZIALNOŚCIĄ" — a Polish limited liability company — with a Polish tax identification number (NIP 5253026819) and a Warsaw address. Three references, two jurisdictions, one brand.
None of this is inherently suspicious. A Delaware holding entity paired with a Polish operating company is an entirely ordinary structure for a European-founded startup selling into the United States market, and the leadership team's Polish names are consistent with that reading. What is a legitimate concern is that the site never explains the relationship. A prospective customer signing a commercial agreement cannot tell from the public site which entity they would be contracting with, which jurisdiction governs, or how the two relate. That is a reasonable question to put to the sales team before signing, and this document flags it rather than resolving it, because the public record does not resolve it.
RMIQ's functional claims cluster around a small number of ideas, each of which is stated explicitly on the site.
SKU-level optimization. The central mechanic, in the company's own words, is to "Win incremental sales by AI-optimizing each SKU with dedicated keywords, bids, and budgets." The emphasis on the individual SKU is the substantive claim here. Many advertisers manage retail media at the campaign or ad-group level, where a single budget covers a basket of products. Managing at the SKU level means each individual product line gets its own keyword set, its own bids, and its own budget allocation — considerably more granular, and considerably more laborious to do by hand, which is the argument for automating it.
Cross-network learning. RMIQ claims advertisers can "Scale faster by leveraging shared cross-network learnings instead of starting from zero." The premise is that performance signal accumulated on one retailer's network — which keywords convert for a given product, which bid levels clear — carries predictive value when launching that same product on a different retailer. Whether this transfer works well in practice depends heavily on how similar the shopper populations and auction dynamics are between networks, and the company publishes no methodology or measured results for the transfer mechanism specifically. Treat it as a stated design goal rather than a verified outcome.
Unified reporting. The platform advertises "Unified Reporting Across Networks," which addresses the fragmentation problem described earlier. Each retail media network reports on its own terms, with its own attribution windows and its own metric definitions; normalizing these into one comparable view is genuinely useful and genuinely difficult.
Configurable objectives. The site lists "Custom Goals: Sales, ROAS, or Customer Growth," indicating the optimization target can be set per advertiser rather than assuming everyone optimizes for the same metric. Sales volume, return on ad spend, and new customer acquisition frequently conflict with one another, so being able to choose is a substantive rather than cosmetic feature.
Managed execution. Distinct from self-service tooling, RMIQ offers "end-to-end execution by RMIQ and full visibility into results." This positions the product as a hybrid: software plus a service layer that operates campaigns on the advertiser's behalf. This matters for evaluation, because a lean team buying RMIQ may be buying an outsourced operations function as much as a dashboard.
Free keyword research tools. The site hosts five publicly accessible keyword tools covering Amazon, Walmart, Instacart, Whole Foods, and eBay. The Amazon tool, for example, "helps advertisers optimize their advertising keyword strategy by recommending keyword match types based on desired brand, product and descriptive search terms," sorting terms into broad, phrase, and exact match while flagging negative keywords. These function as marketing entry points rather than as the core product, and no usage quota or rate limit is published for any of them.
RMIQ dedicates a page to describing its AI approach, and it is worth reporting precisely what that page does and does not say. It defines its subject in general terms — "AI agents are autonomous systems designed to perform specific tasks, make decisions, and optimize outcomes based on data" — and then describes a multi-agent architecture in which "Each agent has a distinct role, from fine-tuning bids and reallocating budgets to analyzing cross-platform data and orchestrating A/B testing."
Five agent types are named: a Network Optimizer Agent, a Cross-Network Learning Agent, a Budget Allocation Agent, a Campaign Strategy Agent, and an A/B Testing and Orchestration Agent. The company states that its system "continuously analyzes performance, adapts bidding strategies, and reallocates budgets" and that it "leverages deep learning and cross-platform intelligence." The page also references the "Web of Agents" paradigm as conceptual inspiration.
Here is the important gap, stated directly: RMIQ does not disclose what models it uses. Across the AI technology page and the rest of the public site, no third-party model provider is named — no OpenAI, no Anthropic, no Google, no open-weight model family, and no claim of a proprietary model of its own either. The word "agent" as RMIQ uses it describes functional roles in an optimization system, and nothing in the published material establishes whether those agents are built on large language models, on classical machine learning and reinforcement learning for bid optimization, or on some combination. For an advertising bid-management system, conventional statistical and reinforcement-learning approaches would be the technically natural choice, and much of what RMIQ describes — bid tuning, budget reallocation, A/B orchestration — is the kind of work that predates the current generation of language models entirely.
This document does not speculate further, because speculation is not evidence. What can be said with confidence is that buyers who need to know the model supply chain — for data residency, vendor risk, or procurement review — will not find that information published and will have to ask. That is a documentation gap rather than a red flag, but it is a gap.
RMIQ's networks page enumerates specific integrations. The named list covers Walmart Connect, Instacart, Target, Sprouts Farmers Market, Home Depot, CVS, Staples, Macy's, Kroger, Thrive Market, Chewy, Whole Foods, DoorDash, Uber Eats, Albertsons, and GoPuff — sixteen named networks, weighted heavily toward grocery, convenience, and delivery, with a secondary cluster in home improvement, pharmacy, office supply, and pet retail.
That list conflicts with another page on the same site. The sellers page headlines coverage of "70+ retail media networks." Sixteen specific names against a claim of more than seventy is a substantial gap, and the site offers no reconciliation. The most charitable reading is that the sixteen represent fully built integrations while the seventy-plus counts networks reachable by some other means, but the site does not say this, and a reader has no basis to assume it. Where the two conflict, the specifically named list is the verifiable one. Any advertiser whose plan depends on a particular retailer should confirm that specific integration directly before committing, rather than relying on the aggregate count.
A geographic observation follows from the list: every named network is a United States retailer. Nothing on the site claims European, Latin American, or Asia-Pacific retail media coverage. Advertisers outside the US market should verify regional availability before evaluating further.
Pricing is where RMIQ's public documentation is at its most inconsistent, and this section reports what each page says rather than reconciling them into a tidier story than the evidence supports.
There is no dedicated pricing page. The path /pricing/ returns a 404 error, despite the site's own navigation referencing pricing in places. What pricing information exists is scattered across audience-specific pages, and it does not agree.
The agencies page carries the only concrete rate. It states a "10% revenue share on results," a "First month free – no fees," an invitation to "Try RMIQ free for 30 days. No risk, just results," and "Discounts for agencies with 10+ clients." A ten percent revenue share aligned to results is a real, specific commercial term.
The brands page gives no numbers at all. It states "No Commitment, No Contract Needed" and describes "flexible payment options tailored to support growing brands," offering "pay-as-you-go options or structured payment plans" without quantifying any of them. It also advertises an equity-oriented program: "Apply for our $1000 campaign, absolutely free!" directed at female and Black-owned startups and underrepresented founders.
The sellers page uses a third framing. It promises "Pay Only for Results" and a "Start Your Free Trial Today" call to action, with neither the revenue share percentage nor the trial length specified.
The keyword audit is priced two different ways on two different pages. The homepage advertises a "$490 Keyword Audit," described as "Credited in full if you continue with RMIQ." The dedicated audit page prices the same service at "$500," delivered in three business days, "Fully credited if you continue." Both figures were confirmed directly in the pages' own text. The gap is small in absolute terms but it is a documented discrepancy in a published price, and the correct advice is to get the figure confirmed in writing before paying.
The practical summary: RMIQ's commercial model is performance-based rather than seat-based or subscription-based, the only published rate is the agency-side ten percent of results, and everything else requires a sales conversation. Advertisers should treat the numbers above as starting points for negotiation and confirm all terms — rate, trial length, minimum spend, contract duration, notice period — directly with the vendor.
RMIQ's own targeting language is "FOR LEAN TEAMS," and taking that seriously clarifies fit considerably.
Good fit: brands selling across several US retailers with no dedicated retail media staff. This is the core case. A consumer packaged goods brand on Walmart, Instacart, Kroger, and Sprouts, with one marketer splitting attention across everything, is exactly the situation where consolidating four consoles into one and outsourcing daily bid management has clear value.
Good fit: agencies managing retail media for multiple clients. The published ten percent revenue-share model and volume discounts at ten or more clients suggest the agency channel is a deliberate focus rather than an afterthought.
Good fit: grocery, convenience, and delivery brands specifically. The integration list is unusually strong in this vertical — Instacart, Whole Foods, Kroger, Sprouts, Albertsons, Thrive Market, GoPuff, DoorDash, Uber Eats. A brand concentrated here will find better coverage than in most other categories.
Poor fit: advertisers outside the United States. Every named network is US-based, and no international coverage is claimed.
Poor fit: teams needing a documented, auditable technology stack. If procurement requires knowing which models process your data, that information is not published.
Poor fit: organizations that require a formal service agreement before evaluation. As detailed below, no standard terms of service exists publicly.
Poor fit: buyers who rely on peer reviews to de-risk a purchase. Independent user reviews are effectively absent.
Based on what the site documents, the realistic path in is as follows.
Start with the free keyword tools. The Amazon, Walmart, Instacart, Whole Foods, and eBay keyword tools are publicly available and require no commercial relationship. They are the lowest-cost way to form a first impression of whether the vendor's keyword thinking is any good, and to get output you can sanity-check against campaigns you already run.
Consider the paid keyword audit as a scoped pilot. At $490 or $500 — confirm which — with three-business-day delivery and full credit against future work, the audit is a bounded way to test the vendor's analytical quality before any platform commitment. The company's marketing claims that "Most brands find ~30% of keywords don't generate any revenue when you take granular, product-level look," which is a testable proposition: run the audit against an account whose performance you already understand and check whether the findings hold up.
Verify your specific networks in the sales conversation. Given the sixteen-versus-seventy discrepancy, name your retailers explicitly and get confirmation of integration depth for each one that matters to you.
Clarify the commercial terms in writing. Which entity contracts with you, what the revenue share is for your segment, what "results" means precisely in the revenue-share calculation, how the free period is defined, and what the exit process looks like.
Run it in parallel before switching. Retail media performance is noisy and seasonal. Keep a control — some products or some networks managed as before — so that any change is measured against something rather than asserted.
Agree on reporting definitions up front. Because each network defines attribution differently, establish which definitions the unified reporting uses before you start comparing its numbers to the ones you get from the retailers directly.
This section is deliberately direct, because the gaps are real and a reader deciding whether to spend money deserves to see them collected in one place.
There is no standard terms of service. This is the most significant documentation gap. The site's footer link labelled "Terms of Service" points to a page that is not a terms of service at all. It resolves to "RMIQ.NET TOMBOLA AT NATURAL PRODUCTS EXPO WEST — OFFICIAL RULES & TERMS," a set of contest rules for a trade-show prize drawing at Natural Products Expo West. That document governs sweepstakes entry, prize eligibility, and credit expiry — the prizes "Have no cash value," are "non-transferable," and "Expire 90 days from the date of issuance." It says nothing about platform use, service levels, payment obligations, intellectual property, termination, or liability. The conventional paths /terms/, /terms-of-service/, and /terms-and-conditions/ all return 404. For a product that takes a percentage of advertising revenue, the absence of published commercial terms is a substantive concern, and any customer should insist on receiving a real agreement before transacting.
The operating entity is presented inconsistently. As covered above, the privacy policy says RMIQ LLC, a press release places that LLC in Delaware, and the contest rules name a Polish company with a Warsaw address and a Polish tax number. The relationship between them is never explained publicly.
Network coverage claims do not agree. Sixteen named integrations against a "70+" headline, unreconciled.
The audit price is published at two figures. $490 on the homepage, $500 on the audit page.
The model stack is undisclosed. No third-party provider is named and no proprietary model is claimed.
Data retention is vague and AI training is unaddressed. The privacy policy commits only to retaining "an archive of such Personal Information for a commercially reasonable time," which is not a defined period. More consequentially for an advertising platform, the policy does not state whether campaign performance data, keyword data, or account data is used to train or improve the vendor's models. Given that "cross-network learning" is an advertised feature, and that such learning is by definition aggregation of signal across accounts, the mechanics of that aggregation are a fair question — particularly for brands whose competitors may also be RMIQ customers. Ask directly whether your performance data informs optimization for other advertisers, and get the answer in writing.
Independent user reviews are essentially nonexistent. A G2 product page exists but carries no user reviews. No meaningful presence was found on Capterra, Trustpilot, or Product Hunt, and no substantive practitioner discussion was found in community forums. Attempts to retrieve the G2 page and the Crunchbase profile were blocked by anti-bot protection on both direct fetch and a full browser fallback, so neither could be read in full. The practical consequence is that the performance figures RMIQ publishes — "+160% ROAS growth" on Instacart, "+80% ROAS growth" on Walmart, "3X ROAS" in a customer testimonial, and figures elsewhere including "35% higher average ROAS," "70% reduction in ad spend waste," and "100x faster insights" — are entirely vendor-reported. No baseline, time period, sample size, or methodology is published for any of them, and no third party has verified them. Treat every one of these numbers as marketing rather than measurement.
Coverage is US-only. No international retail media networks are named.
Advertisers evaluating RMIQ are generally choosing among four approaches, and the honest comparison is about trade-offs rather than a winner.
Managing each retailer's native console directly. Walmart Connect, Instacart Ads, and the rest all provide their own interfaces at no additional platform cost. This gives the deepest access to each network's features and the lowest fees, at the cost of proportionally growing manual effort and no comparable cross-network view. For a brand on one or two retailers, this is often the right answer and RMIQ's value proposition largely evaporates.
Enterprise retail media platforms. Established commerce-advertising vendors serve large brands and holding companies with broad network coverage, mature reporting, and formal enterprise contracting. They typically carry higher minimums and longer implementations. RMIQ's explicit "lean teams" positioning suggests it is not primarily competing here, and a large enterprise with strict procurement requirements would likely find RMIQ's documentation gaps disqualifying.
A specialist agency. An agency provides human expertise and accountability without platform lock-in, usually at a percentage of spend. This is the closest analogue to RMIQ's managed-execution offering, and the comparison is fairly direct: RMIQ argues that automation plus a service layer delivers agency-like outcomes at a performance-linked rate.
Doing less. For brands with small retail media budgets, the arithmetic matters. A ten percent revenue share is only worth paying if the optimization produces more than ten percent incremental lift over what the team would achieve alone. Below a certain spend level, the reasonable choice is to run simpler campaigns manually and revisit tooling when volume justifies it.
RMIQ's distinguishing characteristics within this field are its grocery and delivery network depth, its performance-linked pricing, and its combination of software with hands-on execution. Its weaknesses relative to more established competitors are documentation completeness and the absence of any independent verification.
Do not skip the free tools. They cost nothing and give a real signal about the vendor's keyword methodology.
Scope the paid audit against a known account. Testing an analytical service on data you already understand is the only way to evaluate whether its findings are insightful or generic.
Get a real contract. Given that no standard terms are published, receiving and reviewing an actual service agreement is not optional diligence — it is the diligence.
Pin down the data question. Specifically: is your campaign data used to improve optimization for other advertisers, and is it used for model training? Written answers only.
Verify the entity you are contracting with. Delaware LLC or Polish spółka z o.o. — establish which, and what governing law applies.
Confirm your networks individually. The aggregate coverage claim is not reliable; the named list is.
Measure against a control. Vendor-reported ROAS improvements are unverified. Your own parallel test is the only figure that will mean anything.
RMIQ is a retail media advertising platform. It creates, runs, and optimizes sponsored product campaigns inside retailer-owned ad networks — the placements shoppers see when searching on Walmart, Instacart, Target, Kroger, and similar platforms. Its stated core mechanic is to optimize keywords, bids, and budgets at the individual SKU level, with unified reporting across all connected networks and an option for RMIQ to execute campaigns on your behalf rather than leaving them for you to manage.
That is a catalogue naming artifact, not a discrepancy in the product. The registered domain is rmiq.net and the operator brands itself consistently as RMIQ or Retail Media IQ across its own site, its press materials, and independent industry coverage. The generic word "Retail" is not a name the vendor uses. The catalogue's underlying description of the product — an AI-powered platform for creating, launching, and optimizing campaigns across retail media networks — is accurate; only the name is misleading.
The networks page specifically names Walmart Connect, Instacart, Target, Sprouts Farmers Market, Home Depot, CVS, Staples, Macy's, Kroger, Thrive Market, Chewy, Whole Foods, DoorDash, Uber Eats, Albertsons, and GoPuff. A separate page on the same site claims coverage of more than seventy networks, which the site never reconciles with the named sixteen. Rely on the named list, and confirm any retailer critical to your plan directly with the vendor before committing.
There is no published pricing page — the /pricing/ path returns a 404. The only concrete rate published anywhere on the site appears on the agencies page: a 10% revenue share on results, a first month free, and discounts for agencies with ten or more clients, alongside a thirty-day free trial offer. The brands page mentions flexible pay-as-you-go arrangements with no numbers, and the sellers page promises "Pay Only for Results" without specifying the rate. Expect to negotiate, and get terms in writing.
The site publishes two different prices for the same service: the homepage says $490 and the dedicated audit page says $500. Both include full credit toward future work if you continue with RMIQ, with delivery quoted at three business days. Confirm the actual figure before paying. As a bounded, low-commitment way to evaluate the vendor's analytical quality before signing anything larger, it is a reasonable structure — provided you run it against an account whose performance you already understand well enough to judge the findings.
This is not disclosed. RMIQ's technology page describes a multi-agent architecture with five named agent roles — Network Optimizer, Cross-Network Learning, Budget Allocation, Campaign Strategy, and A/B Testing and Orchestration — and refers to deep learning and cross-platform intelligence. But no third-party model provider is named anywhere on the site, and no proprietary model is claimed either. If knowing the model supply chain matters for your procurement or data-governance review, you will need to ask the vendor directly.
The published privacy policy does not address this. It covers data collection, GDPR and CCPA rights, cookies, and third-party service providers, but says nothing about whether advertising performance data is used for model training. This deserves specific attention because "cross-network learning" is an advertised feature, and learning transferred across accounts implies aggregation of signal from multiple advertisers. If competitors in your category may also be customers, ask explicitly whether your data informs optimization for other accounts, and require a written answer.
Not publicly. The footer link labelled "Terms of Service" leads to official rules for a trade-show prize drawing at Natural Products Expo West, not to platform terms — it governs sweepstakes eligibility and prize credits that expire ninety days from issuance. The conventional paths for terms all return 404 errors. Since the commercial model involves taking a percentage of your advertising revenue, obtaining and reviewing an actual service agreement before transacting is essential rather than optional.
They are vendor-reported and unverified. The figures on the site include +160% ROAS growth on Instacart, +80% on Walmart, a 3X ROAS customer testimonial, and elsewhere 35% higher average ROAS and 70% reduction in ad spend waste. None is accompanied by a baseline, a time period, a sample size, or a methodology, and no independent party has validated them. A G2 profile exists but carries no user reviews, and no substantive coverage was found on Capterra, Trustpilot, Product Hunt, or in practitioner communities. Plan to measure lift yourself against a control group rather than relying on these numbers.
The leadership is publicly named and checkable, which is a positive signal: CEO Mateusz Drela, previously CEO of Zeropark with over fifteen years in performance marketing; COO Jacek Chodźko; and CTO Jan Żywczak, with prior senior engineering roles at Codewise and Ocado, plus additional named engineers and an advisory board. The legal entity is less clear. The privacy policy names RMIQ LLC, a press release places that entity at a Delaware address, and a separate legal page on the same site names a Polish company, RMIQ Spółka z ograniczoną odpowiedzialnością, with a Warsaw address and Polish tax number. A Delaware-plus-Poland structure is ordinary for a European startup selling into the US, but the site never explains it — so establish which entity you are contracting with before you sign.
