A competitor research stack is defined as the ordered set of layers you look at when studying a rival: what they are spending on, what they are charging, what they are claiming, and how their store converts. Each layer answers a different question, and each becomes misleading when read without the others. The CROBenchmark dataset covers 7,000+ websites in 15+ industries, scored against 248+ audit criteria [CROBenchmark Report 2026, Omniconvert]. This guide shows the four layers and how to finish with decisions instead of a folder of screenshots.
- Read 4 layers in order: demand, acquisition, offer and conversion.
- Use each layer to explain the one above it, because each misleads when read on its own.
- Read persistence, not volume, in an ad library, since a long-running ad has survived a decision to keep paying for it.
- Trust the offer layer most, because price, thresholds and guarantees are published.
- End with a decision list with owners, not a document or a folder of screenshots.
Most DTC competitor research produces a folder of screenshots and no decisions. Somebody spends a day in an ad library, saves forty creatives, notes that a rival is running a spring promotion, and presents it. Everyone agrees it is interesting. Nothing changes. The research was not wrong, it simply stopped one layer short of being usable. Last updated: October 2026.
The fix is structural rather than a matter of effort. Omniconvert has scored stores against real competitors in their category and country for 13 years in eCommerce, drawing on a CROBenchmark dataset covering 70,000+ experiments and 2,500+ Shopify stores, and research that changes a roadmap reads its observations in layers, not as a pile. This guide sets out the four layers, what each can honestly tell you, where each one lies, and how to finish with decisions.
What a competitor research stack is
Layer one: demand, or who actually competes with you
Three cheap sources fix this quickly. Post-purchase survey responses to a single question about what else was considered will name brands you did not expect, and often exclude ones you assumed. Search results for your core buying queries in your target country show who is actively competing for that intent. Answer engines, increasingly, name a shortlist directly when asked a buying question in your category, and that shortlist is closer to the modern consideration set than any category report.
Layer two: acquisition, and how to read an ad library honestly
The Meta Ad Library is public, free and the single most useful competitor artefact available to a DTC brand. It is also routinely misread, in a specific and predictable way.
Persistence is defined as how long a creative has stayed continuously live. The misreading is treating quantity as evidence. A rival with sixty live creatives is not sixty times more successful than one with three; they may simply be in a heavy testing phase, or running dynamic variants that expand a single concept. What carries signal is persistence. Most creative gets paused quickly, so an ad that has been continuously live for months has repeatedly survived somebody's decision to keep funding it.
What this layer genuinely cannot tell you is spend. Estimates circulate widely and they are inferences from proxies. Use them for rough ordering if you must, never as a number in a plan, and be openly sceptical when a tool presents one to two decimal places.
Read the source documentation once so you know the limits of the layer. Meta’s own Ad Library research tools documentation sets out what the library carries, and Google’s Ads Transparency Center does the same for Search and YouTube.
Layer three: the offer, which usually explains the creative
Record five things per competitor: the entry price and the price of the most-promoted item, the bundle or subscription structure, the free-shipping threshold, the returns and guarantee terms, and the promotional cadence over a quarter. That last one is the sleeper. A brand that discounts on a predictable monthly rhythm has trained its customers to wait, which is both a weakness you can attack and a trap you should avoid copying.
Baymard Institute's checkout research is the standard reference for why this layer converts into conversion outcomes: unexpected extra cost shown late is a leading driver of abandonment, and average cart abandonment sits near seventy percent across the industry. A rival whose shipping terms are stated early and plainly is not being generous, they are removing the most common reason a full basket is left behind.
Layer four: conversion, walked rather than assumed
Note specifically: how quickly the core claim is made, what reassurance appears above the fold, whether sizing or fit is resolved on the page or deferred, when total cost including shipping first becomes visible, how many fields stand between the basket and payment, and whether guest checkout exists. Those six items account for most observable conversion difference between DTC stores in the same category.
What each layer can and cannot tell you
| Layer | Reliably observable | Commonly misread as | Refresh cadence |
|---|---|---|---|
| Demand | Who the buyer considered | Who shares your category label | Twice a year |
| Acquisition | Which creatives persist | Spend level, from creative volume | Monthly |
| Offer | Price, bundles, thresholds, terms | Margin and unit economics | Fortnightly |
| Conversion | The path a customer walks | Their conversion rate | Quarterly |
| AI visibility | Whether an engine names them | Their organic search rank | Monthly |
| Agentic readiness | Whether their product data resolves | Site quality generally | Quarterly |
The last two rows are where most competitor research has a blind spot. Whether an answer engine names a rival when a shopper asks a buying question in your category is now a competitive fact, observable in minutes by asking. AI Visibility and Agentic Commerce sit alongside Creative and Ads, Reviews and UGC, Competitor Synthesis and CRO in the six-dimension store score.
Turning four layers into decisions
- A claim to test. A persistent rival angle you have never run. It becomes a creative test, not a rebrand. The fact that it survived for them is a hypothesis about your audience, not a conclusion.
- A price or threshold to trial. A shipping threshold, bundle or guarantee that three rivals share and you lack. Convergence across independent competitors is the strongest signal available in this whole exercise.
- An element to add. Something from the walked purchase path: the sizing resolution, the early total, the guest option. Usually small, usually cheap, usually overdue.
- A gap to attack. Something none of them do, that your buyers ask about. The rarest output and the most valuable, because it is the only one that does not make you more similar to your competitors.
Where the ranked list needs experiments rather than opinions to settle it, the same logic extends across the wider growth stack: Omniconvert Commander (formerly Nexus) unifies commerce data, prioritises experiments by True Profit and generates the campaigns and creative a human approves before they go live. And if you want the six-dimension version of this comparison done for you, a free leaderboard score benchmarks your store against real competitors in your category and country, with the paid report covering all six dimensions in depth.
The highest-yield tool in that stack gets its own walkthrough: how to Reverse-Engineer competitor ads with the Meta ad library.
Related reading: spot a winning ad.
A note on what the stack is for, since the tools are easier to assemble than the intent behind them. The discipline this all serves has a definition worth starting from: what is competitive intelligence for eCommerce?
FAQ: the competitor research stack
What is a competitor research stack?
A competitor research stack is defined as the ordered set of layers you look at when studying a rival: what they are spending on, what they are charging, what they are claiming, and how their store converts. Each layer answers a different question, and each becomes misleading when read without the others. The stack exists to stop competitor research being a screenshot collection.
How do you find a competitor’s ads?
The Meta Ad Library is public and shows active ads for any advertiser, which makes it the cheapest starting point. Search the brand, filter to your country, and record which creatives have been running longest. Longevity is the useful signal because most creative is paused quickly, so an ad still live after months has almost certainly earned its place.
Can you tell how much a competitor is spending?
Not accurately, and tools that claim a precise number are estimating from proxies. What you can read reliably is direction and emphasis: how many creatives are live, how long they persist, which products they feature, which markets they run in. Treat spend estimates as rough ordering, never as figures worth quoting in a plan.
How often should you run competitor research?
Run a full pass quarterly and a light pass monthly. The full pass covers all four layers and produces decisions. The light pass checks only what changes fast: new creative angles, price moves, promotional cadence. Weekly monitoring feels diligent and mostly captures noise, because the things that matter move slower than a week.
What should competitor research actually produce?
A short list of decisions with an owner each: a claim to test in your own creative, a price or bundle to trial, a page element to add, a gap to attack. If the output is a document describing rivals rather than a list of things you will now do differently, the research has stopped one step short of being useful.
Most teams that run DTC competitor research jump to the ad library, save forty creatives and present a folder of screenshots. They skip the first layer, so the competitor list holds the brands they admire and not the ones their buyer considered, and every later judgement is made against the wrong set. The research then ends as a document with no decision attached.
A persistent rival angle becomes a hypothesis, and a structured testing programme is how you check it. AliveCor used Omniconvert to run a structured A/B testing programme and achieved +21% conversion rate, +5% revenue per visitor, and 94% statistical relevance across their experiments. [Omniconvert, AliveCor case study]
Keep three rivals, four layers and one slot for the gap
Competitor research earns its keep when it is layered and narrow: three real competitors, four layers, read in order. Establish who genuinely competes for your buyer, read persistence rather than volume in the ad libraries, write down the published offer every fortnight because it explains the creative, and walk the purchase path yourself on a phone. Then force every observation into a decision with an owner, and keep one slot for the gap nobody is filling. Do that quarterly and competitor research stops being a folder of screenshots and starts being the cheapest source of roadmap items you have. To see where you currently stand across all six dimensions, including the AI visibility and agentic readiness most competitor programmes never check, start with a free store benchmark score.