AdSpy Practical Guide: From Ad Search to a Creative Testing Matrix

Figure: A thematic cover generated with ImageGen, not an official AdSpy interface or ad creative. Its elements summarize the research sequence of search, analysis, and validation.

Last updated: August 24, 2026
Intended for: paid social, ecommerce growth, creative strategy, competitive research, and agency teams
Source boundary: AdSpy has limited public help documentation. This guide is based on capabilities verifiable on its public official page; interface fields, database coverage, and account limits must be confirmed in your current account.

AdSpy is a searchable database of Facebook and Instagram ads. It is best suited to answering “What language is the market using?” and “Which patterns are worth validating?” It is not suited to answering “How much did this ad spend?” or “What was its actual ROAS?” Those questions require the advertiser’s own platform data, attribution system, profit data, and incrementality tests.

This guide provides a reproducible workflow: produce one sound hypothesis in 15 minutes, narrow the search systematically, validate findings by evidence level, and finally connect ad structure, landing-page continuity, and a testing matrix in one task brief.

Table of Contents

• 15-Minute Quick Start
• What AdSpy Can Help You Answer
• Separate Officially Verifiable Capabilities from Account-Level Checks
• Evidence Levels: What Is Strong Enough to Support a Conclusion
• Interface Guide: Identify Five Control Areas First
• Complete Example: Build a Reproducible Search from Broad to Narrow
• Analyze an Ad from Hook to CTA
• Follow the Ad Through to the Landing Page
• Turn Findings into a Testing Matrix
• Set Up Competitor Monitoring
• Copyable Creative Analysis Template
• Access, Pricing, and RelayX Notes
• Common Mistakes and Troubleshooting
• 30-Day Implementation Plan
• Frequently Asked Questions
• Sources

15-Minute Quick Start

The goal of this pass is not to find a “winning ad.” It is to produce one testable creative hypothesis. Prepare a research sheet with four columns: query date, filters, evidence level, and next step.

Use this research-question template:

When promoting [category/use case] to [audience] in [country/language], which [hook/proof/media format] patterns recur during [time frame], and what kind of landing page carries the message forward?

There are only three completion criteria: the filters are reproducible, the conclusion does not depend on a single ad, and the next step can be expressed as a test that changes only one primary variable.

What AdSpy Can Help You Answer

The official AdSpy website explicitly lists search or filtering by ad text, URL, page name, likes, media type, Last seen, user reactions, and comments, along with Affiliate Network, Affiliate ID, Offer ID, and landing-page technologies. These are research inputs, not ground truth about campaign performance.

Separate Officially Verifiable Capabilities from Account-Level Checks

Capabilities Explicitly Described on the Official Website

• Search by ad text, URL, and page name.
• Search or sort using likes, media type, Last seen, user reactions, and related dimensions.
• Search ad comments by keyword.
• Use Affiliate Network, Affiliate ID, Offer ID, and landing-page technology clues.
• Access estimated audience dimensions such as location, gender, and age range.
• Search a database of Facebook and Instagram ads.

Items You Must Confirm After Signing In

The following can vary by account, date, data coverage, and product updates, so do not hard-code them based solely on the public marketing page:

• Current filter names, locations, defaults, and combination logic.
• Supported countries, languages, media subtypes, and date ranges.
• Result counts, update frequency, and completeness of ad coverage.
• Save, Export, sort, bulk-action, and team-collaboration capabilities.
• Request, concurrency, download, or export limits.
• The interpretation applied to estimated fields such as Demographics.

Before conducting research at scale, run a calibration search with a known ad or page name. Confirm whether it appears, how fields are displayed, and how Last seen is defined in the current interface.

Evidence Levels: What Is Strong Enough to Support a Conclusion

The most common problem in ad intelligence is not a lack of data; it is presenting a clue as a fact. Assign every finding an A, B, or C evidence level.

Regardless of the evidence level, AdSpy cannot prove actual spend, conversion rate, ROAS, profit, or incremental impact. Those judgments must be made with your own ad-platform, attribution, and experiment data.

Interface Guide: Identify Five Control Areas First

After signing in and opening the ad search interface, do not stack filters continuously. Review five groups in order: search object, market, time and media, engagement and comments, and advanced clues.

Figure: Teaching interface mockup with sample data, not an official live screenshot. Actual entry points, field names, ordering, and availability depend on the current AdSpy account.

Use this reading order: verify the filter scope, review the advertiser and copy, inspect observation dates and media, and only then read comments and the landing page. This reduces the bias of noticing an attractive creative first and inventing a rationale afterward.

Complete Example: Build a Reproducible Search from Broad to Narrow

Example Setup

• Research goal: Find opening-frame and copy hooks for an email automation product aimed at small ecommerce teams.
• Market: United States, English.
• Time: The 90 days preceding the query date.
• Media: Analyze video and image ads separately first.
• Required output: Three creative patterns at evidence level B or higher, plus one executable test.

Keep keywords in the market’s original language; use another language only for annotations:

• Category terms: email automation, email marketing.
• Pain-point terms: manual follow-up, low repeat purchase.
• Outcome terms: recover sales, save time.
• Use-case terms: for ecommerce, for small teams.
• Exclusion candidates: jobs, course, and unrelated brand terms.

Six-Pass Narrowing Method

Figure: Teaching interface mockup with sample data, not an official live screenshot. Counts in the diagram only illustrate narrowing logic and do not represent live AdSpy data.

For every pass, record the new condition, approximate result volume, retained samples, and exit reason. If you apply six filters at once and get no results, you cannot tell which condition made the search too narrow.

Sampling and Deduplication

Select 15–30 ads from the final two passes, then deduplicate in this order:

1. Keep one instance when the advertiser, copy, and media are identical.
2. If the copy is the same but the opening frame or media differs, label the items as creative variants.
3. If the media is the same but the Hook, Proof, or CTA differs, label the items as copy variants.
4. When multiple ads lead to the same landing page, record the ad-to-page relationship instead of counting the page repeatedly.
5. Retain the query date and original links so that a later page update is not mistaken for the evidence available at the time.

Write the Search Finding as a Testable Statement

Not acceptable: Outcome-led opening frames perform best.

Acceptable: In this sample of English-language US video ads observed within the past 90 days, four different advertisers repeatedly used an “outcome first, mechanism second” opening-frame structure. This is currently level-B pattern evidence and should be tested against a feature-led opening frame with the same audience and offer.

Analyze an Ad from Hook to CTA

Analyze each candidate with five layers. Mechanism may be added between Promise and Proof, but it must not substitute for evidence.

Do not copy entire passages of ad copy. Retain only short analytical summaries and structural labels. Images, video, trademarks, likenesses, and comments may be protected by copyright, trademark, publicity rights, or platform terms.

How Comments Fit into the Analysis

AdSpy’s official website explicitly states that comments can be searched by keyword. Classify comments into:

• Needs: What the person is explicitly trying to accomplish.
• Objections: Price, credibility, complexity, time, or fit.
• Questions: Features, shipping, refunds, compatibility, or availability by location.
• Use context: How people describe their situation in their own words.
• Risks: Complaints, misleading claims, service issues, or quality concerns.

Comments are a biased sample and may contain spam, extreme opinions, or personal information. Do not treat comment proportions as market proportions; record themes, and de-identify any quoted language.

Follow the Ad Through to the Landing Page

An ad structure becomes a complete hypothesis only when it is consistent with the landing-page experience. After opening a public URL, review the following in order:

Analyze public pages only. Do not bypass sign-in, paywalls, or security controls, submit forms, or test a real purchase unless you have explicit authorization.

Turn Findings into a Testing Matrix

Write the observation and the decision separately. A sound task brief includes the observation, evidence level, hypothesis, one primary variable, primary metric, guardrail metrics, and stopping rule.

Figure: Teaching interface mockup with sample data, not an official live screenshot. The Hook, Problem, Promise, Proof, CTA, and test results shown are instructional examples.

Example Task Brief

• Observation: Four advertisers show a before-and-after automation outcome in the opening frame of video ads.
• Evidence: Level B; the pattern recurs across advertisers.
• Hypothesis: The audience needs to understand the outcome before it will engage with a feature explanation.
• Variable: Outcome-led opening frame vs feature-led opening frame.
• Hold constant: Audience, offer, main copy, CTA, landing page, placement, and budget rules.
• Primary metric: Qualified landing-page visits, registrations/leads, or purchases.
• Guardrail metrics: Hides, complaints, bounce, refunds, and low-quality leads.
• Stopping rule: Define sample size, duration, budget, and minimum meaningful difference before launch.

Executable Testing Matrix

Where possible, change only one primary variable per test. If creative, audience, placement, bidding, and page all change together, you cannot determine what caused a better result.

Set Up Competitor Monitoring

Maintain five to eight monitored entities per category: three direct competitors, two new entrants, one or two cross-category creative references, and your own brand as a baseline.

Update these fields weekly:

Database coverage, filters, geography, and update timing can all produce gaps. Monitoring reports should say “not found in this query,” not “the advertiser is not running ads.”

Copyable Creative Analysis Template

Copy the Markdown below into your team document. Use one copy per ad. Variants of the same concept may share a research question, but their variables should be recorded separately.

### Ad sample: <internal ID>

- Query date: YYYY-MM-DD
- Research question:
- Market / language / media:
- Complete filter set:
- AdSpy record / public verification link:
- Advertiser / Page:
- First seen / Last seen (record the interface fields exactly):
- Evidence level: A / B / C

#### Creative structure

| Layer | Observation | Testable variable |
| --- | --- | --- |
| Hook |  |  |
| Problem |  |  |
| Promise |  |  |
| Proof |  |  |
| CTA |  |  |

#### Audience and landing page

- Needs found in comments:
- Objections found in comments:
- Ad-to-landing-page continuity:
- Offer and Proof:
- Compliance / rights risk:

#### Decision

- Observation:
- Hypothesis:
- One primary variable:
- Version A / B:
- Hold constant:
- Primary metric:
- Guardrail metrics:
- Sample, duration, and stopping rule:
- Owner / due date:

Access, Pricing, and RelayX Notes

As of August 24, 2026, the official AdSpy website displays promotional language for “$149/month” and “virtually unlimited usage,” while describing the price as an introductory offer. Pricing, tax, renewal terms, and search or export limits may change; refer to the current checkout page and account terms.

RelayX only provides the tool entry point and shared-access notes. See the internal marketing growth solution and AdSpy tool page. For shared use, follow four rules: save the query recipe before running it; do not enter nonpublic client strategy or personal data; do not connect personal advertising or payment accounts; and ensure exports and screenshots comply with rights, platform policy, and team permissions. RelayX request accounting is not the same as AdSpy’s use of the term “usage”; rely on the actual account counter.

Common Mistakes and Troubleshooting

30-Day Implementation Plan

Week 1: Calibration and Baseline

• Calibrate the current account fields with a known ad or page name.
• Choose one country, language, category, and audience.
• Establish five to eight monitored entities.
• Standardize the search recipe and creative analysis template.

Week 2: Systematic Search

• Complete broad search, market, time, and media grouping.
• Select 15–30 candidate ads.
• Deduplicate and cluster by Hook, Problem, Promise, Proof, and CTA.
• Assign an A/B/C evidence level to every finding.

Week 3: Comments and Landing Pages

• Code comment themes for high-priority samples.
• Review message match between each ad and landing page.
• Produce three to five creative hypotheses.
• Complete rights, brand, and compliance review.

Week 4: Small-Scale Validation

• Change only one primary variable in each test.
• Define primary metrics, guardrails, and stopping rules in advance.
• Record audience, offer, page, and delivery conditions.
• Use actual conversion, profit, and lead-quality data to retain, iterate, or stop.

Frequently Asked Questions

Can AdSpy Show an Ad’s Actual Spend and ROAS?

No. The official feature description does not provide access to an advertiser’s backend spend, attribution, profit, or incremental results. Engagement, comments, and continued observation can only help identify candidates.

Does a Recent Last seen Date Mean the Ad Is Live Right Now?

Not necessarily. It means the database observed the ad recently. Validate important conclusions with Meta Ad Library, the advertiser’s public page, or the current landing page, and record the query date.

Can I Download and Reuse the Creative Directly?

Do not assume that visibility grants reuse permission. Study the structure, theme, and communication logic, then create an original execution. Images, video, copy, trademarks, likenesses, and comments may all be protected by rights or platform terms.

Are Comments Useful for Customer Research?

They are useful for discovering language, questions, and objections, but not for estimating market proportions. Important product decisions also require interviews, support data, surveys, and observed behavior.

Does “Observed for a Long Time” Prove an Ad Is Profitable?

No. It may reflect continued delivery, but it may also be influenced by low budgets, database coverage, or observation methods. Profitability requires backend ad data, costs, refunds, repeat purchase, and incrementality data.

Sources

• AdSpy Official Website and Feature Overview
• Meta Ad Library
• Meta Advertising Standards
• RelayX: Marketing Growth Solution
• RelayX: AdSpy Tool Page

This guide is an original synthesis reorganized and written using official pages accessible on August 24, 2026. All instructional images are teaching interface mockups with sample data, not official live AdSpy screenshots. Interface fields, database coverage, pricing, and account limits are subject to the current official pages and account. Ad research does not grant reuse rights; all creative and campaign activity should undergo copyright, platform-policy, and applicable-law review.