# What Features Should a Reliable Ad Spy Tool Have? A 15-Point Buyer Checklist > The free ad libraries already give you the ads. Here's what a paid ad spy tool must add, how to read days live and copy counts, and 15 questions to ask in any demo. - **Author**: Marxx - **Published**: 2026-10-24 - **Category**: Creative strategy - **URL**: https://marxx.ai/posts/ad-spy-tool-checklist --- **What features should a reliable ad spy tool have?** Coverage of the platforms and markets you compete in, and a history of ads that have stopped. Days live and near-identical copy counts on every ad. Honest spend estimates, consistent tags for hook, offer and format, alerts when rivals change course, and client workspaces if you're an agency. Test each one live, on your own competitor. Every ad spy tool looks good in a demo. The sales rep searches a famous brand, scrolls through hundreds of polished ads, and opens a dashboard full of charts. It's impressive, and it tells you almost nothing about whether the tool will help you next Monday. The problem is that most buyers judge ad spy tools by their **library**: how many ads, how many brands. But a library is a commodity. Meta and Google already publish one for free. What you're really paying for is **signal**: the evidence that tells you which of a competitor's ads are working, what changed, and what to do about it. This guide shows you how to tell the difference. It covers: 1. What the free libraries give you, and the exact gaps a paid tool has to fill. 2. The two signals that matter most, how they work, and where they mislead. 3. Fifteen questions to ask in a demo, with what a good answer sounds like and why it matters. 4. A 30-minute demo protocol and a weighted scorecard, so you can compare tools fairly. Five of the 15 interview questions for an ad spy tool demo, with what a good answer and a red flag look like --- ## Part 1: Start with what's free, so you know what you're paying for Two free, official libraries already exist. Any paid tool has to beat them on something specific. **The Meta Ad Library** shows every ad currently running on Facebook and Instagram. But for ordinary commercial ads shown outside the EU, it keeps them **only while they're active**. When an ad stops, it drops out. Political and issue ads are kept for seven years, with spend and impressions shown in ranges. Ads delivered in the EU are kept for a year after they last ran, with targeting and estimated EU reach, because of the Digital Services Act ([PPC Land summary of Meta's Ad Library](https://ppc.land/ads-library/)). **The Google Ads Transparency Center**, launched in March 2023, lets you search ads from verified advertisers by advertiser, region, format and the date an ad was last shown ([Google, 2023](https://blog.google/technology/ads/announcing-the-launch-of-the-new-ads-transparency-center/)). What the free libraries give you and the gaps a paid tool has to fill: history of stopped ads, days live, copy counts, honest spend estimates, fixed labels and alerts | What you need | Meta Ad Library | Google Transparency Center | What a paid tool must add | |---|---|---|---| | Live ads | Yes | Yes | Nothing. This is free | | Ads that stopped (non-EU, non-political) | Removed | Shows last-shown date | A history that keeps them | | How long an ad has run | Start date | Date range | Days live, calculated for you | | Near-identical copies | Not grouped | Not grouped | Copies grouped and counted | | Spend | Ranges for political and EU ads only | Political ads only | An honest, explained estimate | | What the ad is doing | Not tagged | Not tagged | Consistent labels for hook, offer, emotion, format | | What changed | No alerts | No alerts | Alerts, trends and digests | **The key insight:** A paid tool's library size is the least important number in the demo. Its value is in the right-hand column. > **The quotable version:** You're not buying ads. Meta gives those away. You're buying the history, the grouping and the labels that turn ads into evidence. --- ## Part 2: The two signals that matter most Competitors don't publish their results. So everything an ad spy tool tells you is **inference**: reasoning from what you can see to what you can't. Two signals carry most of the weight. How to read days live and copy count: why each works and where each misleads ### Signal 1: Days live **What it is:** How long an ad has been running continuously. **Why it works:** An ad is a recurring cost. Every day it runs, the brand pays for it again. A brand that keeps choosing one creative for a month or more is revealing what it believes works. Economists call this a **revealed preference**: you learn what people value from what they repeatedly pay for, not from what they say. **Where it misleads:** - **Evergreen awareness ads** can run for months because nobody is judging them on sales. - **Forgotten ads** happen, especially in small accounts. - **Relaunches** reset the start date, so a proven idea can look new. **How to use it well:** Treat 30+ days as "worth a closer look", not "proven". It gets much stronger combined with signal 2. ### Signal 2: Copy count **What it is:** How many near-identical versions of the same creative a brand is running at once. **Why it works:** When an ad performs, brands duplicate it. They put it into new ad sets, test small changes, or push it to more audiences. Many copies usually means investment. **Real example:** In the Marxx library in October 2026, Wonderla, an Indian theme park, ran four "Buy 3, Get 1 Free" creatives with **35 to 44 copies each**, live for 99 days. SHOTT, an entertainment venue, ran "Hit a strike, grab a drink" with **6 copies for 48 days**. Those are conviction signals, not one-off posts. Real competitor ads with days live and copies, October 2026: Smaaash 'Bowling at INR 9' 36 days, Smaaash 'Make Monday your Sunday' 37 days, SHOTT 'Hit a strike, grab a drink' 48 days with 6 copies, SHOTT 'Make it epic' 37 days **Where it misleads:** Copies can also be split by region, language or placement for operational reasons. Ten copies across ten cities isn't the same as ten copies in one city. **How to use it well:** Combine the signals. **Long-running and heavily copied** is the strongest public evidence you'll get. **New but heavily copied** means a fast-scaling test, so watch it. **Long-running with one copy** might be evergreen or forgotten. | Days live | Copies | Most likely meaning | What to do | |---|---|---|---| | 30+ | Many | Proven and being scaled | Study it closely; learn its structure | | Under 14 | Many | A new test being pushed | Watch it for two weeks | | 30+ | One | Evergreen, or forgotten | Low priority | | Under 14 | One | An ordinary test | Ignore unless it lasts | ### The signal people ask for most: Spend Spend is the number buyers want and the one platforms mostly don't publish. Outside political ads and EU ranges, there's no official spend figure for Meta or Google commercial ads. Any tool showing exact spend is **estimating** it, usually from impressions, placements and assumed costs. Estimates can be useful for comparing a brand with itself over time. They're risky for budget decisions. A reliable tool explains its method and shows a range, not a single confident number. ### A worked example: The ad that vanished Timeline of Smaaash's 'Countless ways to waste your time' ad: launched 26 August, ran with 8 copies, stopped on 22 October after 56 days, then left the free library Smaaash, an Indian entertainment chain, ran a montage ad with a negative hook ("Countless ways to waste your time") from 26 August 2026. It ran **8 near-identical copies** and stopped on **22 October, after 56 days**. In the free Meta library, outside the EU, a stopped commercial ad drops out. If you only check the free library, you'd never know it existed. A tool that keeps history shows you three useful things: 1. **It worked well enough to keep for eight weeks**, with eight copies. Negative hooks can work in this category. 2. **It stopped.** That's a question, not an answer. Was it fatigue? A seasonal change? A new campaign? 3. **What replaced it?** That's often the most valuable competitive insight of the week. **What you can't conclude:** That it stopped because it failed. Without the brand's data, you only know it stopped. Good analysis keeps those apart. --- ## Part 3: The 15 questions to ask in a demo Treat the demo like a job interview. You're hiring this tool to do a job every week, so ask it what you'd ask a new analyst, using your own competitor, live. The 15 questions grouped into coverage, signal, meaning, change and fit ### Coverage: Does it see what I need? 1. **"Does it cover Meta and Google for the same competitor?"** Good answer: Both, in one view. Why it matters: Brands often test messages in search and scale them on social, or the reverse. Seeing one side hides half the story. 2. **"Does it cover the countries I sell in?"** Good answer: It shows your competitor's ads in your markets. Why it matters: A brand's ads in the US can be completely different from its ads in India. 3. **"Can I see ads that stopped, with dates?"** Good answer: Yes, with start and end dates. Why it matters: Outside the EU, the free Meta library drops stopped commercial ads. Without history, you can't see what a competitor tried and abandoned. ### Signal: Does it show what matters? 4. **"How long has this ad been running?"** Good answer: A start date and days live on every ad. 5. **"How many copies of it are live?"** Good answer: Near-identical versions grouped, with a count. Why it matters: Days live and copies together are the strongest public evidence of what's working (Part 2). 6. **"Can you show me their spend? How?"** Good answer: An honest one. "It's an estimate, here's how we calculate it, and here's the range." Red flag: An exact figure with no method. ### Meaning: Does it explain the ads? 7. **"Are hooks, offers, emotions and formats tagged?"** Good answer: Every ad labelled. A real ad broken into five tagged parts: command hook, Monday blues problem, play all games offer, excitement, branded static format 8. **"If I open the same ad twice, are the tags the same?"** Good answer: Yes, from a fixed list of labels. Why it matters: You can only count and compare what's labelled consistently. If "price hook" is sometimes "discount hook" and sometimes "offer-led", your trend charts are measuring the labeller, not the market. 9. **"Can it read the video, not just the thumbnail?"** Good answer: Transcripts or scene detection for video ads. Why it matters: A video's hook is usually in the first seconds of speech or on-screen text, not the cover frame. ### Change: Does it tell me what moved? 10. **"Will it alert me when a competitor launches, scales or stops something?"** Good answer: Alerts by email or Slack. 11. **"Can it show trends across my category?"** Good answer: Which hooks and formats are rising or declining across competitors. Marxx trending tactics showing hooks and formats rising and declining 12. **"Is there a weekly digest I'd forward to a client or boss?"** Good answer: A short summary with the ads behind each point. ### Fit: Does it work for my team? 13. **"Can I keep clients or brands separate?"** Good answer: A workspace per client, with no data mixing. 14. **"How do I share findings?"** Good answer: Boards or links that people can read without a login. 15. **"Does it work where we already work?"** Good answer: Slack, or inside Claude and ChatGPT through an MCP server. Marxx creative board grouping competitor ads by hook, angle and format ### Red flags in the demo Answers that should make you pause: we show spend, millions of ads, AI insights, export later | What they say | Why it's a red flag | What to ask instead | |---|---|---| | "We show exact spend." | Platforms don't publish it for most ads | "How is it estimated, and what's the range?" | | "We have millions of ads." | Library size is the free part | "How many from my five competitors?" | | "AI insights." | Unlabelled AI summaries can't be counted or compared | "Show me the same ad tagged twice." | | "We can export that for you later." | It may not exist yet | "Can you show me now?" | --- ## Part 4: Run a fair test ### The 30-minute demo protocol Run every tool through the same script, so you're comparing tools rather than salespeople. The 30-minute demo protocol: name your competitor, longest-running ad, what they stopped, open one ad twice, spend method, set up an alert 1. **Minutes 0-5:** Name your competitor. No famous brands. 2. **Minutes 5-10:** Ask for their longest-running ad and how many copies it has. 3. **Minutes 10-15:** Ask what they stopped in the last 30 days. 4. **Minutes 15-20:** Open one ad twice. Are the tags identical? 5. **Minutes 20-25:** Ask exactly how spend is estimated. 6. **Minutes 25-30:** Set up an alert and a share link, live. **Why this order:** The first three steps test the signals you can't get for free. If a tool fails them, the rest doesn't matter. ### Score it, weighted for how you'll use it Give each of the 15 questions **0** (no), **1** (partly, or only described) or **2** (yes, shown live). Then weight the groups by what your team actually needs. Weighted scoring: an in-house brand weights signal three times; an agency weights client workspaces three times | Group | In-house brand weight | Agency weight | |---|---|---| | Signal (days live, copies, spend honesty) | x3 | x2 | | Meaning (consistent tags, video) | x2 | x1 | | Change (alerts, trends, digests) | x2 | x2 | | Coverage (platforms, markets, history) | x1 | x2 | | Fit (workspaces, sharing, integrations) | x1 | x3 | **Why weight at all:** An unweighted score treats "has a Slack integration" as equal to "shows days live". For most teams, it isn't. Weighting forces you to decide what you're buying before the demo persuades you. **Where scoring misleads:** A demo shows the best case. Before you sign, ask for a one- or two-week trial on your own competitors, and re-score after you've used it on a normal Monday. --- ## Six buying mistakes, and how to spot them | Mistake | Symptom | Fix | |---|---|---| | Judging by library size | You chose the tool with "the most ads" | Judge by the right-hand column in Part 1 | | Demo on a famous brand | It looked great on Nike, thin on your rival | Always use your own competitor | | Trusting exact spend | Budget decisions based on a guess | Ask for the method and a range | | Treating long-running as proven | You copied an evergreen awareness ad | Combine days live with copy count | | Inconsistent tags | Trend charts jump around for no reason | Open the same ad twice in the demo | | No history | You can't see what rivals abandoned | Ask what stopped in the last 30 days | ## What to take into your buying meeting - **The library is free. The signal is what you pay for.** History, grouping and consistent labels are the product. - **Days live plus copies is the strongest public evidence.** Either alone can mislead. - **Stopped is a question, not a verdict.** You can see that an ad ended, not why. - **Exact spend is a red flag.** Honest tools show estimates and explain them. - **You can only count what's labelled consistently.** Test the tags, not the AI summary. ## FAQ **What features should I look for in a reliable ad spy tool?** Meta and Google coverage in your markets, a history of stopped ads, days live and copy counts on every ad, honest spend estimates, consistent tagging of hooks, offers and formats, alerts and trends, and client workspaces if you're an agency. **What's the most important feature?** Days live and copy counts together, plus a history of stopped ads. They're the signals the free libraries don't give you. **Do ad spy tools show competitor spend?** Not reliably. For most commercial ads, Meta and Google don't publish spend, so any figure is an estimate. Ask how it's calculated. **Is the free Meta Ad Library enough?** For checking one or two competitors' live ads, often yes. It falls short when you need history, copy counts, tagging, alerts or many competitors. **Should I trust a demo on a famous brand?** No. Ask the rep to use your own competitor, live. --- ## Run the interview on us Book a demo and ask us all 15 questions. We'll set up tracking on your top five rivals and answer them live, on their running ads. **[Book a demo: Track your top 5 rivals ->](https://www.marxx.ai)** *Related: [Best Ad Spy Tools in 2026](/blog/best-ad-spy-tools) - [Free Ad Spy Tools](/blog/free-ad-spy-tools) - [PPC Competitive Analysis Template](/blog/ppc-competitive-analysis-template)* ### Sources - Google (2023). [Announcing the launch of the new Ads Transparency Center](https://blog.google/technology/ads/announcing-the-launch-of-the-new-ads-transparency-center/). - PPC Land. [Meta Ad Library: What it shows and how long ads are kept](https://ppc.land/ads-library/). - Marxx ad library, October 2026. Days live and copy counts were accurate on 25 October 2026. --- - [More Creative strategy articles](https://marxx.ai/posts/category/creative-strategy) - [All articles](https://marxx.ai/posts)