# Scrunch agent delivery pilot Decide whether you need monitoring, page diagnostics or agent specific delivery. Complete one record for each candidate. A crawler fetch, a citation and a conversion are different observations. Worksheet version 1.0 · 17 September 2026. Complete one record per candidate. Blank fields request your evidence; this pack does not report vendor account tests or executed pilots. Complete comparison: https://rankecho.io/compare/scrunch-ai-alternatives ## 1 Define the pilot Start with a bounded public page and an explicit reason for using a delivery layer. **Job and owner** — Record required monitoring, diagnostics or agent delivery, plus the accountable technical reviewer. [Enter your response] **Candidate and entitlement** — Record exact plan, AXP or delivery inclusion, price basis, surfaces, brand and pitch workspace limits and quote link. [Enter your response] **Public page and approved facts** — Record canonical URL, source version, approved claims and the person responsible for accuracy. [Enter your response] **Acceptance and stop conditions** — Define required facts, allowed response differences, latency limit, cache behavior and rollback trigger before testing. [Enter your response] ## 2 Capture both experiences Use authorized requests to your own pilot. A copied user agent string alone does not establish a verified agent identity. **Request identity evidence** — Record request IDs, timestamps, identity verification method and provider or log evidence. [Enter your response] **Paired response captures** — Link human and agent headers and bodies for the same source version; include status, canonical, robots directives, schema and source facts. [Enter your response] | Control | Saved evidence | Pass fail or unknown | | --- | --- | --- | | Content fidelity | [Compare approved facts and required elements] | [Result and reason] | | Cache separation | [Show each visitor receives intended response on miss and hit] | [Result and reason] | | Failure and rollback | [Timeout or failure response, off switch and restored public page] | [Result and reason] | ## 3 Preserve evidence and budget Keep the existing monitoring baseline usable if the pilot stops or the vendor changes. **Observation log** — Record successful fetches, citations, referrals and conversions separately with methods and denominators; do not infer one from another. [Enter your response] **Monitoring continuity** — Freeze prompt IDs, locale, surfaces and collection window; export historical answers and mark changed methods. [Enter your response] **Delivery operations** — Name owners for CDN configuration, cache purge, source updates, alert response and tested rollback; link the runbook. [Enter your response] **Full commitment cost** — Record monitoring plus delivery charges, integration labor, ongoing operations, retained tools and written annual or enterprise terms. [Enter your response] ## 4 Accept or stop Monitoring alone does not establish delivery parity. Require a saved result for each pilot condition. **Acceptance outcomes** — Link each condition to pass, fail or unknown; keep unresolved delivery and identity questions open. [Enter your response] **Decision** — Choose retain monitoring, add delivery, switch monitoring or hold. State the evidence and residual work. [Enter your response] **Handoff** — Record approved owner, rollout scope, incident contact, off switch and rollback receipt. [Enter your response] **Review date** — Schedule a source fidelity check and fixed panel observation; state what would stop expansion. [Enter your response] ## Complete comparison guide Keep Scrunch when its monitoring workflow fits, or when a separately verified AXP edge-delivery proposal meets your requirements. Consider RankEcho for reviewable fixes and shipment-dated observations, OtterlyAI for lower-entry monitoring and audits, or Profound for visibility analysis and content Agents. Decide first whether you are replacing monitoring, improving owned pages, or buying delivery infrastructure: none of these alternatives should be assumed to replace AXP. Original guide and evidence records: 12 September 2026. Companion and pricing recheck: 17 September 2026. Confirm the current plan, billing commitment and exact workload in a written quote. Profound brand pricing rechecked 17 September 2026: its current cards offer a free Trial and custom Enterprise, replacing the previously listed Starter and Growth cards. A trial is limited to 10 ChatGPT prompts run once. Obtain a quote for ongoing tracking; the historical source records below retain their original dates. [Profound pricing](https://www.tryprofound.com/pricing) ## Which option fits the job and budget? Treat monitoring, improving owned pages, and serving agent-adapted HTML as separate purchases. The products overlap, but a monitoring plan, a content workflow, and an edge-delivery proposal are not interchangeable. Published by RankEcho from official documentation, without account-level product tests. Prices and public product claims were checked on September 12, 2026; they exclude tax, implementation, negotiated terms, and unlisted add-ons. | Option | Strong fit | Published cost basis | Poor fit or must-have gate | | --- | --- | --- | --- | | Scrunch | Monitoring with audits, recommendations, and limited content work, plus an optional enterprise edge-delivery product. | Brand Core $250 monthly; Agency Core $500 monthly; Enterprise custom. The FAQ offers Enterprise annual discounts equivalent to two months free; no Core annual amount was verified. | Brand Core: 125 prompts, one brand workspace, five licenses, and five audits monthly. Agency Core: 250 prompts, three brand plus three pitch workspaces, unlimited licenses, and five audits per brand monthly. Both cover ChatGPT, Perplexity, Google AI Overviews, and Copilot; AXP is custom Enterprise scope. | | RankEcho | A smaller Find-Fix-Prove workflow with reviewable artifacts, manual shipment, and same-prompt observations after the ship date. | Base monthly: Solo $49, Growth $149, Agency $399; annual billing gives two months free. Writing add-on monthly: Content Starter $99, Pro $249, Agency $649. | Solo, Growth, and Agency cover 1, 3, and 25 domains. Up to five AI services can be configured, with no Copilot check, edge serving, or documented native GA4 connector. | | OtterlyAI | Lower-entry daily monitoring with audits, recommendations, reports, exports, and unlimited team members. | Monthly: Lite $29 for 15 prompts, Standard $189 for 100, Premium $489 for 400. Annual billing advertises a 15% discount. | ChatGPT, Google AI Overviews, Perplexity, and Copilot are included; Claude, AI Mode, and Gemini are paid add-ons. Lite has one workspace; Standard and Premium list unlimited workspaces, API, MCP, Looker Studio, and Agent Analytics with caps. | | Profound | Visibility analysis, content-oriented Agents, and Agent Analytics for teams that need a broader data program. | Brand pricing rechecked 17 September 2026: free Trial and custom Enterprise. Current brand cards do not list the former Starter or Growth prices; request a written quote for ongoing tracking. | Trial: 10 unique prompts run once on ChatGPT, limited AI Marketer credits, no history, exports or API. Enterprise: custom daily prompt tracking, up to nine answer engines, CSV/JSON exports, API and SSO. Confirm credit allocation and terms; no AXP-equivalent serving was established. | - [Scrunch pricing](https://scrunch.com/pricing/) - [RankEcho pricing](https://rankecho.io/pricing) - [RankEcho content add-on](https://rankecho.io/pricing/content) - [OtterlyAI pricing](https://otterly.ai/pricing) - [Profound pricing](https://www.tryprofound.com/pricing) ## When should you stay with Scrunch? Stay when the current monitoring history, team workflow, and Core capacity fit, or when changing tools would add cost without resolving a failed requirement. Stay for AXP only after its custom proposal passes an infrastructure review. Scrunch describes AXP as separate from Monitoring & Insights and names Cloudflare Snippets, Akamai EdgeWorkers, and AWS CloudFront Functions as deployment options. It maps and transforms approved content at the edge for identified AI agents while humans receive the ordinary site. That documented design was not deployed or tested in this review. Switch when a measured requirement fails: surface or prompt capacity, workspace model, export rights, owned-content workflow, data access, delivery scope, or total contracted cost. Preserve a working incumbent when the alternative only changes the dashboard. - [Review Scrunch's current plan cards](https://scrunch.com/pricing/) - [Review Scrunch's AXP documentation](https://helpcenter.scrunchai.com/en/articles/13656392-agent-experience-platform-axp) ## Where do RankEcho, OtterlyAI and Profound fit? RankEcho fits when a team wants a bounded finding turned into editable fix artifacts and then observed again after manual shipment. Free audits check Perplexity and Gemini once per prompt. Paid and trialing accounts add ChatGPT, Claude, and Google AI Overviews when configured for the account, for up to 5 engines. Google AI Overviews is reported only when an overview is shown and retrievable; otherwise that cell is marked skipped/unavailable rather than counted as a citation miss. RankEcho does not currently run Microsoft Copilot checks. Its sample is fictional and deterministic, its full-page writing is a separate add-on, and its Proof Loop reports correlation with controls rather than causation. It is a poor fit for edge serving or a buyer requiring a broad native analytics-connector catalog. OtterlyAI fits a cost-sensitive monitoring and audit shortlist. It is a poor fit if the selected plan does not pass the required workspace, surface, history, or integration test. Profound fits teams evaluating visibility analysis alongside content Agents and Agent Analytics. Agents perform credit-metered work, Agent Analytics measures activity, and AXP serves adapted content; treat them as separate jobs. Test credit consumption on a representative task and require the data and delivery scope in the order form. - [Inspect RankEcho's sample fix](https://rankecho.io/sample-fix) - [Review the Proof Loop method](https://rankecho.io/product/proof-loop) - [Review OtterlyAI's plan gates](https://otterly.ai/pricing) - [Review Profound's plan gates](https://www.tryprofound.com/pricing) ## How do you compare the same workload? Use a purchase record with pass/fail requirements, not a winner score. List the brands, prompts, surfaces, locales, and collection cadence you need, then map that workload to each vendor's billable units. Include workspaces, seats, exports, content or Agent credits, implementation, and any delivery proposal in the quote. - Freeze 10 buyer prompts, one US-English market setup, the same competitors, and one collection window. Record how each product defines a prompt, answer, surface, run, unavailable result, mention, and citation. - Configure the same brand in each evaluation and record setup time, observations produced, raw answer and source access, history, and every plan gate. Keep failed and unavailable cells visible. - Choose one real owned-page gap and give each applicable option the same approved facts and source pack. Save its diagnosis, recommendation, editable output, approval step, publishing responsibility, and remaining human work. - Request the same executive report and raw export. Verify fields, timestamps, source URLs, branding, viewer access, retention, and API or integration behavior with the actual file or connection. - For credit-metered work, run a representative task and record credits consumed. For AXP, price monitoring and delivery separately. Compare twelve-month cost only after every must-have passes. ## What must you preserve before switching? Before cancelling Scrunch, export the exact prompt text, locale, surfaces, competitors, timestamps, answers, source URLs, metric definitions, workspaces, audits, recommendations, drafts, reports, roles, and contract dates you are permitted to retain. Ask the candidate to import a copy of one project, then compare identifiers, tags, dates, answer text, citations, and report fields. A successful file download does not prove historical compatibility. If sampling, availability rules, deduplication, or denominators differ, close the old series and establish a clearly dated new baseline. Run both systems for one reporting cycle when practical. Keep the old export readable, record the cutover, and do not merge percentages that describe different units. ## How do you accept or reject an agent-delivery pilot? Run this only when agent-facing delivery is a real requirement. Treat the pilot as an infrastructure and content-governance acceptance procedure; do not assume RankEcho, OtterlyAI, Profound, or a monitoring subscription supplies an AXP replacement. - Approve one small fact set first. Record its source owner, transformation rules, reviewer, version, expiry or update trigger, and the content allowed in both human and agent-facing responses. - Verify ordinary human delivery and the vendor-identified agent route with authorized infrastructure evidence. Do not rely on a copied User-Agent string as proof of identity or routing. - Check the existing URL, status, canonical, robots treatment, structured data, visible facts, links, and source ownership in both responses. Record every intentional difference for review. - Test cache variation, stale-content behavior, timeouts, origin or edge failure, fallback, rollback, and the off switch. Assign the person responsible for each control before launch. - Preserve mappings, approvals, response captures, logs, and retention terms. Re-run the checks after a template, CDN, routing, content-source, or classifier change. - Report bot fetch, answer inclusion, citation, identifiable session, and conversion as separate events. Passing a delivery check proves none of the later outcomes. - [Compare rendered and source evidence](https://rankecho.io/resources/find/issues/javascript-only-evidence) - [Separate a change from ordinary variation](https://rankecho.io/resources/prove/change-vs-noise) - [Keep citation and commercial units separate](https://rankecho.io/resources/prove/citation-to-session-to-conversion) ## What should you verify again before signing? Recheck plan names, billing, prompt and workspace units, surfaces, cadence, exports, retention, APIs, content and Agent credits, and contract terms. For Scrunch, also recheck which proposal includes AXP and the delivery, approval, cache, fallback, and off-switch responsibilities. For RankEcho, recheck the engine, Copilot, connector, content-add-on, manual-shipment, sample, and Proof Loop boundaries. Save the dated pricing page, quote, order form, data terms, export sample, pilot record, and answers to unresolved questions. Resolve a decision-critical conflict in the tested account and signed terms before cancelling the incumbent. ## What is the best Scrunch AI alternative? There is no universal best. Consider RankEcho for bounded fix and post-shipment observation work, OtterlyAI for a lower-entry monitoring and audit workflow, or Profound for visibility analysis with content Agents and Agent Analytics. Keep Scrunch when its monitoring or a verified AXP proposal passes the required workflow, technical, and commercial checks. ## Is Scrunch only an AI visibility monitoring product? No. The current Core matrix also lists site audits, recommendations, basic content generation, and limited page optimization. AXP is separately quoted in custom Enterprise scope, so confirm which scope is included in the proposal. ## Do RankEcho, OtterlyAI or Profound replace Scrunch AXP? Do not assume they do. The reviewed material did not establish an equivalent edge-delivery layer for any of the three. Evaluate monitoring, content work, and agent-facing serving as separate requirements and run a delivery pilot when serving is mandatory. ## Can I carry a Scrunch trend into another tool? Only after a field-level migration test. Preserve prompts, configuration, timestamps, answers, sources, and metric definitions. When the new tool samples different surfaces or uses different denominators, begin a dated new baseline rather than joining the series. ## Does an agent fetch prove a citation or conversion? No. A verified fetch only shows that a request reached and received a response from the tested route. Answer inclusion, citation, an identifiable session, and a commercial outcome are separate observations with separate evidence. ## Preserved dated evidence - [Scrunch pricing](https://scrunch.com/pricing/) — checked 2026-09-12. Scrunch's live US-dollar cards list Brand Core at $250 monthly and Agency Core at $500 monthly, with custom Enterprise plans. Core combines monitoring with audit and content capabilities; AXP appears in custom Enterprise scope. The rendered cards and billing FAQ were reviewed. A text-only extraction exposed conflicting Starter, Growth, and Explorer labels, so buyers should confirm the current card, quote, and order form. No account or billing flow was tested. - [Scrunch Agent Experience Platform help](https://helpcenter.scrunchai.com/en/articles/13656392-agent-experience-platform-axp) — checked 2026-09-12. Scrunch documents AXP as a separate product from Monitoring & Insights. It uses edge or CDN functions to serve adapted HTML at existing URLs to identified AI agents while ordinary site content remains for human visitors, with mapping, transformation, versioning, and approval controls. This is a documentation review, not a deployment test. No edge function, agent classification, transformed response, cache behavior, rollback, search-policy treatment, latency, citation, traffic, or outcome was verified. - [OtterlyAI pricing](https://otterly.ai/pricing) — checked 2026-09-12. OtterlyAI lists Lite at $29 monthly for 15 prompts, Standard at $189 for 100, and Premium at $489 for 400, with a 15% annual discount. All three list daily checks, unlimited team members, four included surfaces, audits, recommendations, reports, and exports; additional surfaces and advanced integrations depend on add-ons or tier. The public pricing comparison was reviewed. No account, daily run, audit, recommendation, report, workspace, export, API, MCP, Looker Studio, add-on, or checkout flow was tested. - [Profound pricing](https://www.tryprofound.com/pricing) — checked 2026-09-12. Profound publishes monthly and annual brand pricing with tier-specific prompt, surface, Agents-credit, seat, export, and API entitlements. The rendered cards and billing switch were reviewed. No account, answer-engine run, Agent task, Agent Analytics integration, export, API, credit meter, or checkout flow was tested. - [RankEcho pricing](https://rankecho.io/pricing) — checked 2026-09-12. RankEcho's monthly Solo, Growth, and Agency plans are $49, $149, and $399, with 100, 500, and 2500 tracked-prompt slots across 1, 3, and 25 domains. Paid plans include Fix and Proof workflows, with weekly Solo or daily Growth and Agency re-tests. The public page was checked against billing plan limits in source. No customer account, engine run, fix, report, export, attribution connection, or post-change result was tested. - [RankEcho content pricing](https://rankecho.io/pricing/content) — checked 2026-09-12. RankEcho full-page writing and refreshing is a separate add-on that requires a paid visibility plan. Content Starter is $99 monthly for 10 pieces, Content Pro is $249 for 40, and Content Agency is $649 for 120; one piece is 1,500 generated words, rounded up with a one-piece minimum. The add-on prices, allowances, paid-base dependency, and piece definition were checked against source. No content was generated, reviewed, exported, or published. - [RankEcho sample fix](https://rankecho.io/sample-fix) — checked 2026-09-12. RankEcho's public sample fix is a deterministic example for a fictional brand. It shows editable answer-block, schema, brief, source-plan, and review artifacts, and says the example is not customer data, a universal requirement, or a promised outcome. The sample was inspected as a documented output preview. It does not show that the same package fits another page or that publishing an artifact changes an answer or citation. - [RankEcho Proof Loop](https://rankecho.io/product/proof-loop) — checked 2026-09-12. RankEcho's Proof Loop records a baseline, shipment, and later observations for the same configured prompts and engines. Its public method describes the comparison as correlation with controls, not proof that the shipment caused a ranking, mention, or citation change. The documented workflow and limits were reviewed. No customer series, causal effect, referral, or commercial outcome was tested. ## Product next step Inspect a reviewed change at https://rankecho.io/sample-fix and the Fix Engine at https://rankecho.io/product/fix-engine. Full-page writing is separate and requires a paid base plan plus a content add-on: https://rankecho.io/pricing/content.