Akamai Access Denied reference number: how to trace the block
What the reference number on an Akamai Access Denied page identifies, who can translate it, which Akamai tools do it, and how long the trail stays available.
By Crawlito · 9 min read
In short
- The reference number on an Akamai Access Denied page identifies one denied request, so it is the single most useful thing to save from a failed crawler test.
- Akamai documents its translation tools only inside the site's own Akamai account: the Translate Error String tool in Edge Diagnostics and the Security Event Error Translator in Security Center.
- Edge Diagnostics keeps error logs for the last 6 or 24 hours depending on server and traffic conditions, so translate the string the same day.
- The Application Security API looks up codes from the past 31 days and returns failure categories such as WAF, BOT, CUSTOM_RULE, REPUTATION and NETWORK_LIST.
- An Edge Diagnostics translation shows the client IP and user agent of the denied request, which tells you whether it was your own test or the real crawler.
On this page
The reference number on an Akamai "Access Denied" page is an identifier for one specific denied request. On its own it tells a visitor nothing, but the site's Akamai team can paste it into the Translate Error String tool in Edge Diagnostics, or look it up through the Application Security error translator, and see which property, URL, client and security control produced the block. If you hit one during a crawler test, save the full string and the time, and send both to whoever runs the site's Akamai account.
This post covers the trace itself: what to capture, which Akamai tool to use, what each one returns and how long the data lasts. Why AI crawlers get denied by Bot Manager in the first place, and how to change the actions, is in the Akamai Bot Manager guide.
What the reference number is
A typical Akamai denial page is short. Microsoft's support article on Akamai-hosted sites shows this example:
Access Denied
You don't have permission to access http://www.microsoft.com on this server.
Reference #18.85c5d617.1549635923.277a7a1Microsoft describes the reference number as a unique identifier for that user and that access attempt. Akamai's own documentation calls it an error reference code or error string. Two points follow from that:
- It is per request. Ten blocked fetches produce ten different strings. A string from yesterday's test says nothing about today's.
- It is not a reason code. The dotted string does not encode "bot rule" or "rate limit" in a way Akamai documents for visitors. Akamai does not publicly document what each part means, so do not try to decode it yourself.
Akamai also has a related identifier. The Property Manager Global Request Number behavior can attach a unique ID to every request, as a response header, a request header to the origin, or a variable. Akamai says these GRNs use the same format as error reference strings, and the same translation tool accepts both. If a site sends the GRN as a response header, you get a traceable ID even for requests that were not denied.
Who can translate it
The translation tools Akamai documents live inside Akamai Control Center and the Akamai APIs, which need the site's own account and API credentials. A crawler operator, an SEO agency or a visitor has no documented way to look up the string. The security configuration that produced the denial is the one the site's team set, so the people to contact first are the site's own security or CDN team.
That shapes the workflow:
| Who you are | What you can do with the string |
|---|---|
| Site owner with Control Center access | Translate it yourself in Edge Diagnostics or Security Center |
| Developer or auditor without Akamai access | Send it, with the time and URL, to the site's Akamai administrator |
| Agency testing a client's site | Ask the client to translate it, or to give you Edge Diagnostics access |
| Crawler operator or visitor | Report it through the site's contact or support channel |
What to capture when you see the page
The string is most useful with the context around it. Record these while the test is fresh:
- The full reference string, exactly as shown, including every dotted part. Copy it, do not retype it from a screenshot.
- The time, in UTC, to the minute.
- The full URL you requested, including the host (
wwwor not) and any query string. - The IP address your request came from. For a test from a cloud function or a VPN, that is the egress IP, not your laptop's.
- The user agent you sent, and the HTTP status code you received.
Automated tools often log only the status code and throw the body away, which loses the reference. When testing a page with a crawler user agent, save the body and pull the string out:
curl -s -A "Mozilla/5.0 (compatible; ExampleTestBot/1.0)" \
-o body.html -w "%{http_code}\n" https://www.example.com/pricing
sed -e 's/ / /g' -e 's/#/#/g' -e 's/././g' body.html \
| grep -o "Reference #[0-9a-f.]*"
date -u +"%Y-%m-%dT%H:%M:%SZ"The sed step turns HTML character references for spaces, # and dots back into plain characters, in case the page source writes the string that way. If grep still finds nothing, open the file and look for the reference by eye.
Remember what this test proves. A request from your machine with a copied crawler user agent is not the verified crawler, as the guide to verifying real AI crawlers explains. An Edge Diagnostics translation shows the client IP and user agent, which confirms whether the denied request was yours.
Translate it in Edge Diagnostics
Akamai documents the Translate Error String tool in Edge Diagnostics. The steps, at the time of writing:
- In Control Center, go to Test & troubleshoot > Edge Diagnostics.
- In the side menu, select the tools icon, then Translate Error String.
- Enter between one and ten error reference codes. Akamai asks for them in lowercase.
- Optionally select Trace forward logs to include logs from every edge server involved in the request, not only the one where the error occurred. This can take longer.
- Click Submit. Each code becomes a separate request in the Recent requests table. If one fails, Akamai shows a failure icon in the Actions column that you can click to retry.
The Summary section is what you read first. According to Akamai, it shows the reason for failure, WAF details with a link to the WAF policy, a link to Web Security Analytics covering five minutes either side of the error, and the request facts: property, URL, HTTP response code, date and epoch time, client IP, connecting IP, edge server IP, origin hostname and IP, user agent and request method. A Logs preview follows with the raw edge log lines.
Two details in Akamai's documentation matter for crawler investigations:
- The window is short. Error logs are available for the last 6 or 24 hours, depending on server and traffic conditions. A string from a test last week will most likely return nothing here.
- The WAF link points to the current policy. It opens the current production WAF policy, which may differ from the one that was active when the request was denied. If someone changed the configuration since, compare versions before blaming a rule.
If the request completes but the results are empty, Akamai suggests waiting a few minutes and trying again, because edge servers may need time to save recent logs.
The same lookup through the API
For scripted checks, the Edge Diagnostics API has a translate error string operation. It is asynchronous:
POST /edge-diagnostics/v1/error-translator
Content-Type: application/json
{
"errorCode": "18.85c5d617.1549635923.277a7a1",
"traceForwardLogs": false
}errorCode is the alphanumeric part of the string, without "Reference #". A successful call returns 202 with an executionStatus of IN_PROGRESS, SUCCESS or FAILURE. While it is in progress, you wait retryAfter seconds and then poll GET /edge-diagnostics/v1/error-translator/requests/{requestId}, or follow the returned link. A finished result includes reasonForFailure, httpResponseCode, userAgent, the client, connecting, edge server and origin IPs with ASN and country, propertyName, wafDetails, a wsaUrl for Web Security Analytics, the filtered logLines, and a list of suggestedActions. Calls are authenticated with EdgeGrid, and the endpoint can return 429 when you exceed its rate limit; Akamai's reference shows only example values for the limit headers.
Translate it with the Application Security error translator
When the denial came from a security control rather than a delivery error, the security translator keeps data far longer. Akamai announced the Security Event Error Translator in Security Center in August 2024 for App & API Protector, Bot Manager Premier and Account Protector. It translates the reference into context about your security setup, filters Web Security Analytics down to the exact request, and works for errors up to 31 days old.
The equivalent API call is Translate an error reference code in the Application Security API:
GET /appsec/v1/error-translator/18.713f12c6.1782371012.6acdf833The response includes the event timestamp, a requestInfo block (ASN, connecting IP, country code, the HTTP status returned and the URL), a securityInfo block naming the security configuration, its version and the policy, and a wsaInfo.resultUrl link to the event in Web Security Analytics. A 404 means Akamai found no security event data for that code. The code may be older than 31 days, mistyped, or from a denial that a security product did not log, in which case Edge Diagnostics is the better place to look.
The most useful field for a crawler investigation is failureReasonCategories, an array of the protections that blocked the request. Akamai lists these values:
| Category | Akamai's description | Where to look next |
|---|---|---|
BOT | Blocked by Bot Manager | Bot Manager actions for the bot or category, as in the Bot Manager guide |
WAF | Blocked by Web Application Firewall protections | The WAF policy; open the event in Web Security Analytics |
WAF_EVALUATION | Blocked by evaluation rules | Rules being evaluated in the WAF policy |
CUSTOM_RULE | Blocked by a custom rule | The site's custom rules, which may match on user agent or path |
REPUTATION | Blocked by client reputation protections | Client reputation settings |
NETWORK_LIST | Blocked by Network List protections | The network lists the policy uses |
DOS | Blocked by Denial of Service protections | The policy's denial-of-service settings |
USER_RISK | Blocked by Account Protector's User Risk protections | Account Protector settings |
MALWARE_PROTECTION, UNCLASSIFIED | Malware Protection, or a protection outside the other categories | The event in Web Security Analytics |
The documented response fields name the configuration and policy, not an individual rule. For the exact rule or bot, follow the Web Security Analytics link.
Which tool to use
| Edge Diagnostics Translate Error String | Security Event Error Translator | |
|---|---|---|
| Where | Control Center, Test & troubleshoot | Control Center, Security Center |
| API | POST /edge-diagnostics/v1/error-translator | GET /appsec/v1/error-translator/{code} |
| Lookup window | Last 6 or 24 hours | Up to 31 days |
| Best for | Edge logs, origin and delivery errors, request detail | Which security configuration, policy and category denied it |
| Shows user agent | Yes | Not in the documented API fields |
| Products | Edge Diagnostics in Control Center | App & API Protector, Bot Manager Premier, Account Protector |
In practice: translate same-day strings in Edge Diagnostics to get the full logs, and use the security translator for anything older or when you need to know which security control acted. If the page you got was not an Akamai page at all, the guide to identifying which WAF blocked a crawler shows how to tell vendors apart from response headers, and the F5 BIG-IP rejection page has its own support ID with a different lookup.
Frequently asked questions
Can I look up an Akamai reference number myself?
Only if you have access to the Akamai account that serves the site. Otherwise, send the full string, the UTC time and the URL to the site's owner or security team.
Why does the translation return no results?
Edge Diagnostics keeps error logs for only 6 or 24 hours, and very recent logs can take a few minutes to appear. For older security denials, use the Security Center translator or the Application Security API, which cover 31 days.
Does the reference number tell me why I was blocked?
Not by itself. The translation shows the reason for failure, the security policy and, in the security translator, a category such as BOT, WAF or NETWORK_LIST.
Will Akamai support unblock my crawler if I send them the reference?
The denial comes from the security configuration the site's team set, so the change is normally theirs to make. The reference helps that team find the exact request quickly.
Is an Access Denied page the same as a challenge?
No. A denial returns a short error page with a reference, while a challenge serves an interstitial a crawler cannot complete. The post on WAF challenge pages shows how to recognise those.
Check what your pages return
Once the cause is fixed, confirm the result from outside your network. The free check shows what an important page returns to the main AI crawler user agents. For every template and edge rule reviewed with evidence, an audit covers your CDN and robots.txt together.
Filed under CDN and bot protection. Topics: bot-protection, waf, cdn, troubleshooting