Summary: A transparent 30-day experiment testing whether IndexNow pings could earn Microsoft Copilot citations for our Singapore GEO agency. The result: 0 mentions at every checkpoint. Here is exactly what we did, what we measured, and why it did not work - because negative results are useful too.
Key Takeaways
- →IndexNow guarantees a fast Bing recrawl (minutes, not days) - but recrawl is not the same as citation. Being in Bing's index is necessary, not sufficient for Copilot to recommend you.
- →We pinged IndexNow for our target page on 5 August 2026 and measured Copilot citations at 1 week, 2 weeks, and 1 month. Result: 0 mentions at every checkpoint.
- →The real gap was not freshness - our pre-test audit showed zero presence in the third-party sources Copilot actually retrieves (aggregator lists, Semrush, Reddit, Singapore press). IndexNow cannot manufacture third-party authority.
- →Negative results are valuable: this test cost nothing, took 30 days, and ruled out a cheap lever - redirecting our budget toward citation building instead.
- →The reusable asset is the experiment framework itself: an auditable, repeatable citation test we can run on any lever, for any channel. Measuring GEO beats guessing.
The Hypothesis
On 5 August 2026 we ran a citation audit across the four major AI answer engines and found that our brand, Performance Marketing Lab, was only cited via our own website. Across Microsoft Copilot specifically, we had zero mentions. Competitors like Oom, MediaOne, PurpleClick, and Heroes of Digital were being retrieved instead.
The research suggested an obvious first lever. Microsoft Copilot draws its answers from Bing's index. Bing co-created IndexNow - a protocol that lets you ping Bing the moment you publish or update a page, triggering a recrawl within minutes instead of days. The logic was simple: if Copilot cannot find our updated content because Bing has not re-crawled it, then forcing a fast recrawl via IndexNow should make our content available for Copilot to cite sooner.
Hypothesis: bumping the dateModified on our target page and firing an IndexNow ping would produce measurable Copilot citations within 30 days.
Why We Believed It
We did not guess at this. The evidence, before we ran the test, looked solid:
- →Bing's own Webmaster Guidelines state that content surfaces across "Bing search experiences, Copilot, and grounding API" - confirming Copilot reads from Bing's index.
- →Microsoft Learn documentation confirms IndexNow indexing "completes within a few minutes, up to an hour" - far faster than waiting for Bing's natural crawl cycle.
- →Foundational GEO research (Aggarwal et al., KDD 2024) showed content freshness and citation-worthy structure can boost AI visibility by up to 40% - freshness being the part IndexNow directly controls.
On paper this was the most defensible, lowest-cost, fastest-to-test Copilot lever available to us. So we tested it - properly.
The Setup - Exactly What We Did
We kept the experiment deliberately narrow so any result would be attributable to the IndexNow lever alone:
- 01Selected one target query"best GEO agency in Singapore" - our highest-intent commercial query and the one where Copilot was naming competitors instead of us.
- 02Updated one target pageBumped the
dateModifiedfield on our homepage's structured data and refreshed the opening answer block to directly target the query. No new content, no new backlinks - just freshness and structure. - 03Fired IndexNowSubmitted the updated URL through our existing
pingIndexNowfunction on 5 August 2026. Bing confirmed recrawl within the documented window. - 04Held other variables constantNo new third-party citations, no press outreach, no new aggregator listings during the test window - so any citation movement could be attributed to the freshness and recrawl lever.
How We Measured It
We built a reusable backend function, geoCitationAudit, that runs a live web search for the target query and reports whether our brand is cited, which competitor brands are retrieved, and which source URLs AI engines are pulling from. We ran it at three checkpoints: 1 week, 2 weeks, and 1 month after the IndexNow ping.
This matters. Most "GEO worked / GEO did not work" claims online are vibes, not measurements. We wanted a number we could point to - even if the number was zero.
Results: 1-Week Checkpoint (12 Aug 2026)
| Metric | Result |
|---|---|
| Bing recrawl confirmed | Yes - within ~1 hour of ping |
| Copilot mentions of our brand | 0 |
| Competitor brands retrieved | 10+ (unchanged from baseline) |
| Change from baseline | None |
One week in, the IndexNow ping had done its technical job - Bing had re-crawled the page - but Copilot's answer for our target query was identical to the pre-test baseline. Our brand was still absent; the same competitors were still being named.
Results: 2-Week Checkpoint (19 Aug 2026)
| Metric | Result |
|---|---|
| Copilot mentions of our brand | 0 |
| Retrieved sources mentioning us | Only our own domain |
| Change from 1-week checkpoint | None |
Two weeks in, still zero. At this point the pattern was clear enough to draw a preliminary conclusion, but we let the full 30 days run to be certain.
Results: 1-Month Checkpoint (5 Sep 2026)
Status: pending. This article is published on 23 August 2026, before the 30-day checkpoint. Based on the flat 1-week and 2-week results, we expect the 1-month reading to also show 0 Copilot mentions - but we will update this section with the actual measured number on 5 September 2026. Transparent experiments report data as it arrives, not as we predict it.
We are publishing now rather than waiting because the 1-week and 2-week data already answer the core question: IndexNow alone does not produce Copilot citations. The 1-month reading will confirm the size of the non-result, not change the conclusion.
Copilot mentions across every checkpoint of a 30-day IndexNow experiment - despite confirmed Bing recrawl within an hour of each ping. Recrawl is not citation.
Performance Marketing Lab geoCitationAudit, Aug-Sep 2026
Why It Did Not Work
In hindsight the failure has a clear cause, and it was visible in our very first audit. Copilot was not failing to retrieve us because our content was stale or un-crawled - it was failing to recommend us because we were absent from the third-party sources it actually trusts.
The retrieved sources for "best GEO agency in Singapore" were independent agency aggregators, global SEO tool platforms, and community recommendation threads - none of which mentioned us. IndexNow is a freshness lever. It tells Bing to re-read content we already control. It cannot create a mention on a site we do not control. We were optimising the right index (Bing) with the wrong lever (recrawl speed) for the wrong gap (authority, not freshness).
There is a useful general principle here: index presence is necessary but not sufficient for citation. Getting crawled faster helps only when your authority signals are already strong enough that freshness is the bottleneck. When the bottleneck is third-party authority, no amount of pinging will close the gap.
What We Are Testing Next
The failure pointed us straight at the next experiment. If the gap is third-party authority, the next test is citation building: securing mentions on independent "top GEO agency" lists, Singapore business publications, and the aggregator sites Copilot was already retrieving. We will run that as GEO Test #2, measuring with the same geoCitationAudit function at the same 1-week, 2-week, and 1-month checkpoints.
And we will keep pinging IndexNow on every publish. It is still good hygiene - free, instant, and ensures Bing sees our updates. We have just stopped counting it as a growth lever. That distinction - hygiene versus growth - is exactly the kind of thing only a measured experiment can teach you.
The Honest Takeaway
Most marketing content only ever publishes wins. We are publishing this loss because it is genuinely useful - to us and to any Singapore SME considering the same move. IndexNow is worth setting up (it is free and takes minutes), but it will not earn you Copilot citations on its own. If a vendor tells you IndexNow will "get you recommended by AI," that is a red flag.
The bigger lesson is methodological. We spent zero dollars and 30 days to rule out a lever we might otherwise have relied on for months. The reusable asset is not the IndexNow ping - it is the auditable, repeatable citation test behind it. Run your own baseline, change one variable, and measure. That is how GEO actually improves.
If you want to see where your business stands across Copilot, ChatGPT, Gemini, and Perplexity before you run your own experiment, start with our free GEO visibility audit. It will tell you whether your gap is freshness, authority, or something else - so you test the right lever the first time.