Update a page, write a new article, or earn a mention: how to choose

When an AI answer cites other pages instead of yours, choose the action by asking where the missing evidence lives. If a page of yours already covers the topic but lacks something the cited pages have, update that page. If you have no page that answers the question, write a new article. If the pages being cited are third-party pages, such as a publisher's roundup or a review site, no edit on your own site can put your brand into them, so the task is an earned mention: a reason for that publisher to include you. The gap is an observation, and each choice is a hypothesis to test.

In short

• Same gap, three possible jobs. The deciding question is whether the missing evidence belongs on your site or on someone else's.
• Update a page when you have a matching page with a specific, nameable shortfall. Write a new article when no page answers the question. Pursue an earned mention when third-party pages carry the answer.
• A gap can call for more than one action. Choose the one that removes the biggest visible difference first, and log the others.
• "Hold off" is a legitimate outcome. Some gaps are not worth a task yet.
• Record what you chose, why, and what you will check afterwards. Without that record you cannot tell later whether the choice worked.

Why does the same gap call for different work?

Because a citation gap describes a difference in evidence, not a task. A citation is a link in an AI answer pointing at a specific URL as a source. A citation gap is the difference between what the pages engines cite for a prompt contain and what your own page offers. In DiscoveredBy, a prompt becomes a gap candidate when your brand appears in under 30% of that prompt's own runs over the trailing window and at least one citation in that window points at a domain that is not yours. See the citation gap glossary entry for the short definition.

That description says nothing about who should fix it or where. Two teams can look at the same list of cited URLs and disagree: one wants to rewrite the pricing page, the other wants to pitch a publisher. Both may be right about the evidence and wrong about the fit. The choice below makes the disagreement explicit and settles it with a few observable facts.

This post assumes you have already found a gap worth working on. If not, start with finding the citation gaps to fix first. If you have chosen the action and need to hand the work to a writer, see turning a citation gap into a content brief.

Enlarge image

What are the three actions, exactly?

Each action changes a different thing, takes a different kind of effort, and is checked differently.

• Update a page means editing a page you already own so that it contains something the cited pages contain: a comparison table, an FAQ, a figure, a clearer heading, structured data, a fresher date. It is usually the smallest change, and easier to compare before and after because the page and its address stay the same.
• Write a new article means publishing a page that does not exist yet, for a question none of your pages answers. It is more work than an edit, and there is no earlier version of the page to compare against.
• Earn a mention means getting your brand into a third-party page that AI engines already cite. You do not control the page; another party does. The work is outreach, evidence and relationships, not editing.

There is also a fourth outcome, hold off. The DiscoveredBy Articles docs list "holding off entirely" as one of the asset outcomes a brief can recommend. A gap with weak evidence, a prompt that does not reflect a real buying question, or a page already being reworked can all justify waiting.

Does the missing evidence belong on your site or someone else's?

This is the first and most important cut. If the pages carrying the answer are third-party pages, editing your own site cannot fix that. If they are competitor pages or general pages on topics you could cover yourself, your site is the right place.

Look at the cited URLs for the prompt and sort each by whose site it sits on and what kind of page it is.

Do not read this table as a rule about what engines prefer. It sorts observed citations by where you can act, and nothing more. The Sources area and the source types docs help you see what kinds of pages are being cited.

Enlarge image

When should you update an existing page?

Update a page when two things are true: a page of yours is already the natural answer, and you can name what it lacks. The name matters. "Improve the content" is not a shortfall; "the cited pages all have a comparison table and ours has none" is.

DiscoveredBy looks for exactly this. For each candidate prompt it reads the most-cited competing URLs and looks for the page of yours most likely to compete: an explicit mapping if you set one, otherwise a page of yours already cited for that prompt, otherwise your page with the closest keyword overlap. It then compares the pages for concrete signals and records each difference as evidence. When a page of yours is behind the gap, the Optimizations screen (on plans that include it) can draft the change for that page, and each proposed change has to carry at least one real quote from a scraped source page. The docs are explicit that this checks the evidence, not the draft body, so read a draft as a starting point.

Two cautions belong here:

• If your page is being retrieved but not credited, the problem may not be content. Check that first with the retrieved versus cited docs.
• A page update is easy to make and hard to judge afterwards. Decide the before-and-after method before you edit; did your content update help covers it.

When should you write a new article?

Write a new article when no page of yours answers the question at all. In DiscoveredBy's terms, a gap with no matching page of yours is one where the opportunity is to create one, and such a gap never becomes an optimization, because there is no page to optimize.

The Articles docs describe the flow: accepting a citation gap creates a candidate topic, or you type your own title. Nothing is written until you pick a topic. The brief decides the asset: update a page you own, publish something new, or hold off. Only a new article or a comparison page gets a drafted body. For other outcomes, the brief is the deliverable.

A new article has a specific risk: it can duplicate a page you already have under a different name. Before you commit, search your own site for the question. If a weaker page exists, an update is usually the cheaper and cleaner choice.

Some drafts stop and ask for input first. For a brand new page on plans that ground drafts in sources, the docs say a brief that needs proof only you can supply, such as pricing or customer proof, can pause generation at "Needs your input". Plan for that: the facts have to come from you.

When is an earned mention the right task?

An earned mention is the right task when third-party pages carry the answer and your brand is absent from them. The evidence you need does not live on your site, so there is nothing to edit.

DiscoveredBy's Earned sources workflow starts from this observation. It lists third-party pages cited in completed answers in the window you pick, excluding your own domain and your active tracked competitors, ordered by how many distinct completed outputs cite them. You can save a page as an opportunity, assign a teammate, add notes and a follow-up date, and track a status: new, planned, outreach in progress, placement reported, or dismissed.

The limits are worth stating plainly:

• The tool does not send messages, find contacts, buy placements or publish anything. Outreach happens outside it.
• The counts are a way to find observed sources. The docs say they are not a source-authority score, a likelihood of placement, or a claim that the page is missing your brand.
• To check whether a page already names you, use the literal-match view in Source mentions, and remember that missing or old scraped text limits what it can establish.
• A reported placement is a team-entered claim, not independent verification.

For the practical side of choosing which pages to approach, see finding third-party pages worth approaching.

The decision matrix

Answer these questions in order and stop at the first row that fits. Copy the table into your working document.

Question 2 can send you in two directions at once. When both a publisher's roundup and competitor product pages are cited, log two tasks: an earned mention and a page update or article. They are separate work with separate owners.

A worked example

Illustrative example: Quillstone and its competitors are fictional, and the numbers are made up to show the method.

Quillstone sells document-review software to mid-sized legal and compliance teams. Its competitors are Brieflane and Clausewise. Over 28 days, its team reviews three gaps.

Working through the matrix:

• Gap A. The prompt is a real buying question. The cited pages are competitor feature pages, so the evidence belongs on Quillstone's site. Quillstone has a features page that should answer it, and the shortfall is nameable: the competitors show a clause-comparison table and Quillstone's page describes the feature in prose only. The action is update a page. The hypothesis is that adding a table changes what is retrieved and credited; the team writes down the before-and-after basis first.
• Gap B. The cited pages are guides, and Quillstone has no page on the topic. The team searches its own site to confirm nothing weaker exists. The action is write a new article. The brief will ask for real proof, such as a sample checklist from an actual review, so someone with that material is named as the input owner.
• Gap C. One third-party roundup appears in 9 of 24 answers, and it omits Quillstone. Nothing on Quillstone's own site can alter that page. The action is earned mention: save the roundup as an opportunity, use the literal-match view to check whether the saved text of that page names the brand (missing or old scraped text limits what it can show), and assign a teammate to prepare a factual, non-promotional case for the editor.

Gap C's 5 of 24 is 20.8%, shown as "about 21%". All three prompts sit under the 30% threshold. The team also logs a secondary task for Gap C: the competitor product pages cited alongside the roundup might justify a page update later, but they chose not to do both at once, so they could tell which action moved anything.

Note that none of these choices is a prediction. Each is a reasoned bet, logged with the way it will be checked.

The next-task log

Copy this into a spreadsheet or document. One row per gap, filled in before work starts.

NEXT-TASK LOG

Gap / prompt (exact text):
Date reviewed:
Window and engines examined:
Completed answers in window:
Answers naming my brand:                 (count and share)
Answers citing my domain:                (count and share, if relevant)
Cited pages, grouped by site type:       (mine / competitor / publisher / review site / other)
Page of mine that should answer this:    (URL, or "none found after searching my site")
Specific shortfall vs cited pages:       (or "cannot name one")

DECISION (choose one; log a second row if two apply)
[ ] Update a page:        URL:            Change:
[ ] Write a new article:  Working title:  Input owner for proof:
[ ] Earn a mention:       Target page:    Owner:      Follow-up date:
[ ] Hold off:             Reason:         Re-check date:

WHY THIS ONE
Where the missing evidence lives:         (my site / other site / both)
What I am NOT claiming:                   (e.g. "this will not guarantee a citation")

HOW I WILL CHECK
Baseline (before):                        (brand visibility and citation rate for this prompt, date range)
Check dates:                              (e.g. day 7, 14, 30)
Comparison basis:                         (same prompts, same engines, same window length)
Result and date:
Notes on other changes in the same period:

If you use DiscoveredBy, some of this lives in the product already. The To do list under Actions puts optimizations, citation gaps, earned sources and other open work in one prioritised list, and splits them into tabs by where the work happens: "On your site" for optimizations and gaps (with Search Console items), "Other sites" for earned sources. Priority bands there come from one rule per kind of item (a score for optimizations and gaps, the status for earned sources), and the docs are clear that scores from different sources measure different things. Page issues groups the same work by page, which helps when several gaps point at one URL.

Enlarge image

How do you check whether the choice worked?

Check at fixed points against a baseline you recorded before the change, and read the result as an observation, not a proof.

DiscoveredBy's docs describe this for each route. A citation gap marked implemented, or an article published and linked back, is re-read at 7, 14 and 30 days. For a reported earned placement, the Earned sources screen compares equal periods before and after, excluding the placement day, and marks a period with fewer than 30 completed outputs as provisional. The docs also state that a higher rate does not establish that the outreach caused it.

The same discipline applies to every action. Change one thing at a time where you can. Note anything else that shifted in the same period, such as new competitor pages or engine changes. If the result is flat, log that as a result too.

Common mistakes and what this cannot tell you

• Treating the gap as the answer. A gap shows a difference in evidence. It does not show that closing it will win a citation. Every task in the log is a hypothesis.
• Defaulting to a new article. New pages feel productive, but if a weaker page already exists on the topic, an update is often the more direct test.
• Defaulting to an edit when the evidence is elsewhere. If a publisher's roundup carries the answer, polishing your own page addresses a different problem.
• Reading discovery counts as authority. A page cited often is a page worth understanding, not a page that is guaranteed to accept you.
• Skipping the retrieval and access check. A page not credited may be a delivery problem, not a content problem.
• Bundling actions. If you update the page, publish an article and pitch a publisher in the same week, you cannot tell which one mattered.
• Overstating the reasons. Cited sources and stated reasons describe an answer. They do not prove why an engine produced it.

Frequently asked questions

Can one citation gap need more than one action?

Yes. If a publisher's roundup and competitor product pages are both cited, the roundup calls for an earned mention and the product pages point to an update on your site. Log them as separate tasks with separate owners. If you can, start with one so you can see its effect.

Is updating a page always cheaper than writing a new article?

Usually in effort, and often in time, but not always. An update needs a page that should answer the prompt and a shortfall you can name. If no such page exists, or the change amounts to a rewrite, a new article may be the more honest description of the work.

Does DiscoveredBy tell me which of the three to choose?

Partly. Citation gaps distinguish a gap with a matching page of yours from one without. Optimizations cover the first case, and Articles covers the second, where its brief can also recommend updating an owned page or holding off. Earned sources is a separate list of cited third-party pages. The choice between the three routes is still a judgement your team makes, using the evidence each screen holds.

Can I earn a mention without contacting a publisher?

Sometimes, for example by maintaining your own accurate profile on a directory or review site where that is allowed. For editorial pages, a publisher decides what to include, and DiscoveredBy does not send outreach or place content for you. Keep any approach factual and useful to that publisher's readers.

How long before I know whether it worked?

The product's own checkpoints for a marked-implemented gap are 7, 14 and 30 days, and a seven-day earned-source comparison can only be read after the full post-placement window has passed. Short windows are noisy, so treat early reads as provisional.

Next step

Open your highest-impact citation gap, fill in one row of the next-task log, and make the choice before opening an editor. DiscoveredBy's Optimize features bring gaps, drafts, articles and earned sources together so the evidence for each choice is on hand. To start, sign in to DiscoveredBy.