Links, Partnerships and Digital PR

Overview

Almost everything written about link building was written to move a page up a results list. If that outcome becomes less decisive, most of the tactics lose their purpose — and the ones that were always disguised payment become pure risk with no upside.

What remains valuable is narrower and harder: producing something worth referencing, and knowing the people who reference things. Both survive any change in how answers are assembled, because both are ultimately about human attention. This course is the outbound counterpart to Course 6: there you made yourself citable, here you make yourself known.

10.1 What do links actually do now?

What is it?

A reassessment of the link as three separate things that used to be bundled together: a possible ranking input, a traffic source, and a piece of public evidence that someone credible referenced you.

Why it matters

Only the third is reliably durable. Ranking effects are outside your knowledge and subject to change; referral traffic from most links is negligible; but a public reference from a credible source is Authoritativeness evidence (Course 7.2), it is discoverable by anything assembling answers, and it does not expire. Separating the three stops you paying for the wrong one.

How to do it

1. Audit existing links by what they actually deliver: measurable referral traffic, credible public evidence, or nothing.
2. Classify how each was acquired: earned by the work, exchanged, paid, or self-placed. Be honest — this determines your risk.
3. Identify anything that would embarrass you if published. Paid links presented as editorial are both a policy and a disclosure problem.
4. Stop measuring aggregate authority scores as an objective. Vendor scores are estimates of an unknown function, not a business metric.
5. Rebuild the target list around people and publications your actual audience reads — the sources whose mention would matter to a reader.
6. Record, for each link, whether it would still be worth having if it passed no ranking value at all. If not, do not pursue that kind again.

Tools needed

Your referral analytics, a list of existing links, and the Course 7.2 evidence map.

Lab 10.1

Audit every link you know about. For each, record measured referral traffic, whether it constitutes credible public evidence, and how it was acquired. Then answer the key question in one column: would this still be worth having if it passed no ranking value?

Artifact

Link value audit
  Source | Acquisition (earned / partnership / exchanged / paid / self-placed)
  Measured referral visits (90 days) | Credible evidence? Y/N | Audience overlap
  Worth having with ZERO ranking value? Y/N
  Count of links that fail that test: ___ | Tactics to stop: ___

Common mistakes

• Treating a vendor authority score as a business objective, then optimising an estimate of an unknown function.
• Pursuing links from sites your audience never reads, which delivers neither traffic nor credible evidence.
• Paying for placements without disclosure, accepting real exposure for a benefit that may already be diminishing.

Pro astuces

• "Would I still want this with zero ranking value?" is the cleanest filter available. It eliminates most of the industry's tactics in one question.
• A mention without a link from a source your audience trusts is often worth more than a link from one they have never heard of.

10.2 How do you run an original-data campaign?

What is it?

Turning the primary research from Course 6.2 into a campaign: a specific finding, a documented method, a usable dataset, ready-made assets, and a small list of people to whom the finding is genuinely relevant.

Why it matters

Original data is the only outreach that gives the recipient something they need. Everyone in your niche receives dozens of requests to link to a guide; almost nobody receives a defensible number about their own market that they can quote. This is also the highest-leverage reuse in the whole curriculum: one collection routine feeds the research asset, the API, the newsletter and the campaign.

How to do it

1. Choose a question with a genuinely unknown answer that matters commercially to your audience. If the answer is already published, there is no campaign.
2. Design the smallest method that can answer it defensibly, and write down its limitations before you collect anything.
3. Collect, then report the finding as one specific sentence with a number, a period and a sample size.
4. Publish the full dataset and the method alongside the finding — via the versioned API from Course 9.5 if you have one.
5. Build the pack: the finding, the method note, a chart, the raw data, and a two-line summary someone can quote without misrepresenting you.
6. Approach a short list of people for whom the finding is directly relevant, and never claim more than the data supports.

Tools needed

Your dataset, the Course 9.5 endpoint, a charting approach, and the claim register for every number you publish.

Lab 10.2

Run one campaign end to end from data you collected yourself. Build the full pack. Approach ten genuinely relevant people. Record every response, including the silences, and note which element of the pack recipients actually used.

Artifact

Data campaign pack
  Question | Why the answer was unknown | Method | Sample | Collection window | Limitations
  Finding (one sentence, with number and period)
  Assets: chart | dataset/API URL | quotable two-line summary
  Approaches: 10 | responses | mentions | which pack element was used
  Every published number logged in the claim register? Y/N

Common mistakes

• Surveying an unrepresentative sample and reporting a headline the data cannot support — which is how a campaign becomes a correction.
• Publishing the finding but withholding the dataset, removing the reason anyone would cite it.
• Approaching a mass list rather than the small number of people to whom the finding actually matters.

Pro astuces

• Stating the limitations in the pack increases pickup. It shows the recipient exactly what they can safely say, which is the thing they are worried about.
• A longitudinal campaign repeated annually becomes the reference in a niche, and the second year costs a fraction of the first.

10.3 What does ethical outreach look like?

What is it?

Small volumes of researched, specific, honest approaches to named people, with a written policy covering what you will never do — and full disclosure of any commercial element.

Why it matters

Mass outreach has a dismal response rate and it damages the reputation you are trying to build, in a niche where the relevant people all know each other. Twenty researched approaches routinely outperform a thousand templated ones. And where anything of value changes hands, disclosure is not optional — read the FTC endorsement and review guidance before you offer anyone anything.

How to do it

1. Research before writing. Read what the person actually published and reference it accurately, or do not contact them.
2. Lead with what is useful to them — the finding, the dataset, the correction — not with what you want.
3. Be explicit about who you are, which site you represent, and that the three sites are commonly owned if that is relevant.
4. Never offer or accept payment, product or reciprocal placement for editorial coverage without clear disclosure by both parties.
5. Follow up once, politely, then stop. A written policy on follow-ups prevents the drift into harassment.
6. Log every approach and every outcome, including refusals and requests not to be contacted — and honour those permanently.

Tools needed

A simple contact log, your data campaign pack, and your published editorial policy.

Lab 10.3

Write the outreach policy including the "never" list. Then send twenty researched, individually written approaches for your Course 10.2 campaign. Log every outcome. Compare the response rate to any templated campaign you have run before.

Artifact

Outreach log & policy
  POLICY: what we always disclose | what we never do (the "never" list) | follow-up rule (once)
  LOG: person | publication | what they published that made them relevant | date sent
      | response | outcome | do-not-contact? (permanent)
  Response rate: ___ / 20  | Comparison to templated approach: ___

Common mistakes

• Personalisation tokens standing in for actual research, which recipients recognise immediately.
• Repeated follow-ups, converting a neutral contact into a permanent negative.
• Undisclosed paid placements, which risk both platform policy and consumer-protection exposure.

Pro astuces

• The single most effective opening is a genuine, specific correction or addition to something the person recently published — offered without asking for anything.
• Twenty approaches you would be happy to have published verbatim is the right volume for a three-site portfolio.

10.4 How do partnerships beat link building?

What is it?

Ongoing relationships with complementary publishers, tool vendors, associations and practitioners — producing co-created assets, shared datasets, cross-promotion to owned audiences, and mutual referrals.

Why it matters

A partnership produces repeated value from one negotiation, and it operates on channels no intermediary controls: their newsletter, their community, their audience. It is the most direct answer to the governing question available from outside your own site — and unlike a link, it cannot be devalued by an algorithm change.

How to do it

1. Target complementary rather than competing: a tool vendor whose users need your analysis, an association whose members need your data.
2. Lead with a concrete co-creation proposal, not a request for exposure. Bring the dataset, the tool or the audience.
3. Write down the terms: what each side contributes, who owns the output, how it is disclosed, how either side exits.
4. Prefer owned-channel exchange — newsletter cross-promotion reaches engaged people directly and is measurable.
5. Disclose the relationship publicly wherever the partnership touches editorial or commercial content.
6. Review annually against the concentration limits: a partner supplying more than a quarter of your traffic or revenue has become a dependency.

Tools needed

Your dataset and tools, your newsletter, the reviewer register, and written agreements.

Lab 10.4

Identify five complementary organisations, approach three with a specific co-creation proposal, and take one to a written agreement covering contribution, ownership, disclosure and exit. Then ship the co-created asset.

Artifact

Partnership agreement
  Partner | Why complementary (not competing) | What each side contributes
  Output & who owns it | Public disclosure wording | Review date | Exit terms
  Concentration check: partner share of traffic ___% revenue ___% (limits: ≤60% source, ≤25% sponsor)

Common mistakes

• Partnering with direct competitors for the same keywords, which produces friction rather than reach.
• Verbal agreements with no ownership clause, which fail exactly when the asset becomes valuable.
• Letting one partner become a dependency, breaching a concentration limit while it feels like success.

Pro astuces

• Tool vendors are the most under-used partners for a data publisher: they have the users, you have the analysis, and neither of you competes.
• A recurring annual joint dataset is the strongest partnership structure — both parties are invested in its accuracy and both must cite it.

10.5 How do you become the source people name?

What is it?

Positioning a real, named person as the go-to expert for a specific narrow question — so that journalists, podcasters, community moderators and answer engines all have an obvious entity to attribute.

Why it matters

Attribution attaches to people more readily than to brands. A named expert with a track record is quotable, invitable and rememberable in a way that "the editorial team" never is. It also compounds: each appearance makes the next one easier, and none of it depends on a ranking.

How to do it

1. Choose one narrow question you can credibly own — narrow enough that you are the obvious choice, not the fifth option.
2. Build the proof: the recurring dataset, the tool, the tracked measurement that makes you the person with the numbers.
3. Publish a real source page: who you are, exactly what you can speak to, what you will not speak to, and how to reach you fast.
4. Be genuinely responsive. Availability at short notice is the single largest factor in being asked again.
5. Show up consistently in the communities your audience already uses, being useful without a link in every message.
6. Keep a record of every appearance and citation — this is the Authoritativeness evidence from Course 7.2 accumulating.

Source positioning kit
  The one question I own: ___
  Proof I hold: recurring dataset / tool / tracked measurement (with URLs)
  Named expert: name | credentials (verifiable) | canonical page URL
  Will speak to: ___    Will NOT speak to: ___
  Contact route and typical response time: ___
  Appearance log: outlet | date | topic | named? | link? | audience overlap

Tools needed

Your author page, your recurring dataset, and a public citation log.

Lab 10.5

Write the source positioning kit, publish the source page, and make three genuine contributions in communities your audience uses — with no link in any of them. Then record what happened over the following month.

Artifact

Source positioning kit, live source page, and a 30-day record of contributions and inbound requests

Common mistakes

• Positioning too broadly — "digital marketing expert" is nobody's obvious choice.
• Claiming expertise without the underlying measurement, which collapses on the first serious question.
• Slow responses, which quietly remove you from the shortlist without any feedback.

Pro astuces

• Stating what you will not comment on makes you more credible, not less — it signals that your yes means something.
• Answering within an hour, even to decline and refer someone better, is what makes you the first call next time.

10.6 How do you measure PR without vanity metrics?

What is it?

Judging outreach and PR by durable outcomes — relationships that recur, subscribers acquired, citations that persist, partnerships signed — rather than by mention counts, reach estimates or advertising-value equivalents.

Why it matters

PR is the easiest activity in the portfolio to fake success at, because the available metrics are large, flattering and mostly meaningless. "Reach" is usually an audience estimate multiplied by optimism. Given the Course 7.5 cost of a data campaign, you need to know whether it produced anything durable.

How to do it

1. Define durable outcomes before the campaign: subscribers acquired, partnerships opened, citations that persist, repeat relationships.
2. Measure subscribers attributable to the campaign, because that is the outcome you keep.
3. Count recurring relationships: how many people referenced you a second time without being asked. This is the real PR metric.
4. Check whether citations persist after twelve months. A mention that survives is evidence; one that vanishes was traffic.
5. Compute cost per durable outcome using the fully loaded campaign cost from Course 7.5.
6. Never report reach or advertising-value equivalent internally. If it would not survive a sceptical question, do not put it on the dashboard.

Tools needed

Your email platform, the revenue calculator, your citation log, and a twelve-month reminder.

Lab 10.6

Take your Course 10.2 campaign and compute its cost per durable outcome using the loaded cost. Then decide, in writing, whether to repeat it — and set a twelve-month reminder to check which citations still exist.

Artifact

PR outcome record
  Campaign | Fully loaded cost (Course 7.5) | Subscribers acquired | Partnerships opened
  Citations at 30 days ___ | at 12 months ___ | Repeat referencers ___
  COST PER DURABLE OUTCOME: ___
  Repeat decision: yes / no / modify — and why (written before the next campaign)

Common mistakes

• Reporting reach to yourself, which makes an unprofitable campaign look like a success and guarantees you repeat it.
• Never checking citation persistence, so a one-week spike is remembered as lasting authority.
• Omitting the loaded cost, which makes cost per outcome unknowable and the repeat decision arbitrary.

Pro astuces

• Repeat referencers is the most predictive PR metric there is. Someone who cited you twice unprompted will likely do it again — and that is a relationship, not a placement.
• If a campaign produced no subscribers and no partnerships, it produced nothing durable, regardless of how the mention count looked.

Scenario assessment: the agency package and the flattering report

An agency offers 15 guest placements a month on "high-authority" sites for a flat monthly fee, with a report showing authority scores and estimated reach. Separately, your own data campaign — total loaded cost equivalent to four months of that fee — produced 6 mentions, 2 of them from publications your audience actually reads, 140 attributed subscribers, and one association that now wants to co-publish the dataset annually. Your Month 6 gate is approaching. The agency's report shows numbers that dwarf your campaign's on every metric it uses. A separate offer arrives: a vendor will pay for a favourable placement in your comparison table, "no need to mark it as sponsored".

Final project: one campaign, one partnership, one named source

Build attention that does not depend on ranking: a defensible finding, a real relationship, and a person people can name.

Deliverable
1. Link value audit with the "worth having with zero ranking value?" column completed for every link
2. One original-data campaign run end to end: question, method, limitations, dataset published, full pack built
3. Outreach policy including the written "never" list, plus 20 logged researched approaches
4. One written partnership agreement: contribution, ownership, disclosure, review date, exit terms
5. One co-created asset shipped with the partner
6. Source positioning kit and a live source page for a named expert, with the narrow question stated
7. Three genuine community contributions made with no link included
8. PR outcome record with cost per durable outcome and a written repeat decision

Pass standard: you can state cost per durable outcome for the campaign, you have one signed partnership with an ownership clause, and no tactic in your programme fails the zero-ranking-value test.

Course 10 checklist

• Every known link audited by acquisition method and measured referral traffic
• "Worth having with zero ranking value?" answered for every link
• Tactics that fail that test stopped, and listed
• Vendor authority score removed as an objective
• Campaign question genuinely unanswered before you measured it
• Method and limitations written BEFORE collection
• Full dataset published, not gated
• Finding stated as one sentence with number, period and sample size
• Every published number logged in the claim register
• Outreach policy written, including the "never" list
• 20 individually researched approaches sent and logged
• One follow-up maximum; do-not-contact requests honoured permanently
• Common ownership of the three sites disclosed where relevant
• No undisclosed paid or exchanged editorial placements anywhere
• Partnership agreement in writing with an ownership clause and exit terms
• Partnership disclosed publicly where it touches editorial or commercial content
• Partner concentration checked against the ≤25% sponsor and ≤60% source limits
• One narrow question chosen that a named expert can credibly own
• Source page states what the expert will and will not speak to
• Cost per durable outcome computed from the fully loaded campaign cost
• Reach and advertising-value-equivalent excluded from all reporting
• 12-month citation-persistence check scheduled

AI-agent prompt for Course 10

Act as a digital PR and partnerships strategist for an independent publisher. Operating assumption: the ranking value of a link is uncertain and may be declining, so the only outcomes that count are durable ones — subscribers, recurring relationships, persistent citations, signed partnerships.

I will paste: my existing link list, my datasets, my niche, my audience, and my fully loaded cost per asset from Course 7.5.

Do this:
1. Audit my links. For each, classify acquisition (earned / partnership / exchanged / paid / self-placed), state measured referral value, and answer: WOULD THIS BE WORTH HAVING IF IT PASSED NO RANKING VALUE? List the tactics I should stop.
2. Design one original-data campaign from data I already hold or could collect cheaply. Give me: the unanswered question, the smallest defensible method, the limitations to publish, and the finding stated as ONE sentence with a number, a period and a sample size.
3. Build the campaign pack outline: finding, method note, chart, dataset/API link, and a quotable two-line summary a recipient can use without misrepresenting me.
4. Produce a target list of at most 20 NAMED people with, for each, the specific thing they published that makes them relevant. No mass lists.
5. Draft my outreach policy including an explicit "never" list, and a one-follow-up rule.
6. Propose 5 COMPLEMENTARY (never competing) partnership targets with a specific co-creation proposal each, and give me an agreement outline covering contribution, output ownership, public disclosure, review date and exit.
7. Write a source positioning kit for one named expert: the single narrow question they can own, the proof required, and what they will explicitly NOT speak to.
8. Define the PR measurement: subscribers acquired, partnerships opened, citations persisting at 12 months, repeat referencers, and cost per durable outcome using my loaded cost.

Rules:
- Never propose paid links, link exchanges, undisclosed sponsored placements, fake personas, invented credentials, or scraped personal email addresses. Refuse if I ask.
- Never report or forecast reach, estimated impressions, advertising value equivalent, or vendor authority scores. Say explicitly that these are excluded and why.
- Treat disclosure as legal: state that requirements vary by jurisdiction, reference that endorsement and review guidance applies, and note you are not giving legal advice.
- Do not invent response rates, pickup rates or benchmark link counts. Mark anything I have not measured as UNMEASURED.
- Flag any partner that would exceed 25% of revenue or 60% of traffic as a concentration breach before recommending it.
Output as tables, with the stopped tactics, refusals and excluded metrics called out in plain text above the tables.

Primary sources

• FTC endorsement, influencer and review guidance — the primary reference for disclosure whenever anything of value changes hands for coverage.
• Search spam policies — including the treatment of link schemes and paid links.
• Creating helpful, reliable, people-first content — the originality and first-hand-experience questions that a data campaign is designed to satisfy.

Platform policies on links and disclosure requirements both change. Every claim here is a dated claim with a 90-day expiry — log it in your register and re-verify against the primary source.