Ecommerce link building. We commit to the share that lands on pages that sell.

Most ecommerce campaigns quietly turn into blog links. They are easier to place and they count the same on a report. So before you sign we agree a minimum share of placements that has to land on category and product URLs, we report the actual figure against it every month, and we do not invoice the shortfall.

Here is the thing that goes wrong with ecommerce link building, and it goes wrong quietly.

Publishers will link to a guide. They will hesitate over a collection page and usually refuse a product URL, because there is no editorial reason to send a reader to a checkout. So a campaign that is paid by the link drifts toward the placements that are easy to get. Twelve months later the report shows ninety links and the pages that take the money have not moved.

Nobody in this category will put a number on it. We do.

Before you sign, we agree a floor: the minimum percentage of placements that must resolve to a commercial URL, meaning a category, collection or product page and not a blog post or a guide. It is written into the engagement. Every monthly report shows the actual figure next to it. If we come in under, the shortfall is either replaced or it is not invoiced.

That costs us money, and it should. Commercial destinations take more pitches, more rewrites and more refusals per placement than editorial ones.

What the floor is

It depends on your category and how much linkable material your site already has, so we set it during the audit and not on this page. What we will say is that we have never agreed one below half, and that a campaign proposing a floor in the low tens is telling you something about itself.

The failure log

Every report carries a section listing placements where the target URL we asked for could not be used, what went live instead, and why the publisher refused. It is the least flattering page in the document. It exists because a destination column without it lets us quietly redefine what counts, and you would have no way of seeing it.

One redacted report from live work. No form, no email address, no calendar booking to see it. Zero of the seven pages we could read on this search result show one, and several describe reporting in detail without ever exhibiting it.

Every column is in it, including the ones that make a month look worse: publication, live URL, the link attribute, the target page and its type, whether that target is commercial or editorial, the running share against your agreed floor, the date it went live, and whether it was still live at thirty days.

Take it to whoever you use now and ask for the same columns.

A sample proves the format, not the performance. It is one month of one engagement, and we chose it.

Link reports answer the agency's question, which is what we did. Yours is different. You want to know whether the pages we pointed at made more money, and a list of URLs cannot tell you that.

So the measurement is designed before the campaign starts, not assembled afterwards from whatever looks good.

We name the collection and product URLs receiving links. We name a control set. Those are comparable pages on your own site receiving nothing, matched as closely as your catalogue allows, and then we report organic sessions and organic revenue on both sets over a stated window, from a named tool with the view documented.

That design goes in the proposal, before you have paid anything, which means you can judge whether it is a fair test while it still costs you nothing to walk away.

It also means the result can come back bad, in a document we have already committed to publishing. Matched controls are not flattering. That is the point of using them, and it is why almost nobody does.

Product roundups and gift guides are the one common format that links straight to a product page, which is why every agency on this search result sells them and why we do too.

What none of them tells you is how those pages make money. Open any "best [category]" guide and inspect an outbound link. A large share of them carry an affiliate tag, a redirect through a tracking domain, or a rel attribute that stops equity passing.

That is not a scandal. It is the business model, and it produces three facts that change what you are buying.

Many publishers will only include a product that carries an affiliate programme on a network they already work with, at or above the commission rate their existing entries pay. If you have no programme, or a rate below the incumbents, a large share of this inventory is closed to you at any budget. We tell you which share at scoping, not in month three.

A cloaked or nofollowed roundup link still puts you in front of a buyer with a card in their hand, which is often worth more than the equity would have been. But it is not a link, and we do not invoice it as one. The attribute is in the report on every placement, and attributed placements are listed separately.

Auto-affiliatisation tools can rewrite an existing editorial link into a cloaked redirect months after publication, without the writer touching the page. A link you paid for stops passing equity and nothing announces it. We re-check every placement at thirty days and again at six months, and we tell you when one has been converted.

Our whole argument is that authority lands on category pages and assets and then has to travel inward to the pages that convert. If that is true, the travelling is not a detail. It is the step where the work becomes revenue.

Which means an agency that sells you the links and hands you the transmission has kept the easy half and given you the part that decides whether any of it worked.

Internal link changes, within ten working days

We give you the map. You have to be able to ship it. If your theme changes sit in a dev queue for a quarter, the placements alone will not move the pages you care about, and we would be taking money for a result we know will not arrive. Check this against your own release process before you talk to us. We check implementation at thirty and sixty days and record it in the report, including when it has not happened.

The things a publisher will ask for

Roundup and gift guide placement depends on things outreach cannot fix. The product in stock through the season. A price inside the band that page covers. Shipping to the publisher's market. Usable imagery. Samples available on a lead time. If a product goes out of stock the week a guide publishes, the entry gets pulled and nobody is at fault.

Long-lead print and the big publishers commission four to six months out. Miss that and you have missed it for the year.

New digital commissions are compiled roughly two to three months before the peak, which is where most of the pitching happens.

Refreshes of guides that already rank are updated through October and November, and this is the honest part: a store arriving late can still get into these, and they are the weakest of the three routes. The page already ranks, so you inherit its position instead of helping build it, and the entry usually goes to the bottom of a list somebody else has held for two years.

Q4 capacity

We hold a fixed number of Q4 pitch slots and publish how many remain, with the date they were last counted. When they are gone we stop selling gift guide work for that season, including to people already paying us. A principle would not survive an October sales call. A number does, and you can watch whether it ever moves.

One store per category per season

A gift guide has a fixed number of slots and we are not going to pitch two of your competitors into the same list. For Q4 roundup work we take one store per product category per season. If yours is taken you will hear that on the first call, not after a proposal.

The refusal that matters more

Money spent in month one does not move revenue until roughly months four to eight. So a store that arrives in Q3 hoping to change this Q4 cannot be helped at any price, by us or by anybody, and we will say so instead of selling you a campaign that pays out next year while you are looking at this year's numbers.

Coupon and voucher directories

This is the largest cheap inventory in ecommerce and the argument against it is not link quality. Those pages seed branded coupon demand. They teach your buyers that a code exists, and a shopper who leaves checkout to hunt for one either finds a discount you did not want to give or does not come back. You can check this yourself in ten minutes: look at your branded plus "coupon" or "discount code" query volume, and your code redemption rate against orders. That is a gross margin cost on transactions you had already won. We do not place these and we do not invoice them.

Product directory submissions

Sold by the hundred and read by nobody, and indistinguishable in a link profile from the networks that get devalued. There is no version of this that is worth your money, and the only reason it persists is that it makes a link count look good. If a proposal you are comparing includes volume at a price that seems impossible, this is usually what is filling it out.

Undisclosed incentivised reviews

If we send a publisher your product, that is an incentive, and disclosure rules in the US, UK and EU put the exposure on the advertiser. That is you. Not the publisher, and not us. So a disclosure goes on, we do not decide its wording, and the publisher keeps the final say. We also do not invoice a placement where the disclosure was removed after publication at anyone's request. Seeded product also costs you goods, shipping and returns per placement. We put that in the plan as a line so it does not arrive later as a surprise. This is our practice, not legal advice.

When the arithmetic does not close

Link building is a fixed cost that pays back through incremental organic sessions. Below a certain average order value and contribution margin, the sessions a realistic campaign can add will not repay the spend inside a year. Paid or conversion rate work is the better buy. We run that calculation with you during the audit, and we have talked stores out of buying on it.

When the search results are already owned

On bare transactional and SKU-level queries, shopping units and marketplace listings hold everything above the fold, so an organic win sits where nobody looks. What is worth buying is the space around them. Comparison queries, best-of-category, buyer's guides, use-case searches. And we will not take a campaign whose primary targets are owned end to end by a marketplace for that product type, which turns away a large and well-funded slice of this market.

When you are priced above the marketplace on the same SKU

Publishers compare. So do their readers, and a store consistently more expensive than the listing one click away is materially harder to place however good the outreach is. That sits in your pricing strategy and not in ours. Better said now than in month four.

Count your own pages. The order follows from what you find.

We keep a live register of every URL we have ever pointed a link at, and you get it at migration and again if you leave us. Hand it to whoever writes your redirect map. The problem mostly disappears at that point, and it costs nobody a fee.

Links that break because you deleted the page are not our replacement liability. Better said here than argued about later. What we do is flag them in the next report. Somebody then decides whether to redirect them or let them go.

Six regulated-category DTC stores, and what that does and does not prove

Our ecommerce work has been in restricted categories. CBD, kratom and adjacent products. Engagements ran anywhere from one year to four, in a market where a store cannot buy paid social or Google Ads and has to earn every visit it gets.

Four of the seven pages we could read publish real numbers, and the dominant model here is a per-link price banded by domain rating. We price differently. The variable is how many category pages the campaign has to move, because that is the thing this page has spent two thousand words arguing about.

We invoice on live placement. Nothing before, and a placement bought for a season that publishes after it closes is not invoiced.

Our floor is [$X,XXX] a month. Below that we cannot buy enough placements to move a competitive collection page. Taking the money would be taking it for nothing.