Most link building dashboards are built from whatever the tool exports easily. That is why they are full of numbers that move without meaning anything and empty of the two or three that decide whether the programme continues.
Here is the split we use: six worth a slot, four worth deleting, and one that predicts more than the rest combined.
The six worth a dashboard slot
Referring-domain gap on the target page. The number the programme exists to close: page-level referring domains for the four URLs above yours, median, minus yours. Everything else is a proxy for this. If it is not on the dashboard, the dashboard is not measuring the job.
Placements live and indexed, with the index date. Not secured, not scheduled, not "in editorial". Live and indexed. The date matters because anything not indexed by day thirty should not be billed — that moves the cost of a bad host onto the supplier where it belongs.
Rejections, with reason codes. Count and breakdown. This is the only metric on the list that costs the agency money to report honestly, which is exactly why it is the most informative one. A programme with a zero rejection rate is a programme with no standard.
Anchor distribution over the trailing three months. Against the model set at kickoff. All-time distribution drifts slowly and hides a bad quarter; trailing three months shows a supplier starting to lean on exact-match because rankings were flat and somebody asked why.
Attrition and replacement. Links lost, links replaced, elapsed time. Roughly four percent a year is normal in B2B SaaS. A supplier reporting zero is not checking.
Position and impressions for the one target page. Singular. A site-wide average moves for reasons that have nothing to do with the programme, and it is the easiest number in this discipline to hide behind.
The four that are decorative
Total backlinks. Counts every link from every page of every domain. One publication that links to you from a sitewide footer produces four thousand backlinks and one referring domain. The second number is the real one.
Domain rating, yours. It moves late, it moves in jumps, it is a vendor's normalised model rather than anything a search engine uses, and it says nothing about the page you are trying to rank. Watching your own DR is watching a lagging proxy of a proxy.
Domain rating, theirs — as a purchasing criterion. Useful as a cheap floor when screening six hundred candidates. Useless as a target, because the moment it became a purchasing criterion an industry grew up manufacturing it. High authority alongside near-zero page traffic is the signature of inventory built to be sold.
Traffic to the placement page. Interesting, occasionally, but it is not why you bought the placement and it varies by two orders of magnitude for reasons outside anyone's control. Treat it as a screening input before purchase, not a performance metric after.
The one almost nobody tracks
Whether sales starts sending the placements to prospects, unprompted.
It sounds soft. It is the most predictive signal we have found, and it arrives in months two to four — long before rankings move and much longer before revenue does.
The mechanism is not mysterious. A placement that is genuinely good makes the category argument better than your own sales deck does, because it comes from a third party and it is written for readers rather than for buyers. Account executives find those pieces and use them. When that starts happening, the programme is producing something real regardless of what the ranking chart says yet.
When it does not happen, you have learned something equally valuable: the placements are technically fine and commercially inert. More volume will not fix that. Different publications and different arguments might.
Measuring it is unsophisticated and works. Ask the sales lead once a month which pieces they sent to prospects, and count. If the answer is consistently none after four months, that is a finding, not a gap in your instrumentation.
Two metrics to hold quarterly, not monthly
Blended cost per referring domain. Total spend divided by domains closed. This is a supplier-efficiency measure rather than a business one, but it is what makes proposals comparable. Credible B2B SaaS placements run roughly $150 to $500 across the market. A blended cost far below that band is not a bargain; it means the arithmetic of producing a real editorial placement does not close, so the inventory is coming from somewhere else.
Reachable list depth remaining. How many qualified, reachable publications have not been used yet. This is the metric that tells you when to stop, and it is the reason so many programmes overstay: in month fourteen the list is empty, the quota is not, and the quality bar quietly drops to fill the gap.
What a good monthly report actually looks like
One page. Six numbers with last month beside them. The rejection log. The anchor table with target against actual. Two sentences on what changed and why.
No screenshots of tool dashboards, no traffic charts for the whole domain, no domain rating. If a number cannot be tied to a decision somebody might make this month, it is decoration, and decoration in a report is how a programme goes eight months without anyone noticing it stopped working.
The test we apply: for each metric on the page, name the decision it would change. Anything that survives that question belongs. Most dashboards lose half their contents, and every one of them reads better afterwards.