What to record for keyword research return on investment reviews

What to take away

Decide what return means before you score it

Most teams in England measure keyword research return on investment as rankings gained. Rankings are an output, not a return. Return compares what you paid with what the work produced.

Settle on a money definition first. A Manchester software firm paying £900 a month for tooling plus £1,500 in analyst time spends £2,400 a month. If its clusters produce six qualified leads and two close at £6,000 each, the return is £12,000, a labelled illustrative example rather than a benchmark.

Write the definition down and get finance to agree it. A definition agreed in writing survives staff changes and quarterly restructures.

Build a weighted scoring rubric

Use a fixed rubric so every cluster is judged the same way. Weight each factor, score each cluster from one to five, then multiply.

A cluster that ranks well but feeds nothing valuable should not score above three. The rubric only works if low scores can survive a meeting.

Review the scores with whoever writes the briefs, not only with the person who bought the tool. The writer usually sees first whether a cluster produces usable work.

Cost the inputs properly

Return is only as good as the cost side. List every input in pounds: subscription seats, one-off data purchases, contractor days and internal hours at a stated day rate.

For example, a Birmingham agency charging clients £75 an hour and spending 20 hours a month on keyword work carries £1,500 of internal cost before any tool fee. Add a £150 seat and the true monthly input is £1,650. Teams that skip internal hours overstate return, sometimes by half.

Record the day rate you would pay a contractor for the same work, since finance will use that comparison whether you do or not. Budget for what people leave out, such as rework after a brief misses its mark. Our guide to keyword research hidden costs in England lists the usual suspects and how to catch them early.

Capture the evidence as you go

You cannot score return retrospectively from memory. Log four things each month: invoices, hours by person, briefs produced and outcomes attributed to each cluster.

Attribution is imperfect. Record a consistent rule, such as last-click on the enquiry form, rather than switching methods when results look weak.

A shared ledger beats a lost email thread when a budget review arrives. If you buy tools through procurement, expect a demonstration stage before sign-off, as described in the Semrush demo request process.

Review quarterly and reset the weights

Score every cluster each quarter, then compare totals with the money spent. Clusters below two for two consecutive quarters should be cut or rewritten.

Reset the weights once a year, not every month. Changing them mid-quarter makes the trend line meaningless. If you need the full cost picture first, work through the keyword research costs and budget guide for England.

Name one owner and put the next review date in the calendar. Compliance belongs in the review too. Claims about performance sit inside consumer protection expectations, and the competition and consumer protection guidance shows how regulators treat unsupported statements.

Common questions

How often should we rescore return on investment?

Quarterly suits most England teams, matching typical reporting cycles and giving clusters time to produce leads. Monthly scoring tends to chase noise.

What if we cannot attribute leads to keywords?

Use a stated rule and apply it consistently, such as first enquiry after an organic landing page view. Record the limitation in the ledger.

Do we need a tool to do this scoring?

No. A shared spreadsheet with the rubric, invoices and hours will do. Tools help with collection, not with the judgement about what return means.

Should internal hours be included in the cost?

Yes. Internal time carries a cost even when no invoice arrives. Excluding it flatters the return and invites challenge from finance.