Write the Gap Analysis Report (Template)

Full transcript

1,034 words · script draft, under review.

Intro

You've spent two weeks on an analysis. Search Console exports, results-page snapshots, competitor keyword lists, a scored gap list, and a ranked plan of five items.

Then you send it to the person who decides. They read the first page, maybe the second. And they ask: "So what should we do?"

Good analysis dies in bad reports all the time. Not because the work was wrong, but because the decision was buried under the evidence.

In this episode, I'll walk through the report template I use for a gap analysis: the structure, how to back every recommendation with evidence, how to write numbers honestly, and how to present it. The example is our made-up shop again.

Three readers, one report

A gap analysis report has three readers, and they want different things.

The decision-maker, a client or a boss, wants a decision. What to do, what it costs, and what happens if we don't.

The doer, a writer or a developer, wants specifics. Which URLs, which keywords, which template, which links.

And the third reader is you, three months from now. You want to know what you believed at the time, what the baseline was, and where the data came from. Without that, you can't tell in episode forty-six whether anything worked.

So the report is layered. The decision-maker can stop after page one. The doer reads the recommendations. You keep the method and the appendix.

The structure, section by section

Here's the structure.

Section one: the summary. Five sentences, no more. What we looked at, the main finding, the top recommendation, what it costs, and when we'll know if it worked. Write this last, put it first.

Section two: the question and the scope. Which site, which market and language, which date range, and which competitors, with one line on why you picked them. That line saves you from the "why didn't you compare us to the big brand" discussion later.

Section three: the baseline. The traffic picture from part one of the framework. Clicks, impressions, the pages that matter, and the exact date range. Filtered for the right hostname, as we did in episode fifteen.

Section four: findings. The gaps, grouped by type: keyword gaps, page-type gaps, structural gaps. Short. This is what you found, not what to do about it.

Section five: recommendations. The ranked five from episode forty-four. This is the heart of the report, and we'll look at its format in a moment.

Section six: method and limitations. Data sources, dates of every pull, what's estimated and what's measured, what you didn't check.

Section seven: the re-measure plan. For each recommendation, the metric, the comparison pages, and the date.

And then an appendix: the full gap list, the dropped items with reasons, and the raw exports.

One card per recommendation

Every recommendation gets the same card.

What: one sentence, concrete. "Add links from the main hub page to the twelve buying guides that currently sit four clicks deep."

Why: the evidence. And here's the rule I follow: every claim must point to a file and a date. The crawl depth comes from a crawl export on a specific day. The competitor comparison comes from a stored results-page snapshot, with its date. Results pages change, as we saw in episode thirty-three, so an undated screenshot is weak evidence.

Expected effect: a direction and a range, not a single number. "More of these guides crawled and indexed, and more impressions on their queries within six to ten weeks." Made-up range, but the form is the point.

Effort, dependencies and confidence come straight from the prioritization sheet.

How we measure: the metric, and the control pages we'll compare against.

If you can't fill in the "why" field with a file and a date, the recommendation isn't ready. Move it to a "to investigate" list instead.

Writing numbers honestly

Numbers are where reports lose trust.

Separate measured from estimated. Clicks and impressions from Search Console are measured, for your own site. Search volumes and competitor rankings from a keyword database are estimates, and ranking databases can be stale. Label them, so nobody mixes them up.

Explain the zeros. A search volume of zero usually means under about ten searches a month, not no demand. One line in the method section saves a confused question.

Use ranges and bands, not false decimals. A traffic forecast of exactly two thousand three hundred and forty visits a month is not more credible than "a few thousand". It's less credible, because everyone knows you can't know that.

And never promise rankings. Promise actions, a measurement plan, and an honest verdict afterwards.

Presenting it

When you present, start with the decision, not the method. "I recommend these five things, in this order. Here's why number one is first." Then walk through one recommendation per slide.

Keep the dropped list ready. The most common question is "why aren't we doing this other thing?" If you can answer with the reason you wrote down when you dropped it, the discussion takes a minute.

And end by asking for the decision explicitly. Which items are approved, and who owns each one. A report that ends without a decision is just reading material.

Where reports go wrong

Four pitfalls I see again and again.

The data dump. Twenty charts and no conclusion. Every chart in the main report should support a recommendation. If it doesn't, it goes to the appendix.

Forty recommendations. That's not a plan, it's a backlog. The report carries five. The rest is in the appendix, ranked, with reasons.

No dates. Search data changes every week. A report without the date range and the date of every pull can't be checked later, and it can't be compared with the next one.

And the hidden method. If the reader can't see where a number came from, they either trust it blindly or not at all. Both are bad. A short method section, with what's measured and what's estimated, is what makes the rest believable.

One last habit: version the report. When you run the analysis again in three months, the old report becomes the baseline for the new one.

Wrap-up

Decision first, evidence for every claim, honest numbers. The full report template is linked in the description. Next episode: after you ship, how to measure whether the fix actually worked. See you there.