Almost nobody who asks for this has a shortage of tactics
Most organizations that ask for a search strategy already have a list. It is long, mostly correct, and stalled. It was assembled from an audit, a tool export, a conference talk, two vendors and an internal argument, and not one item on it carries an owner, a position in a queue or a date.
Strategy, in the only sense worth paying for, is the work of turning that pile into a sequence. What gets done, in what order, funded from where, blocked by what, judged against which number — and the part almost every plan omits, what gets stopped to make room. It is a decision-making artifact. It is not a document of aspirations with a slide about personas at the front.
The plain test: if a plan cannot tell you what you are not doing this quarter, it has not decided anything. A list of good ideas ranked by how good they are is the same list you already had.
What gets established before a single recommendation is written
Four things, in this order, because each one constrains the next.
What the demand actually looks like on the page, not in a keyword tool. Volume is the least interesting property of a query. What matters is what the result page returns for it: whether it is answered in place by an AI Overview, whether a local pack sits above the organic results, how many ads occupy the top, whether the intent is informational or transactional at all. Two queries with identical volume can be worth wildly different amounts because one of them still sends clicks and the other has been answered before anyone scrolls. On the keyword set researched for this site — 1,264 queries — 870 returned an AI Overview. That is 69%, and it is a property of this subject area, not a universal constant. Yours has to be measured, not assumed.
Where the site currently stands. A dated baseline from Search Console, taken before anything changes, inside the 16 months of history the tool retains. Without it, every later claim about improvement is an assertion.
What can actually ship. Development capacity, release cadence, content capacity, legal or brand review, and platform limits that no amount of strategy overrides. A plan that assumes a template change your platform does not permit is fiction.
What an outcome is worth. Lead value, margin, sales cycle length. You supply this. Nobody outside your business can.
What each channel is genuinely good at, which is not what it is usually sold as
Paid search buys presence on the day you decide to buy it and stops on the day you stop paying. Organic search accrues slowly, does not switch off, and cannot be turned on in a hurry. Everything else about the allocation follows from those two sentences and from your sales cycle, your margin, and how quickly you need to learn.
The underrated use of paid is as an instrument rather than a channel. A search terms report tells you the exact language buyers use, which of those phrasings convert, and what a conversion costs — within weeks, on your own site, with your own offer. Organic research cannot produce that evidence at that speed, and it is frequently worth running paid on a segment purely to settle an argument about intent before committing a quarter of content production to it.
The failure I see most often is not a bad split. It is two channels bought from two providers, judged on two different numbers, neither of which is revenue — so paid claims the conversions organic warmed up, organic claims the brand queries paid created, and nobody can say what either is worth. Fixing that is usually worth more than reallocating anything.
Sequencing, which is where most of the value actually sits
The order is not a matter of taste. It is a dependency graph, and getting it wrong wastes real money.
- Nothing gets published into a template that cannot be crawled, rendered or indexed. Commissioning content for a section search engines cannot read is the most expensive avoidable mistake in this field, because the invoice is paid before the problem is visible.
- Measurement is fixed before anything is judged. If conversion tracking is wrong, every subsequent decision rests on fiction, and the more disciplined the team, the more confidently they will act on it.
- A scheduled replatform or migration freezes almost everything else. Work done on URLs that are about to change is work done twice. Either it waits, or it is specified to survive the move.
- Authority and content programs start last and finish never. They have the longest clock, so they start once the things with short clocks are underway, not instead of them.
Sequencing also decides what you learn first. Early items should be chosen partly because they resolve an uncertainty cheaply — you want the question of whether a category responds at all answered in month two, not month nine.
What the document actually contains
Six parts, and it is written to be ticketed rather than admired.
- The ordered queue. Each item with the reason it sits where it sits, the owner, the dependency that must clear first, and the evidence that prompted it.
- The channel split, with what each channel is accountable for stated separately, so neither can take credit for the other.
- A written indexation and content policy — what belongs in the index, what does not, and what happens to pages that stop being useful. Most sites have never written this down, which is why it drifts.
- The measurement plan: the baseline, the leading indicators, the lagging ones, and the review dates.
- The deferred list. Decisions deliberately not made, each with the condition that would reopen it. This is the section that stops the same argument recurring monthly.
- The ruled-out list, with reasons. Knowing why something was rejected is what keeps it rejected after the next vendor pitch.
The decisions in here that are not mine to make
Budget. Brand positioning. Appetite for legal and reputational risk. What a lead is worth. Whether to enter a market at all. Whether to keep a product line that search demand says is dying.
A consultant who settles those for you is guessing with your money. What I can do is state the search consequence of each option precisely enough that the person who owns the decision can make it with the trade-off in front of them: this structure costs six weeks of development and removes an entire class of indexation problem; that one costs nothing now and caps what the site can rank for later.
I would rather hand back a plan with three open decisions clearly framed than one that quietly assumes an answer to all three. The assumed version reads better and ages worse.
How this is engaged, and what makes one plan more work than another
It is advisory. It is bought either as a defined piece of thinking with an end date, or as part of a standing seat where the plan gets revisited as conditions move. No implementation is included — that is deliberate, and it is worth saying plainly, because a strategy priced with delivery attached tends to recommend the delivery.
What moves the effort, and therefore the cost: the number of markets and languages in scope, because each one multiplies the demand research and the technical validation; the number of channels; whether analytics and conversion tracking can be trusted, since a plan cannot rest on numbers that are wrong and fixing them comes first; how many stakeholders have to agree, which is a real cost and an unpopular thing to admit; and whether a current diagnosis exists or the site has to be examined before anything can be recommended.
If a recent, competent audit already exists, say so. Redoing diagnosis you have already paid for is the easiest line item to remove from any proposal, including mine.
When you should not buy this, and how you will know it worked
Skip it if the diagnosis is already clear and the constraint is execution. A small single-market site with a known technical defect does not need a strategy; it needs the defect fixed, and a strategy engagement in that situation is an expensive way to delay the fix by a month.
Skip it also if there is nobody to execute anything. A plan handed to an organization with no development capacity and no content capacity becomes a document that gets quoted in meetings for a year. Buy capacity first, or buy a fixed-scope project that includes the work.
The honest measures of whether the plan was worth buying are mostly about behavior, and they show up before any traffic number does. Did the order hold for two quarters, or did it collapse into whatever was urgent? Did decisions get faster, because the framing already existed? Did the deferred items get reopened on their stated trigger rather than at random? After that, the lagging measures: impressions on the query classes the plan targeted, then position, then clicks, then the commercial number you nominated at the start — which was written down before the work began precisely so that nobody gets to choose the winning metric afterwards.
Frequently Asked Questions
What is the difference between a search marketing strategy and an SEO audit?
An audit tells you what is wrong with a site. A strategy tells you what to do about it, in what order, alongside everything else competing for the same budget and the same developers. An audit is diagnostic and technical; a strategy is allocative and commercial, and it covers paid search, measurement and content production as well as the site itself. They are often bought together, in that order, because a strategy written without a diagnosis is guessing about the starting position.How much does a search marketing strategy cost, and how is it priced?
It is advisory work, bought either as a defined piece of thinking with an end date or inside a standing seat. What moves the figure is scope rather than company size: how many markets and languages are in play, how many channels, whether conversion tracking is trustworthy or has to be repaired before anything can be measured, how many stakeholders must agree, and whether a recent competent diagnosis already exists. If one does, that work should come out of the proposal rather than being repeated.Do I need a strategy if an agency is already running the work?
Sometimes, and the reason is structural rather than a comment on the agency. Any provider that both recommends and delivers the work has an interest in what gets recommended. A separate plan, written by someone who is not going to execute it, gives you an order of work you can hold the delivery provider to. It is also the cheapest way to settle whether the current program is aimed at anything, which is a different question from whether it is being executed competently.How far ahead should a search plan look?
Far enough that the slowest item has room to work, which usually means several quarters, and reviewed often enough that it does not become a relic. Google states plainly that some changes take effect in days while others can take several months, and offers no guarantee of an outcome, so a plan pinned to precise dates for ranking movements is pretending to a precision nobody has. Plan the sequence and the review points; forecast the leading indicators; leave the lagging ones as measurements rather than promises.What if the plan concludes that work already being paid for should stop?
Then that goes in the document, with the reasoning and the evidence, because it is usually the single most valuable line in it. Stopping something is the fastest way to fund something else, and it needs no development time. The common candidates are content produced for query classes that no longer return clicks, link acquisition that would not survive a spam policy reading, and reporting work that nobody reads. Reallocation is easier to approve than new budget, and it moves faster.Can a plan be produced without access to analytics and Search Console?
A worse one, yes. Without Search Console there is no impression or position data at all, no baseline, and no way to tell which queries the site is already close on — which is normally where the cheapest early wins are. Without analytics there is nothing to weight recommendations by commercially, so the plan gets ordered by technical severity instead of by consequence. Read-only access to both, plus whatever crawl and log data exists, is the minimum for the work to be worth what it costs.Published