Most SEO plans are abandoned before they were ever tested
We inherit a lot of accounts, and the pattern in the handover documents is remarkably consistent. Three different keyword strategies in eighteen months. Two half-built topic clusters. A pillar page with nine articles pointing at it and a second pillar page, on a different theme, with two.
Nobody did anything stupid. Each pivot was defensible in the meeting where it happened. The problem is that SEO has a delay structure that makes a plan almost impossible to judge on the timeline most companies review things.
Here's the delay, in order. A new page has to be discovered, which takes days to weeks depending on your internal links and sitemap. Then indexed, which is a separate decision Google makes and sometimes declines. Then it accumulates impressions, and its position bounces around while the ranking systems work out what it is. Then, if it's genuinely good, other people find it and reference it, which is the slowest step of all.
Add that up and a page's first honest read is at three to six months. Which means a strategy changed at week eight wasn't judged — it was interrupted. You paid for the research, the briefs and the setup, learned nothing about whether the thesis was right, and then paid for all three again.
So the default answer to "should we change direction" is no. Not because sitting still is virtuous, but because the alternative is usually paying twice for a question you never answered.
What a pivot actually costs
Founders think of a strategy change as free because no new invoice appears. It isn't free — it just gets billed inside the retainer you were already paying, which makes it invisible.
On a ₹75,000/mo engagement, here's roughly what gets re-spent when you change direction mid-programme.
- Two pivots in a year is common and rarely questioned. That's ₹1,34,000–₹1,82,800 of a ₹9,00,000 annual budget spent on work you'd already bought once.
- The abandoned cluster doesn't sit quietly. Its pages stay indexed, keep their internal links, and if the new theme overlaps at all you've just built keyword cannibalisation into your own site.
- The clock restarts, not pauses. New pages begin the three-to-six-month ramp from scratch. You don't resume at month eight; you resume at month zero with eight months of costs behind you.
| What gets re-done | Effort | Cost |
|---|---|---|
| Keyword research and intent mapping for the new cluster | 12–20 hrs | ₹14,400 – ₹24,000 |
| Briefs already written for articles now cancelled — a quarter's worth at ₹1,500 each | 12 briefs | ₹18,000 |
| Pillar page rebuilt for the new theme, plus its internal link map | — | ₹25,000 – ₹35,000 |
| Re-plumbing internal links so the old cluster doesn't orphan itself | 8–12 hrs | ₹9,600 – ₹14,400 |
| Per pivot | 3–5 weeks of programme time | ₹67,000 – ₹91,400 |
Four events that genuinely justify tearing up the plan
These aren't symptoms. They're events — something changed in the world, not in a chart. That distinction is the whole argument of this page, because symptoms are noisy and events aren't.
- The business moved. New ideal customer, new price point, a new city or country, a product you retired. The test is one question: does the keyword list still describe the person who signs the cheque? If your average deal value went from ₹40,000 to ₹4,00,000, the queries that convert changed with it — bigger buyers search differently, later, and with more comparison intent. Pivot immediately and without guilt. The old plan is now optimising for a customer you no longer want.
- You rank, and the CRM shows nothing. The clearest possible signal, and the most under-used. Concrete rule: if you hold top-ten positions for ten or more target queries for two full months and qualified organic leads still haven't moved off the frozen baseline, the failure is in demand selection, not execution. You picked queries people search but don't buy from. Change the keyword set, not the agency — a new agency will inherit the same wrong list.
- The SERP for your money queries changed shape. Go and look at the actual results page for your five most important commercial queries, today. If a map pack now owns the top, if an AI Overview answers the question completely, if a marketplace bought the first three slots, or if Google shipped its own comparison unit, then the page you planned cannot win the click that's left. The response isn't to try harder on the same query — it's to move up or down the funnel, or into a format the results page still rewards.
- A platform or technical event. Replatform, domain change, merger, acquisition, a manual action, or a migration somebody ran without telling you. Every assumption in the plan about URLs, templates and internal linking is now void. Stop the content calendar, fix the foundation, then restart. Publishing into a broken site is the fastest way to waste a quarter of content budget — the replatform that killed your traffic has the failure modes in detail.
Three symptoms that look like triggers and aren't
Each of these has ended a perfectly good SEO programme in a meeting that lasted twenty minutes. All three are the compounding curve doing its job, or normal search volatility, or both.
"It's month three and traffic is flat"
Traffic is a lagging indicator of a lagging indicator. Impressions move before clicks, clicks move before leads, and each gap is weeks wide.
Before you touch the strategy, open Search Console and look at impressions and average position for your tracked query set on a 28-day comparison. Rising impressions with flat clicks means the pages are being served and not yet chosen — that's a title, intent-match and position problem, and it fixes itself as positions climb. Flat impressions at month three with pages fully indexed is a real problem, and it's the one case in this section where you should act.
"A competitor overtook us on our biggest keyword"
A single-keyword check is close to meaningless. Results vary by location, device, personalisation and the hour you looked, and rankings for competitive commercial queries move daily without anything having happened.
The honest read is aggregate position across your full tracked set over 28 days, not one query on your laptop. If the whole set slid over two months, that's a signal. One keyword is weather.
"A core update hit us"
Broad core updates roll out over a period of weeks, and Google's own guidance is to wait for a rollout to finish before assessing the damage, because positions move in both directions while it runs.
There's also no repair kit designed into a core update — it isn't a penalty with a fix. Ripping up content mid-rollout is the worst available option, because afterwards you can't attribute anything: you changed the site and the index at the same time. Wait, then read it by template and by section rather than sitewide. Usually one content type took the hit and the rest is noise.
The check to run before you act
Print this. It converts a panicked meeting into a twenty-minute investigation, and roughly half the time the investigation ends the argument.
| Symptom | First check | Act only if |
|---|---|---|
| Traffic flat at month three | Impressions and average position for the tracked set, 28-day comparison | Impressions are also flat and indexing is complete |
| A competitor overtook us | Aggregate position across the full tracked set, not one query | The whole set slipped across two consecutive months |
| Rankings dropped after a core update | Whether the rollout has finished, and whether the drop is sitewide or one template | Still down six to eight weeks after the rollout completed |
| Rankings up, leads flat | Which queries the leads should be coming from, checked in the CRM | Ten-plus top-ten queries have converted at zero for two months |
| New pages aren't getting traffic | Search Console Pages report — the exact exclusion reason for each URL | The reason is quality-related, not a crawl, canonical or redirect bug |
What survives a change of direction
When one of the four events does land, the mistake is to treat a pivot as a demolition. Most of what you built is topic-agnostic, and throwing it away is how a justified pivot turns into a wasted quarter.
- The baseline. Never re-baseline to hide a bad quarter — that's the single most common piece of quiet dishonesty in agency reporting. If the business genuinely changed, state the new baseline alongside the old one and explain why, in writing, before the quarter starts.
- Every technical fix. Crawlability, speed, schema, redirect hygiene. None of it cares which keywords you target, and it was probably the highest-return work in the engagement anyway.
- Pages that earn links, even off-strategy. If a page attracts references, it's doing authority work for the whole domain. Keep it, maintain it, and point its internal links at the new cluster.
- The internal link graph. You re-point it. You don't rebuild it. This is a day of work, and skipping it is how the abandoned cluster starts competing with the new one.
- The measurement setup. Analytics, conversion definitions, CRM source tracking, rank tracking history. Rebuilding this costs weeks and, worse, it breaks comparability with everything before the pivot.
- The research you're about to throw away. The old keyword list is evidence. It tells you which queries you tested and what happened, and that's the only way the second plan is better than the first rather than just different.
Review monthly, pivot yearly
The confusion that causes most thrash is treating "review" and "pivot" as the same activity at different intensities. They're different things on different clocks.
Monthly is a review. Five leading indicators, twenty minutes: impressions for the tracked set, average position, indexed pages versus published pages, new referring domains, and qualified organic leads in the CRM. The output of a review is an adjustment to next month's roadmap — which article ships, which page gets refreshed, what technical thing jumped the queue. That's normal operations, not a strategy change.
Quarterly is a decision point. This is the only meeting where a pivot is even on the agenda, and it should require one of the four events, not a feeling. Ninety days is also the shortest window in which any SEO thesis can be fairly judged.
Annually is a rewrite. Once a year, the whole plan gets re-argued from the business down: who the buyer is, what they search, what the results pages look like now, what compounded and what didn't.
The rough rule we'd hold ourselves to: most businesses should genuinely pivot at most once a year, and adjust the roadmap continuously. If you've pivoted twice in twelve months and the business hasn't changed, the problem is probably the review cadence, not the strategy.
What no lock-in does to the pressure to pivot
Contract length quietly sets everyone's incentives here, and both extremes fail in opposite directions.
A twelve-month lock-in with a long notice period means the agency's revenue is safe for two quarters no matter what happens. That buys patience, which SEO genuinely needs — and it also buys the agency the freedom to coast, because the invoice clears whether or not the plan was ever right.
Pure month-to-month flips it. Now the agency has to show something in thirty days or lose the account, and thirty days is far too short for anything real to compound. That's the structural pressure that produces bought links, thin pages published at volume, and reports that lead with traffic. The agency isn't being dishonest so much as responding to the clock you handed them.
Our version is month-to-month after the first quarter, with 30 days' notice, and the client keeps every asset. That only works because of what sits next to it: we freeze your trailing-90-day qualified organic lead count on day one, and if we haven't beaten it in 90 days we keep working free until we do. The thing being defended is the number, not the invoice, so neither side has a reason to redecorate the strategy for the sake of looking busy.
The honest caveat is that no-lock-in cuts both ways. A nervous founder on a monthly contract can cause thrash all by themselves — month four is where that pressure peaks, which is why clients fire their agency in month four more than any other month. A written baseline is the only thing we've found that stops both parties panicking on the same schedule. Details are on our SEO page.